Teardown : Fluke 845A/AB/AR nullmeter/HZ voltmeter tweaks and mods (and repairs)

[TiN]

Got a new toy..
Since mostly these units require some tweaking and modifications, I think this should belong to here, rather than just repair section..



Really clean and nice condition outisde:



Let's see inside:



Guarded area:



Woohoo 

I was expected to see infamous neon tubes, light guides and photoresistor chopper amplifier, but was greeted insteal with alien-looking green PCB with micropower regulator Linear LT1120CN8 and few HP 2731 optocouplers(?) !
LT1120 dates with 1993 week 52. Is it that optically-isolated Bilateral Analog chopper, which mentioned in Fluke 845AB manual's note in 7/1993 print?

Repairs

* Meter does not read anything in BAT OPR mode. It's just dead on zero, like no power is supplied, even though BAT CHK shows good battery
* In LINE OPR mode meter is unable to zero on ranges below 30µV

First issue was resolved by fixing blown track on power PCB.

As usual with equipment with age over 15 years I replace all electrolytic capacitors to avoid electrolyte leaks and further desctruction.



There is plenty space inside, so I just used usual good quality Chemicon radial capacitors instead of axial. I used 470uF 63V instead of original 470uF 25V and 6.3V caps.



Green PCB with chopper assembly have blue multiturn trimmer pot, accessible from the side. That pot adjustment allowed to shift zero, so now meter can be zeroed on lowest 1 µV range.

Test



Settings: Fluke 845AB Range: 1 µV, OPR switch = ZERO, 3458A DCV auto, 10 NPLC, AZERO ON

I will post YouTube video later.

Perhaps worth to reverse that green PCB schematics?

Future plans:

* Add backlight for meter
* Remove input resistance to have high impedance input
* Study chopper operation ,perhaps replace with something more stable?
* Ultimate mod - replace analog meter with digital PCB module with battery-powered Sigma-Delta ADC to show readings in digital?

+ As usual article with bit more details on my site.

[manganin]

Is it that optically-isolated Bilateral Analog chopper, which mentioned in Fluke 845AB manual's note in 7/1993 print?

Yes. The manual supplied with the modified meters was not updated and therefore didn't match the instrument.

Perhaps worth to reverse that green PCB schematics?

That would be very interesting.

No improvement in specifications, but the new version solved the well known neon lamp issue. The oxidation of the new style binding posts caused some problems because they were plain copper and not gold plated like the old ones.

[Dr. Frank]

Hi TiN,

Great find of this new version, and a reverse engineering is greatly appreciated!

You already have repaired this instrument, as you measure its output?

Then, I also would be interested in the leakage / bias current of this 845AR..
This important parameter is lacking in all specifications, only a high / infinite resistance is proclaimed, due to zero volt input at bridge zero.

I am really skeptical about the bias current, maybe modern chopper amps or low bias bench DMMs like the 3458A , 2002 and 8598A really are on par, at least.

Frank

[Kleinstein]

I won't expect very good performance from the circuit (from xdevs website): there is a rather high series resistance (300 K total) at the input, which provides quite some noise. Also the LDRs won't be very high impedance.

I can understand they did an update to the old circuit. So the impedance might be higher with the modified chopper part.

When it comes to input bias, the low chopper frequency helps. Also as the input range for the amplifier is so small input protection can use just two back to back diode and this way cause less leakage. So the bias / leakage can be really low.

If it really comes to mods - for better performance / stability, this might end up as a build from scratch.

[Jay_Diddy_B]

Hi group,

I had a look at the manual on the ko4bb website. The are changes in the manual to reduce the input resistance on the 1uV thru 1mV from 10M Ohms to 1M Ohm by reducing the value of R104 from 10M to 1M.



There is also reference in the manual to increase the input resistance by removing R104 completely.


I will have to have a look at my 845A.

Regards,

Jay_Diddy_B

[quarks]

Nice score, congratulations. In my my 845 I only replaced the batteries so far. Looking forward to see your video.
Thanks for sharing.

[Jay_Diddy_B]

Hi group,
I opened mine. the unit I have was made in 1974. It has neon bulbs and plastic rods, light pipes.

Here are few pictures of working:





 
[Very Sexy] 

I suspect that the green board in TiNs unit is an oscillator and driver for Vactrol, Led/LDR. I don't think the op-amp is part of the precision circuitry.

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Hi,

I looked at TiN's photograph. I think the green board gets timing information from where the neon lamps used to be connected. The board probably has Vactrols H11F1 Photo FET Opto-couplers where I have indicated. Excuse the scribble writing, I am not very neat using the mouse.



Regards,

Jay_Diddy_B

[TimFox]

I thought I had a bad 845, but found that the input transfer switch needed a mechanical tweak.  The rotate/push mechanism is a simple switch without contacts, that drives the real transfer switch through an insulating rod.  I needed to bend the leaf holding the contact from the front panel connector so that the moving leaf would contact it appropriately.  Before that, I could zero the unit but it did not respond to external voltage inputs.

[Squantor]

Hi there,

Fluke 845A owner but it is still under repair. I completely revised all caps/resistors, replaced the neons by leds and also changed the zero circuit to using two 1.25V references. But still having problem with what I think an unstable zero. I am also noticing when I put the zero completely left or right that the +15V and -15V vary by a volt or so. Anybody else experiencing this? This was before and after the reference modification. Does anybody see a lot of variation in the +15 and -15V in general?

@TiN:
Holy crap, a Fluke 845 with the fabled electronic chopper! Please reverse engineer so I can upgrade my 845 too! It is unusable as it is now... Seems like a more complicated circuit then I expected, lots of components there. I wonder why, if it only needs to drive the optoFET leds... Maybe it has to do with the dutycycle and timing?

PS: I revised all the caps, because they where the old philips variants and they all had the bulge on the underside that makes them dry out. All the resistors where replaced as almost all of them where 20% or more high. A normal problem with carbon composition resistors.

[doktor pyta]

Only for reference here is  measurement of my Fluke 845AR on 1uV range, input shorted. Sampled every 10s, using K2001 2VDC 1NPLC Filter=AVER10, total record time ca.11h

[TiN]

doktor pyta
Your unit seem to be much more stable. Look like I was too quick saying it's fixed! Doh 
I'll have to check more in detail. I can see needle sway slowly from -4 to 4 reading on 1uV range, with OPR switch set to ZERO.

[picburner]

This is a part of the new chopper circuit diagram derived from the photos.
Unfortunately there are things that I can't see and, although I tried to turn inside the pictures like in the Blade Runner movie, I'm not really able to do it....
Someone wants to continue correcting errors and filling in the missing values?
TiN, I lend you my endoscopic camera if you want....

[TiN]

Wow, somebody did my work. I'm impressed, really. I will be out of the country this week, but will definately remove covers and post more pics next week, no worries.

[enut11]

Hi All. Just purchased an 845AB from the States and awaiting delivery. Mine has black knobs and looks older so probably has the neon tubes.
This is an interesting post and am looking forward to further discussions. Will post if I find something of interest in mine when it arrives.

Questions
1) Here do I buy Chemicon caps in Australia?
2) Can I replace the carbon resistors with 1% metal film?
3) The Fluke manual says not to clean the rotary switches. Any thoughts?

enut11

[Vgkid]

1) Here do I buy Chemicon caps in Australia?
2) Can I replace the carbon resistors with 1% metal film?
3) The Fluke manual says not to clean the rotary switches. Any thoughts?

1) Mouser, Digikey, RS Australia, Element 14. Any high end capacitors should work.
2) Looks like sqauntor, has done it. No problems there.
3) Not sure, maybe if it causes issues.

[enut11]

2) Can I replace the carbon resistors with 1% metal film?

2) Looks like sqauntor, has done it. No problems there.

Thanks vgkid. I missed that in the posting. I need to work out the best way to preserve the PCB when I do it.
enut11

[Squantor]

2) Can I replace the carbon resistors with 1% metal film?

2) Looks like sqauntor, has done it. No problems there.

Thanks vgkid. I missed that in the posting. I need to work out the best way to preserve the PCB when I do it.
enut11

Hi enut11. Indeed I replaced all the carbon composite resistors in the Fluke, most of them have drifted upwards, especially the higher resistance values. The ones that stay hot like the 150Ohms series resistor was okay, but I replaced it anyways. I dread measurement equipment with carbon composite resistors, I have quite a few HP instruments, that are loaded with them, during repairs I always most of them have drifted upwards. Some of the flakey behaviour is usually to these resistors drifiting upwards making the equipment work maginally. I also have a sizable collection of philips equipment and they really had the rule "eat your own dogfood" with components, they use carbon film and most of them are in spec except for some high resistors. Not a lot repair needed on those, except for some transformers that become leaky.

When replacing them in the Fluke845 take notice of ALL the resistors, a few are underneath the wafer switch and it can be a hassle in removing them, use good strong tweezers. The vias are not plated through so with a good higher power soldering iron and solder sucker you will do fine, at least I did . I used a weller 50Watt iron with a fat chisel tip.

Some of the components are hard to remove due to ground planes and such. Just cut the leads off the old components off and desolder lead by lead.

The shielded part pay attention to that that after soldering, clean the polyethylene grommets with a Q tip and some denatured alcohol.

Unfortunatly the replacement of resistors, the neons by leds and the caps have not solved my problem with a very high drift. It seems that the zero setpoint is drifting a lot and I feel it has to do with the -15 that moves 300mV when the zero is fully clockwise and counterclockwise..

Can anybody measure the powersupply values under the following settings:
Put range in 30uV, set the OPR switch to ZERO.
Now measure the voltages of the supply when the zero is fully clockwise, zeroed and fully counterclockwise.

[enut11]

Hi Squantor
Thanks for the very comprehensive reply. This will be my first time upgrading a piece of antique metrology gear and looking forward to it.
My 845AB is still in transit from the USA and could take up to 4 weeks.

I hope someone answers your question but failing that I will respond as soon as I can. Did you replace the neons because they were faulty or were you chasing the drift problem?

With regard to wattage, modern resistors seem physically smaller for the same rating and I often wonder if this is realistic given that surface area plays a part in heat dissipation. Your thoughts?
enut11

[Squantor]

@enut11:

Indeed, I was chasing the problem and upon first opening I noticed that one of the tubes flickered while the other was a solid orange. So a replacement was in order. I followed the guide mentioned here: http://www.ko4bb.com/getsimple/index.php?id=manuals&dir=Fluke/Fluke_845A_AB

As I can see, most of the resistors have such small loads, especially in the amplifier region that it should not be an issue, the only thing I would pay attention is sufficient wattage for the resistors in the led drive and the powersupply section. The 150Ohm and zener diode get quite toasty.

[Squantor]

Wow, somebody did my work. I'm impressed, really. I will be out of the country this week, but will definately remove covers and post more pics next week, no worries.

Hi TiN, did you find the time to make some more detailed photos of your fluke? Also, mine has a lot of germanium transistors that I suspect of having problems but finding replacements is hard. Can you look at the type numbers of your transistors what type they are?

[TiN]

Sorry everyone, I got busy with other stuff. Just opened this 845AB again and took photos around chopper board.
I had updated article with new pics.



And in process I damaged one of those TO-92 NPNs on base board. Was trying to tilt the thing so can see label on photo and middle pin just snapped. Oh, well, more repairs... 

[Squantor]

@TiN:

Thank you very much, lets start reverse engineering this!
I am sorry to hear about the transistor. I find this Null detector a pain to service, you need to almost completely disassemble it before you can get at the critical parts. It is quite a chore and then you have two boards flopping around. Well, lets find some time to start trouble shooting mine :-).

[zlymex]

It is always a pleasure looking at TiN's thread. Nice and big photo, thorough analysis and repair, and a good test chart at the end.
I had several 845A/AB/AR too but didn't have time/patient to analysis/modify except change line voltage from 110V to 220V.
The chopper amplifier is complex, someone replace it with a chopper stabilized opam such as LTC2054, don't know how it perform against the original.
Also looking forward the reverse engineering.

[enut11]

"Can anybody measure the powersupply values under the following settings:
Put range in 30uV, set the OPR switch to ZERO.
Now measure the voltages of the supply when the zero is fully clockwise, zeroed and fully counterclockwise."
Squantor

Hi Squantor
Received my 845AB today but it was DOA. Lots of corrosion in the power supply area caused by leaking Nicads. It will be some time before I get it going.
enut11

[enut11]

Hi All
Just purchased a Fluke 845AB High Impedance Voltmeter-Null Detector on eBay. The 845AB usually sells for >$1000US so was happy to get one delivered for under $275AU. Of course, with any good deal, you may not get a working instrument and this was the case for me.

Externally, the unit was in good condition for its age (circa 1971?) with no major scratches or dents.

The 845AB was powered by a Nicad battery pack and these are notorious for leaking. Getting to the pack required a fair bit of disassembly. The cells are held against the back of the power supply PCB by a plastic moulding. Even before I had removed the battery holder I could see evidence of battery electrolyte corrosion. On removing the holder, all the cells fell out! What was strange was that someone had replaced one of the Sub C cells with a C cell and did a very bad job of securing the holder back onto the PCB.

One of the pictures that I have posted shows the extent of the corrosion on the tinned copper tracks. This board also holds the transformer and other power supply components. Any advice on how to clean this board would be appreciated.

Thankfully, the main PCB with the voltmeter components is housed in a separate aluminium shield and did not show evidence of corrosion.

This looks to be a long time restoration. Will post more info as I progress further into the disassembly and clean/repair phase.

enut11

[builder]

Hi Enut11,

I have just done a complete rebuild of an 895A DC differential voltmeter, on which the voltmeter circuit (called TVM in the 895A manual) of your 845A will, if made in 1971, is very likely to be based.

I replaced the neon/LDR chopper isolated with ILED/FER isolators ( later 845 models used these).  Worked brilliantly. Can send more detail if required.

Basher

[enut11]

Hi basher. As the unit is currently dead I am not able to verify if the chopper works or not. Am interested in your mod. Currently researching best way to clean corrosion off PCB. Enut11

[Squantor]

I replaced the neon/LDR chopper isolated with ILED/FER isolators ( later 845 models used these).  Worked brilliantly. Can send more detail if required.

Basher

Hi Basher,

Would be nicer to share it with all of us with ailing 845's

[TiN]

I'd be interested to see as well, it's never bad to know more details and tricks about these old but still useable tools.

[enut11]

Well, I cleaned the corroded power supply PCB of nicad residue as best I could, replaced the two electros but no luck in getting any meaningful response from my F845AB.
No warm glow from the neons. Random meter movement - at least a good sign that the movement is working. No response from the zero control. No response from input voltage. Even the battery test (8.4V) reads low at 6v.. Lots to be done to get this unit operational.

As this unit is difficult to get apart for service, I decided to add connectors from the power supply to the main PCB.
enut11

[Squantor]

Well, I cleaned the corroded power supply PCB of nicad residue as best I could, replaced the two electros but no luck in getting any meaningful response from my F845AB.
No warm glow from the neons. Random meter movement - at least a good sign that the movement is working. No response from the zero control. No response from input voltage. Even the battery test (8.4V) reads low at 6v.. Lots to be done to get this unit operational.

As this unit is difficult to get apart for service, I decided to add connectors from the power supply to the main PCB.
enut11

First check the +15 -15 supplies, I am suspecting the 83Hz supply oscillator if the neons are not burning. Good that the movement is working, I am always scared with shipping equipment with mechanical indicators due to shock sensitivity.

When you get the +15 -15 supplies, the neons should glow unless they are completely shot, but usually they do not fail that abruptly.

Unfortunatly, cant help you with the battery issues. I have the 845A, so no batteries there.

[enut11]

Power supply is way out, 13v and -8.5v. Still looking into it.

Does anyone have experience repairing nicad damaged boards. As reported above, I scrubbed the PCB first with diluted vinegar to neutralise the electrolyte, then hot soapy water followed by a pure water rinse then drying in a low temp oven. There is still some residual corrosion around the component leads solder pads. See previous photo.

I am trying to decide whether I have to remove every component from the PS PCB, clean the bare board and resolder the parts back on.
enut11

[Squantor]

Power supply is way out, 13v and -8.5v. Still looking into it.

Does anyone have experience repairing nicad damaged boards. As reported above, I scrubbed the PCB first with diluted vinegar to neutralise the electrolyte, then hot soapy water followed by a pure water rinse then drying in a low temp oven. There is still some residual corrosion around the component leads solder pads. See previous photo.

I am trying to decide whether I have to remove every component from the PS PCB, clean the bare board and resolder the parts back on.
enut11

I have no experience with cleaning PCB's affected by corrosion from nicads, only from electrolytic capacitors deciding to spray their contents inside a Philips lab powersupply.
Some traces where completely eaten away and component leads where damaged. Re-soldering every joint would be smart. If a trace is not shiny metal scrub away the oxidation until it shines. I used one of those fibreglass cleaning pens (use gloves! those fibres will get everywhere and mess up your day with itching!) until the trace metal was visible or the trace was completely corroded through. In your case the traces look good, no "crust" of gray or reddish residue.

Also check the transistor sockets if some of the electrolyte has wicked up into the socket messing with the transistors. Remove the transistors and check, clean if needed. Did you power it on before you cleaned up the electrolyte? That might have damaged some components.

Have you measured the waveforms of the 83Hz Oscillator?

[enut11]

Thanks squantor. Unfortunately I did power it up when I received it. Currently away for 10 days. Will try your suggestions when I get back. Enut11

[FlyingHacker]

I just recently scored a Fluke 883A locally. It is from ~1966 and has a mechanical chopper you can hear. (Thread here: https://www.eevblog.com/forum/repair/fluke-883a-differential-voltmeter-%28pics%29/ )

I used the same trick with radial caps in most places rather than axials (I had a few axials).

The thing is pretty much spot on with my Keithley 6 1/2 digit DMM.

Is the 845 a newer unit? I see your guys mentioned 1971. The weird optical chopper seems like an older technique rather than a newer one. Anyone up on the history of these?

[Squantor]

I just recently scored a Fluke 883A locally. It is from ~1966 and has a mechanical chopper you can hear. (Thread here: https://www.eevblog.com/forum/repair/fluke-883a-differential-voltmeter-%28pics%29/ )

I used the same trick with radial caps in most places rather than axials (I had a few axials).

The thing is pretty much spot on with my Keithley 6 1/2 digit DMM.

Is the 845 a newer unit? I see your guys mentioned 1971. The weird optical chopper seems like an older technique rather than a newer one. Anyone up on the history of these?

Hi there, I also have a similar unit, the 883AB (with battery pack). The pack was completely shot and needs replacement. It uses a big stack of coincells and a bunch of C sized cells. Bit of a pain to replace but the meter non functional as it is. Still need to look into it as it is a bit of a pain to work on but not as much as the 845.

The 845 is not the same as the 883. The 883 is a differential volt meter that uses a Kelvin varley divider. The 845 is just the null detector circuit with a divider up in front to give it some voltmeter capability. To give the same functionality as the 833 you need to pair it with a kelvin varley divider like the fluke 720 and a voltage reference.

I have experienced that a lot of precision equipment with mechanical or neon choppers usually have problems with the chopper. You are quite lucky that yours is fine :-). I am planning to give the 883 a good look and see if its functional and maybe experiment with replacing the mechanical chopper by something using optofets.

I am reconstructing part of the circuits from the 845 onto vero board to experiment with different chopper circuits to make my own replacement plugin for all the mechanical/neon solutions. Either by using the H11F1 optofets, mosfets (like in the Philips PM2434), or analog switches.

[FlyingHacker]

I just recently scored a Fluke 883A locally. It is from ~1966 and has a mechanical chopper you can hear. (Thread here: https://www.eevblog.com/forum/repair/fluke-883a-differential-voltmeter-%28pics%29/ )

The 845 is not the same as the 883. The 883 is a differential volt meter that uses a Kelvin varley divider. The 845 is just the null detector circuit with a divider up in front to give it some voltmeter capability. To give the same functionality as the 833 you need to pair it with a kelvin varley divider like the fluke 720 and a voltage reference.

I see. So the 845 works like the 883A in TVM mode?

I have experienced that a lot of precision equipment with mechanical or neon choppers usually have problems with the chopper. You are quite lucky that yours is fine :-). I am planning to give the 883 a good look and see if its functional and maybe experiment with replacing the mechanical chopper by something using optofets.

I am reconstructing part of the circuits from the 845 onto vero board to experiment with different chopper circuits to make my own replacement plugin for all the mechanical/neon solutions. Either by using the H11F1 optofets, mosfets (like in the Philips PM2434), or analog switches.

I do feel quite fortunate that the chopper works, especially for the price I paid for the whole unit. I have dealt with some mechanical choppers before, and they have failed. The thing runs audibly whenever the unit is turned on. I am watching this thread closely for ideas on a replacement chopper.

What are the challenges in the chopper? Looks like it has to both act as an low impedance input switch and control the transistor that pulls the output to common. It has to do that at 84 HZ with a temperature stable 50% duty cycle.

[Kleinstein]

Yes the Fluke 833 in TVM mode works a little like the 845 and similar meters, though likely less sensitive and maybe without the dividers for the higher voltages. So the Fluke 833 is one instrument with
1) a good reference source
2) a Kelvin Varlay divider
3) a reasonable quality null-meter

Each part is not absolutely highest quality, but still reasonably good. Together this is a good quality DVM of it's time, slow but much higher resolution than a simple analog meter.

I am not sure about using a chopper for the null meter in the 833 - the resolution does not look that high to make it an requirement. Just a zero pot and manual reversal would be enough.

There are not that big chalanges with a chopper - no need for exact 50% duty cycle. It's just a slightly more complicated circuit with quite some parts, that made is expensive back than, when every transistor cost's 10s of bucks. The difficulty was having good switching devices, not such a problem today.  Today you get copper stabilized OPs for a few bucks, that may outperform many of the old choppers.

[Squantor]

What are the challenges in the chopper? Looks like it has to both act as an low impedance input switch and control the transistor that pulls the output to common. It has to do that at 84 HZ with a temperature stable 50% duty cycle.

Well it depends, I am quite charmed of a chopper based on the optofets H11F1 but there seem to be challenges with that. See the circuit that TiN found in his 845. I am not done completely reverse engineering it but I think it has a duty cycle control to make sure that it does break before make and some kind of biasing the chopped side with the trimpot. I am first going to try with a mechanical chopper test to make my design is sane and then move to a H11F1 design.

Here is another story[1] of a person replacing the neon choppers in his hp 419a null detector:

If you move on to H11F instead of one of the other analog switches, beware that they have a significant offset voltage, on the order of 100uV, which Fairchild does not test.  It is affected by temperature and LED drive current.  The modulator pair must be matched to within 15uV.  Age them first, by operating them, heating and cooling them, and soldering their pins, over and over until they stop drifting.  Then look for a pair with similar offset, and glue them together for thermal bonding.

[1]: https://groups.yahoo.com/neo/groups/hp_agilent_equipment/conversations/messages/76318

[FlyingHacker]

I wonder if increased modern line voltage has much to do with frequent chopper failure. At least on the 883 the chopper is powered by the unregulated 6V supply. I was measuring just shy of 9V on the output cap of the 6V supply with my line voltage set to 120V. I can't see a voltage rating marked on the chopper. It is possible removing it from the mounting strap would reveal it, but I don't want to risk damage.

I tapped into the 6V supply to add some amber LED lighting to the meter. I'll post pics in the 883 repair thread. I don't feel bad about doing this since it actually brought the voltage down to around 7.5V.

[enut11]

Quote: "I have no experience with cleaning PCB's affected by corrosion from nicads, only from electrolytic capacitors deciding to spray their contents inside a Philips lab powersupply.
Some traces where completely eaten away and component leads where damaged. Re-soldering every joint would be smart. If a trace is not shiny metal scrub away the oxidation until it shines. I used one of those fibreglass cleaning pens (use gloves! those fibres will get everywhere and mess up your day with itching!) until the trace metal was visible or the trace was completely corroded through. In your case the traces look good, no "crust" of gray or reddish residue.

Also check the transistor sockets if some of the electrolyte has wicked up into the socket messing with the transistors. Remove the transistors and check, clean if needed. Did you power it on before you cleaned up the electrolyte? That might have damaged some components.

Have you measured the waveforms of the 83Hz Oscillator?"

Hi Squantor.
Fixed the PS board. Removed most components and resoldered the pads. Had to dismantle the transistor sockets to clean out some green deposits most likely from the nicads. The 84Hz oscillator now works. Also measured 90vAC RMS on the output of transformer T202 which should be enough to drive the neons. However, still no glow from the neons. Now wondering if the neons are shot?? Still getting unbalanced 15v rails. Mine are +10.2v and -12.7v.
enut11

[enut11]

Wow, somebody did my work. I'm impressed, really. I will be out of the country this week, but will definately remove covers and post more pics next week, no worries.

Hi TiN, did you find the time to make some more detailed photos of your fluke? Also, mine has a lot of germanium transistors that I suspect of having problems but finding replacements is hard. Can you look at the type numbers of your transistors what type they are?

My 845AB also uses germanium transistors. I tested the ones on the power supply board:
Q201 PNP Beta 140, Vf 160mV
Q202 PNP Beta 104, Vf 135mV
Q203 NPN Beta 320, Vf 150mV
Q204 NPN Beta 130, Vf 135mV

Most of the transistors in my unit are the old TO39 metal can type

The 845AB parts list specifies general purpose Si transistors so perhaps any modern transistors will work?
enut11

[Kleinstein]

The circuit should also work with modern Si transistors. Germanium transistors do age, especially if they get hot. Chances are that aged transistors get a higher gain and start to conduct to much, so they get more hot than normal.
In most respects modern Si transistors have better specs. Just the higher U_BE might need some adjustments in some places.

I don't see a problem with Q201 to Q204 in this respect. Q201/Q202  have a pot to adjust the operating point (frequency) already. They also work in switching mode, so trouble with the DC level.

Q203/Q204 might actually profit from the lower CE saturation voltage, but this should not be a problem at the level of accuracy an analog meter provides.

[picburner]

Q106, Q113,Q114,Q203 and Q204 are chopper transistors whith a Vebo>16V.
They can be replaced with silicum type MPSA16/MPSA17 for npn and MPS404A/2N2944A/2N2945A for pnp.
Common transistors don't go well there.
The manual, in part list, say "factory matched" so they have to be selected.

[enut11]

Thanks a lot guys. Gives me something to work with. Looking forward to getting this old test gear up and running.

[enut11]

There is an oddity in the F845AB parts list vs the schematic - they differ in specifying NPN vs PNP for Q201 and Q202. I am taking the schematic to be correct, ie PNP.
enut11

[enut11]

This info came from basher [builder]. It relates to modifying the 845 with LEDs in place of the neons and is something that I will have a look at.

"I got my idea for my 895A (similar chopper circuit to your 845A) from the following,

[volt-nuts] Finally got around to modifying my Fluke 845ab with   LED's
Dallas Smith Tue, 09 Sep 2014 08:57:07 -0700

Finally got around to modify my Fluke 845ab with LED 's for the chopper circuit. Used the 17 volt windings for LED's (Mouser 941-C513AMSNCW0Y0511 Warm White Round LED) instead of the 130 volt, move red wire on transformer pin 9 to pin 7.This winding is 180 degrees out of phase, so I reversed the steering diodes (CR106 & CR107) I left in to help make sure the phase was correct for the LED's when connecting. Change R154 to 6K to set the brightness, selected for good operation of the zero control. Then install jumper to replace C119. Also changed the filter integration response caps C111 to .022uF and C116 to 47uF, this stabilized the jitter to a manageable mode of operation. Meter now works as well or better when the original neon's worked.

As the meter originally had this problem, why is the offset reading different when polarity is reversed at the meter input? About 10uV's.

Lamp Blocks.

However, I used H11F3M  IRED/FET optocouplers and increased R154 as I now had IREDs not neons.  My TVM now has less than 1/2 micro volt jitter and drift, of the TVM, is much less the a microvolt in 20 to 30 minutes.

 I can't recall more as I'm away at present.

Hope this helps, Builder"

[enut11]

Fixed the PS board. Removed most components and resoldered the pads. Had to dismantle the transistor sockets to clean out some green deposits most likely from the nicads. The 84Hz oscillator now works. Also measured 90vAC RMS on the output of transformer T202 which should be enough to drive the neons. However, still no glow from the neons. Now wondering if the neons are shot?? Still getting unbalanced 15v rails. Mine are +10.2v and -12.7v.
enut11

Finally got around to removing main PCB (not easy!). Replaced all electros. Desoldered the neons. They look sick to me. How to test them?

Next step is to try Dallas' LED mod to replace LEDs in chopper circuit.
enut11

[tronde]

Desoldered the neons. They look sick to me. How to test them?

150 kohm series resistor and 100 - 230 VAC should make them glow if basically OK.
I guess yours will be slow to ignite and quite unstable since they have turned dark after many hours of glowing.

New neons are sold on ebay.

[enut11]

Desoldered the neons. They look sick to me. How to test them?

150 kohm series resistor and 100 - 230 VAC should make them glow if basically OK.
I guess yours will be slow to ignite and quite unstable since they have turned dark after many hours of glowing.

New neons are sold on ebay.

Thanks. I have a feeling that the 845AB  neons need certain colour characteristics. Anyway, for the moment, I will try an LED mod and see if that works.
enut11

[enut11]

I have removed the neons, light rods and posts. Changed several components as outlined above from Dallas Smith. Added LEDs to directly activate the LDRs (in the yellow box).

Note that CR106 and CR107 diodes have been reversed, capacitor C119 replaced by a link, LED resistor R154 now 6K, all driven from the low voltage winding on T202.

I am using off-white LEDs that I already had and spent some time finding a matched pair for current and forward voltage.

Fingers crossed.
enut11

[SeanB]

I would not use white leds because of the slow turn off and variability of the phosphor in them. Use a blue LED instead, which will do the same with the fast turn on and turn off, plus it will activate any LDR as fast as possible, but here any LED from blue to green, orange or red will work just as well, as the LDR resistance will work with the red neon lighting of the original.

[Dr. Frank]

LDRs usually are most sensitive in the IR range.

Therefore I would avoid blue, green and especially LEDs with phosphor.

Go for red or IR LEDs instead

Frank

[tronde]

The original neons will most likely have a kind of orange light colour. I don't think they have a selected colour, but the circuit will not work when they become unstable.

[Kleinstein]

There are different types of LDRs. Some are IR sensitive, but the normal ones are not IR sensitive at all, they may not even react much to red light. So an orange or green LED would be the best choice - yellow might be the most effective for the LDR, but yellow LEDs are often less efficient. As the frequency is not that high the phosphor in the white LED is not a problem. The color of the LED should not make a big difference. In theory red light will penetrate deeper and thus give more volume effect, while blue light is more a surface effect. It might be more important to get an even light distribution on the LDRs - gradients in the light intensity could cause some minor offset.

The color of neon lamps is fixed by the neon gas. But this does not mean that this color is best. The problem is more that neon lamps get darker and might start to flicker if old.

[enut11]

Thanks gents. My LED mod did not work but it may not be the LEDs at fault. I am now finding that the output of the 84Hz oscillator is weak so am looking at replacing the germanium with silicon transistors. Then I will try the various coloured LEDs suggested.
enut11

[enut11]

I have had some success with changing the neons (which were dead) for LEDs in the 845AB. I have now gotten a response from the LEDs. I measured around +-12v on the rails (a little low) and I was able to zero the meter on the lowest ranges. The meter responded to a DC input on the 3v range (only one tested). This is a big step forward so am happy.

Only problem encountered is trying to set the 84Hz oscillator - I get a fluctuating frequency that I need to look at. It is not a nice clean square wave and seems to have lower and higher frequencies mixed in. I am not yet in a position of save oscilloscope screens but working on it.

Meanwhile, any suggestions on stabilising the circuit below would be appreciated.
enut11

[Kleinstein]

The LEDs might be a rather high load for the oscillator. This can also make the frequency less stable and explain the lower voltage. So it might be a good idea to reduce the current to the LEDs a little and thus high efficiency LEDs may be needed.
One could also use the supply half of the transformer  - the LEDs don't need 100 V.

[enut11]

Hi. I posted the wrong circuit. The LEDs are powered from the low voltage winding. Will look at reducing the LED current further. Thanks.
enut11

[enut11]

Finally got an oscilloscope screen capture program working. Had to revert to a PC running Win XP and with an RS232 port. I have a TDS220 scope and am using OpenChoice Desktop from Tek. Below is the scope trace for the Fluke 845AB oscillator taken from pin1 of the transformer (TP10). It is supposed to be an 84Hz square wave! As you can see it hardly resembles a square wave nor is it running at 84Hz.

The trace was unstable and I was not able to adjust it to 84 Hz using the 2000 ohm trimmer pot.

Power supplies are +-11.8v (nominal 15v) and the meter is reading input on all ranges, although out of calibration.

I would like to know why the square wave is missing/distorted and how I can get the frequency right.
enut11

[Kleinstein]

The exact frequency is not critical - so the 65 Hz would not be a problem.

But the waveform is really bad - so bad it might damage some parts.

It is well possible that old Ge transistors (Q201,Q202) with to much gain are a problem. Thus range of the 2 K pot is not sufficient anymore to control the oscillator gain. The missing capacitive load of C119 with the neons might be a factor - though it should tend to increase the frequency. It might be a good idea to have the neons connected, even if not used for the chopper. They also act to reduce the peak voltage a little. They also provide some capacitive loading - C119 might be important to set the frequency. Alternative just C119 a series resistor and two zeners / suppressor diode might work.

Also to much capacitance for C201 could be a problem. But I am not so shure of the exact operation of the oscillator.

[enut11]

Hi All. Thanks for your input. Your suggestions have all been worthwhile. I am happy to report that my Fluke 845AB is now working after much time spent on cleaning, troubleshooting, mods and repairs.

This turned out to be my most challenging project as this device has a very complicated physical structure and must be dismantled to get to the main board. I inserted a couple of connectors into the wiring looms to facilitate disassembly and testing.

For the main oscillator, I substituted a couple of switching Si transistors for the Ge. The LEDs for neons mod suggested by Dallas Smith also seems to work well.

Power wise, in the end I decided to run mine on 2S LiPo battery pack as the total meter current draw is under 20mA. There is provision for 230vAC operation later on if I need it.
enut11

[Squantor]

This turned out to be my most challenging project as this device has a very complicated physical structure and must be dismantled to get to the main board. I inserted a couple of connectors into the wiring looms to facilitate disassembly and testing.

For the main oscillator, I substituted a couple of switching Si transistors for the Ge. The LEDs for neons mod suggested by Dallas Smith also seems to work well.

What transistor substitution did you do? Just generic 2n3904/2n3906?

I have troubles with fluctuating voltage rails on my fluke, I still need to revisit my Fluke one of these days. Maybe I am experiencing the same problem, your information/waveforms are going to be a big help. I have already done the LED mod, replaced all the resistors and caps (except for the tantalums and films). This only leaves the transistors.

I wish I had a list of the transistors used in TiN's 845 so I could start substitutions .

[enut11]

Hi squantor. For Q201/Q202 I tried MPS404a pnp [ref picburner] and they worked. I also tried common BC350 pnp and they worked. I suspect the transistors are not critical as long as they are matched. The MPS404a had higher gain (>150) compared to BC350 (<100). 

I could not make sense from some of the waveforms shown in the 845AR manual. For example, TP10 on pin 1 of the transformer T202 should show a square wave. In fact pin 7 on the the transformer secondary shows a beautiful 12vRMS square wave at 84Hz.

My voltage rails are just under 12vDC compared to 15v nominal but since these are unregulated I am not worried as it is probably related to the lower battery voltage (8v LiPo) that I am using.

With the LEDs mod, I elected to mount mine up against the LDRs. Dallas mounted his LEDs where the neons were so he used the original optical rods. I am using efficient LEDs and the current is ~2mA (12v/6K). I am thinking of reducing this to 1mA to see if it has an effect on pointer stability (currently very slowly moving +-3 minor divisions around zero on the 1uV range with a shorted input).
enut11

[Squantor]

Hi enut11, hope your Fluke845 is still open. Could you try something out for me? Can you put it on 3uV mode, short the input with the zero switch and then turn the zero pot fully clockwise and counter clockwise and measure the voltage rails at both settings. Do they differ a lot from when you have Nulled the pointer? Mine fluctuates a full volt and a half on the negative rail. I want to know if that is normal.

Thanks for the transistor tips. Maybe I am suffering from the same problem but not so severe as yours.

Do you get a big difference between line powered and battery powered?

[enut11]

Hi squantor. It is all together now as I wanted to measure performance with the internal shields in place. However I am already thinking of some changes so it wil be apart again soon. I do not recall any bias in the 15v supplies as a result of moving the null control. I will measure this next time around.

I have not powered it with AC mains yet because it requires a different power socket for 230vAC.

[enut11]

I measured the 'total noise' of my fully assembled Fluke 845AB running off a 2S LiPo pack (8.2vDC). I set the output pot to deliver 1v while injecting 10.0v on the 10v range. Once set, the full scale output is 1v on all other ranges. I then shorted the input by selecting OPR-Zero, selected the 1uV range and centered the pointer to zero. The chart below shows a gradual deviation with a small negative slope. Not sure what is causing this but it seems to be linear with time. Temperature over the ~30 min test was 21C, humidity low (winter).

The data from the 845AB was measured with my MS8040 and sent to my XP RS232 port using the Mastech PC program.
enut11

[enut11]

Hi enut11, hope your Fluke845 is still open. Could you try something out for me? Can you put it on 3uV mode, short the input with the zero switch and then turn the zero pot fully clockwise and counter clockwise and measure the voltage rails at both settings. Do they differ a lot from when you have Nulled the pointer? Mine fluctuates a full volt and a half on the negative rail. I want to know if that is normal.

Thanks for the transistor tips. Maybe I am suffering from the same problem but not so severe as yours.

Hi Squantor. Rails tested as requested, 3uV range:
Zero pot centered: +11.28v and -11.14v
Pot fully clockwise: +9.42v and -11.31v
Pot fully anticlockwise: +11.55v and -9.45v

So yes, the zero pot does affect the power rails, the further you turn it the greater the deviation.
enut11

[Squantor]

Hello enut11.

Thanks a lot! Well it saves me from a possible chasing a problem that is not actually a problem. Hopefully in the summer I have time to revisit the 845 and finally fixing it :-).

[TiN]

Doh, I replaced carbon (which were 5-10% high) resistors with modern ones in my 845AB, and now unit is dead
It was working before, just noisy, but now it's just overflow either way and just stays there, on any range.
Battery is ok +9.4V, +15V and -15V are ok too.

[2N3055]

Doh, I replaced carbon (which were 5-10% high) resistors with modern ones in my 845AB, and now unit is dead
It was working before, just noisy, but now it's just overflow either way and just stays there, on any range.
Battery is ok +9.4V, +15V and -15V are ok too.

I  know you know it but did you check for oscillations? Carbon resistors are noninductive, low parasitics... Horrible noise and tempco too;-)

[enut11]

Hi TiN. Have you checked the 84 Hz oscillator?

[martinr33]

Started looking into my 845AB. It is reasonably stable, but it could do better.

Bottom line: The FET opto isolator, something to tighten up the opto drive, and then removing that

1) This is an AC circuit!
The meter reading comes off of a transformer, and the chopper amplifier circuit shifts the center point of the waveform. The chopper frequency makes it all the way through to  the display; the meter is averaging the output waveform. Very clever stuff that no doubt has its roots in vacuum tube designs. This architecture implies that we can't add a chopper amp in and get anything useful. That said, there is a DC segment, but it is wrapped in that AC technology.

2) That new board is simpler than it looks. It uses very good FET optoswitches to replace the LDRs and neons. It should be more consistent, because there's no dependency on the randomness or deterioration of a neon tube.  You could come off of the transformer, but I think that Fluke wanted the extra control of a regulated voltage supplying the optos. The FET optoswitch is absolutely the way to go.

3) Looks like U3 (and thanks to TiN for the pictures, and Picburner for the schematic) simply provides a DC voltage to offset the zero in the front end chopper circuit. Oddly, all the other front end wiring is carefully isolated with standoffs and air wires. But this little offset circuit is just plain PCB wiring.

4) I'm not sure how important asymmetry is in the optoisolators. Mostly, I see trouble in the switching threshhold, as asymmetric switching will offset the zero. The input shaping that Fluke added should keep that to a minimum.

5) Looking at the input, if the 1M resistor is disconnected, the only DC path is through a dark LDR. Again, the opto FETs should help hereby eliminating the leakage path through the LDRs. Another point for the AC pathway in a DC instrument!

6) If anybody does get into an assist PCB for this, a small microcontroller could do a pretty good job of autozero. It would need a comparator to read the zero, and a way to slightly move the offset voltage in the front end. Could probably use pwm.

[Squantor]

Hello Martin R,

My 845A is lying mostly disassembled in a corner, looking at me, whispering "why am I not repaired yet". I wanted to replace the neon+LDR chopper for a long time but I am probably going with a different route.

I have a bunch of equipment that either uses mechanical or LDR+Neon choppers to create the autozero functionality. I am investigating a replacement for them all by using a small custom wound transformer (using a high Al pot core with shielding) and optofets to recreate the chopping function. The reason I am for a small custom wound transformer is that I want to reuse the neon drive windings but the current is too low on those. The optofets need a substantial current to really go into low resistance mode (30ma according to the datasheet of the H11F1).

Have you reverse engineered the optofet chopper PCB that was posted here? I am interested in the residual zero trimming part. So the 5V regulator is there just to provide a DC bias so the differences between the optofets is canceled?
This might help that you dont need to match the optofets so tightly.

[TiN]

My 845A is lying mostly disassembled in a corner, looking at me, whispering "why am I not repaired yet".

Mine already gave up whispering... just looking with one remaining eye...slowly getting covered by layer of dust. 

[Squantor]

Mine already gave up whispering... just looking with one remaining eye...slowly getting covered by layer of dust. 

One of your transistors had some broken off pins, right?

[TiN]

Yes, but after fixing that something else got broken, as after assembly it wouldn't work at all.

[Squantor]

I found the Fluke 845 a pain to work with, how it is put together is annoying and reaching the various parts or replacing them requires a lot of work.

I recently repaired a hp419A, it is much easier to take apart and reach all the components. It is now my main null detector and it has a new rebuilt batterypack too. I only have the 845A version and I have noticed that to get low noise levels a battery pack is very useful.

[Conrad Hoffman]

Yeah, I've got a couple 419s and they're really a pretty decent null meter. They're on the shelf right now with various problems, and I've been using the 845 for quite a while. I think its performance has deteriorated, as the output is very noisy. Have to decide what to fix, as the 845 is such a pain. One of my 419s developed leakage between the front banana jack and the panel, probably some corrosion in the hole. Not sure what shape the batteries are in- I rebuilt the pack, but it's not been charged in a couple years. Probably dead and gone. Sometimes Fluke just has a special knack for mechanical design... not!

[Vtile]

Every poor 845s can be send to me, so I can expose them properly. 

[martinr33]

I looked at the posted schematic, and TiN's pictures, to understand the circuit. .

You could use the other AC winding to drive the optofets. Fluke drove their board off of the rectified DC supply that also uses this winding, so we know there is plenty of current.

The 5V regulator on the Fluke update board is the power source for the optocoupler emitters. The optocoupler photodiodes are controlled by the transistors on the board, driven off of the AC feed through an optoisolator. This design should make  the chopper really consistent, and more independent of input voltage shifts.

For a minimalist solution, you could replace the LDRs with the optocoupler transistors, and drive the emitter side from the 12V winding through a 500 ohm resistor. Wire the emitter diodes back to back, and you only need one resistor, as each diode will only conduct on its half cycle.  This approach will be well superior to neon or CdS cells, even without the regulator circuits.

The residual trimmer is surprisingly simple - it just feeds a little current into the zero side of the chopper input. I am thinking it is there to deal with coarse differences between the optoisolators.   



"Hello Martin R,

My 845A is lying mostly disassembled in a corner, looking at me, whispering "why am I not repaired yet". I wanted to replace the neon+LDR chopper for a long time but I am probably going with a different route.

I have a bunch of equipment that either uses mechanical or LDR+Neon choppers to create the autozero functionality. I am investigating a replacement for them all by using a small custom wound transformer (using a high Al pot core with shielding) and optofets to recreate the chopping function. The reason I am for a small custom wound transformer is that I want to reuse the neon drive windings but the current is too low on those. The optofets need a substantial current to really go into low resistance mode (30ma according to the datasheet of the H11F1).

Have you reverse engineered the optofet chopper PCB that was posted here? I am interested in the residual zero trimming part. So the 5V regulator is there just to provide a DC bias so the differences between the optofets is canceled?
This might help that you dont need to match the optofets so tightly."

[Conrad Hoffman]

I don't know if this trick is applicable here, but when I use LEDs for something where neon bulbs are used, I try to make them turn on in a similar manner. That is, a neon bulb only turns on and conducts suddenly when you exceed 90 volts or so. With a sine wave, the bulb may only flash near the peaks. Scaling everything down in voltage, I'll put a zener in series with the LED so it turns on later in the driving waveform. Two LEDs and zeners if I need bipolar operation like a neon. The pairs also protect each other from excess reverse voltage. I do this with strobe lights for turntables to get a shorter flash and better resolve the bars. No idea if the technique is useful for choppers, but throwing it out there if needed. If the lamps are driven from a square wave, it shouldn't be, but with a sine wave it might help.

[Squantor]

I don't know if this trick is applicable here, but when I use LEDs for something where neon bulbs are used, I try to make them turn on in a similar manner. That is, a neon bulb only turns on and conducts suddenly when you exceed 90 volts or so. With a sine wave, the bulb may only flash near the peaks. Scaling everything down in voltage, I'll put a zener in series with the LED so it turns on later in the driving waveform. Two LEDs and zeners if I need bipolar operation like a neon. The pairs also protect each other from excess reverse voltage. I do this with strobe lights for turntables to get a shorter flash and better resolve the bars. No idea if the technique is useful for choppers, but throwing it out there if needed. If the lamps are driven from a square wave, it shouldn't be, but with a sine wave it might help.

I have actually did some tests with replacing the mechanical chopper of a GM6020 DC millivolt meter from philips. As a mechanical chopper does not switch instantaniously from signal to the ground, using a zener to introduce a similar dead time is a good idea, not that the chopper worked that well but I did not pay that much attention to matching and the meter needed some some repairs first. I always wondered/worried about possible cross conduction creating a large offset.

If you look for instance at the original neon circuit of the Fluke 845A(B) is that they use a capacitor in series. So it is no hard turn on or off, but is this to momentarly boost the voltage so the neon ignites?

[Kleinstein]

The oscillator might depend on the capacitor used to couple the neon tubes. At least one can expect the neon tube part of the circuit influence the frequency and amplitude. I would expect the waveform to be something in between a sine and a square, so some dead time for the neon tubes, but not that much.

With modifying the circuit (replace the neons) it is a question on how much of the original circuit one would really like to keep. Keeping much of the circuit would be think of keeping it as an historic instrument. In this case the logic path would be to get some new neons.

I don't think that even with replacing the LDR/neons with a photo-mos or similar parts would give really good performance. It is not just the neons that are outdated. I new design based on an AZ OP can likely give better performance and could be low power battery operated - just the analog output would be a little more difficult and power hungry.

[lukier]

I new design based on an AZ OP can likely give better performance and could be low power battery operated - just the analog output would be a little more difficult and power hungry.

Do you have any example schematic of a modern null-meter? I would rather DIY something instead of buying antiques. Every time I'm thinking how it would work I end up with some sort of a multimeter-like front end (ranging resistors -> op amp comparing that vs fixed reference and outputing zero if equal, all battery powered). I don't see where would the very high impedance came from - it would have the impedance like multimeters, 10M or whatever the setting of the input divider. 

[Conrad Hoffman]

The 845 is nice because of the high input impedance, but remember that in a true null application it often isn't necessary. After all, with zero voltage difference, current approaches zero, even if the meter is low impedance.

[lukier]

The 845 is nice because of the high input impedance, but remember that in a true null application it often isn't necessary. After all, with zero voltage difference, current approaches zero, even if the meter is low impedance.

I see. So, for example, if I want to compare 2 references and connect them back to back, then a normal DMM is fine to measure the voltage difference?
Also, Fluke 5440B manual asks for a null-meter to calibrate some of the ranges (with the help of Fluke 752A - I'll try to DIY a Hamon divider instead, thanks to your website  ), but again this is around zero comparison (divided reference vs calibrator) so a decent DMM should be fine? (and on these low ranges the impedance is huge anyway).

I guess that explains why null-meters are rather extinct species, so what are the modern uses where one is still required?

[Vtile]

It were known early on as poggendorff compensation method.  ...Well similar current cancellation method that is.

[Dr. Frank]

The 845 is nice because of the high input impedance, but remember that in a true null application it often isn't necessary. After all, with zero voltage difference, current approaches zero, even if the meter is low impedance.

I see. So, for example, if I want to compare 2 references and connect them back to back, then a normal DMM is fine to measure the voltage difference?
Also, Fluke 5440B manual asks for a null-meter to calibrate some of the ranges (with the help of Fluke 752A - I'll try to DIY a Hamon divider instead, thanks to your website  ), but again this is around zero comparison (divided reference vs calibrator) so a decent DMM should be fine? (and on these low ranges the impedance is huge anyway).

I guess that explains why null-meters are rather extinct species, so what are the modern uses where one is still required?

Are you aware, that the 5440/42 are Autocal /artefact calibrators in reality?
Once calibrated by a 752A for all 4 direct volt ranges (10, 22, 250, 1000V), the gain constants will not change greatly (old instruments), and the internal gain calibration is fully sufficient.
Only the 10V basic calibration might be necessary, only a small drift is to be expected, but the divided 2.2V/ 220mV ranges need frequent re-calibration.
carefully read the addendum of the service manual..

By the way, nobody here, or in the volt-nuts mailing list has ever checked the real leakage / bias current of this famous 845A.. it's not specified in this sense, so i think it's overrated.
Frank

[lukier]

Are you aware, that the 5440/42 are Autocal /artefact calibrators in reality?
Once calibrated by a 752A for all 4 direct volt ranges (10, 22, 250, 1000V), the gain constants will not change greatly (old instruments), and the internal gain calibration is fully sufficient.
Only the 10V basic calibration might be necessary, small drift expected,, and the divided 2.2V/ 220mV ranges need frequent re-calibration.
carefully read the addendum of the service manual..

Yes I know, but well I got my 5440B rather cheap and it had a lot of issues. The most minor one is that it is missing the Display Board PCA, so while I'm designing a replacement out of 16 segment green LED displays, shift registers (digits), small MCU and 20x2 LCD (text display), I'm using it via GPIB only.

The real problems were that the +-17 inguard supply was dead, due to the power resistors - one of the 1K balancing resistors (R22) for the +-17V. While I was repairing that I've noticed that 0.15 Ohm resistor for 5V outguard looks like it has seen the better days.

After repairing that, cleaning the relays with isopropanol and pieces of paper and replacing all the electrolytic caps except the huge Sprague ones on filter boards. I've picked quality caps, min 105 deg C (originally there was a lot of 85 deg C) - this machine thermal management is ridiculous, everything runs super hot, not only the ovens and one finds darkened spots on PCBs a lot.
Then I discovered (thanks self-tests!) the unit had more problems - around Analog DAC PCA.

The voltage reference was very low - it turned out the DAC wanted to draw ~100mA from the reference, loading it. I quickly removed the Vref board from the oven and provided 13V from a linear PSU when debugging not to stress these poor SZA263 anymore.

Long story short, it turned out Q9, Q10 and Q12 JFETs failed on short or open. Some NPN (AFAIR Q13) also failed. Unfortunately, Q9, Q10, Q29 and Q30 are rather special matched set.
I've picked AFAIR J107 JFETs and matched the parameters (thanks nifty transistor tester!) and on resistance as good I could (Fluke specifies within 1 Ohm) and soldered them. Very difficult to get JFETs these days. I've replaced also the NPNs with the same part - 2N2369A.

I bet it was the Analog DAC PCA that has taken out the +-17V supply.

Now all the self-tests pass without any problems. Therefore there are few things left for me to do:
- finish display board replacements,
- measure the huge filter boards caps, I hope these don't need replacement as they are extremely expensive,
- replace the fan with brand new EBM Papst 4800Z (already bought),
- calibrate!

For the last point I need to finish my LTZ1000 based 10V standard first and maybe someone from the calibration club thread, with a recent Volt at home, can help me with that. Once I'll bring 10V home I'll try to run only the internal calibration, but I'm prepared that after such repairs the full calibration might be necessary - therefore the need for 752A or DIY alternative. Also, for the 2.2V/220mV ranges.

[Edwin G. Pettis]

Dr. Frank:

I have to disagree with you, null meters was/are still are an important instrument, I still use them and I know others who also use them.  There is a procedure in the Fluke manual, paragraph 4-22 which tells how to measure the leakage resistance and at null, a properly operating null meter does achieve an input resistance of near infinity at balance, no DVM or auto-zero Op Amp is going to get anywhere near that plus there is more bias currents from the DVM and auto-zero op amps than a null meter at balance.  The null meter is the superior instrument for use with bridges, I use them in my lab all the time, all of my bridges use null meters, whether on board or external.  They can achieve a voltage measurement of as little as 10nV resolution, a 3458A cannot do any better than 50nV resolution and the null meter is continuous unlike the DVMs.  I've tried using DVMs in place of null meters, not near as good, I've used auto-zero op amps for null detectors, good but still lacking a bit.  I even use a vacuum tube null detector for special purposes which actually outperforms the solid state units.

I have several 845 units and two 419As, while the 419 is a bit easier to work on inside, I still prefer the 845, for a few years, ESI put the 419As in their 242 bridges (mine has one, it hasn't worked in some time), ESI dropped it (unknown reason), I'm using a Keithley 155 as an out board detector right now.  If you're wondering, I simply didn't have time to fix the 419s and 845s are all busy.

A null meter can make a world of difference when attempting to make such sensitive measurements, forget the DVMs and put the null meter back where the procedures call for them.

[Conrad Hoffman]

OK, drifting far OT and out to sea, but it's also amazing the sensitivity of measurements made in the past using mirror galvanometers and telescope or projection methods. A true null meter, though not very high Z by todays standards.

[Kleinstein]

A mirror galvanometer can be really sensitive to low DC voltages - no problem measuring thermocouples and similar. The big advantage over an electronic solution like the Fluke  845 is that there is definitely no bias current.  Usually the input impedance is not that high - but this may not be such a big problem with a null meter. It is only a problem before getting the adjustment right. With a galvanometer noise limit and impedance are coupled - electronic amplifiers can get better than that limit, at least at not so low frequencies.

The 845 still uses a kind of chopper amplifier at the input. So one has to expect some input bias current, though not that much. At least expect something like input offset (without compensation) times input resistance. Even the LDR can create some offset / bias, due to unwanted photovoltaic effects.
A chopper stabilized OP in a non inverting amplifier configuration would have a similar (or like higher than the LDR version) high input impedance in the > 10 G Ohms range. Not many AZ OPs seem to specify input impedance. For the ADA 4522 (low noise rather high bias) a value of 100 GOhms (common mode) is given.

This would apply to low voltages like up to maybe 2 V. Only higher voltages would need an input divider and would thus be lower impedance, this also applies to the 845 !. A tricky part might be to add protection to ensure high impedance also with higher voltages. Just a series resistor like in the 845 is a compromise toward input noise.

To get a very low bias, one might want to combine a modern Null-meter with an TIA to measure pA range currents as well, so one could adjust / check the bias.

[martinr33]

On the 845 and impedance...

The 845 is an AC design. It is bizarre by modern standards, but it is what you did with awful transistors - and it works well. The input comes through a capacitor, and the output - to the meter - is through a transformer. The display meter does rectification and averaging to give the result. There is a DC section in the middle of all this, but it carries the AC signal.

The input impedance is mostly going to be the DC leakage through the two LDRs (one of them is always on). Then, there's an AC component to the input impedance through a 0.1uF capacitor that is grounded through a 9M resistor. At 86Hz, this won't be much.

Aside from the ranging resistors, there's not much in this unit that matters from a precision perspective. I'd suspect that the neons are the biggest problem, followed by noise in the front-end transistor.

It would also take a really high input voltage to do any damage. I think you could put 1kV in on the 1uV range, and not see problems. The manual warns against doing that if you disconnect the 1M resistor in the input stage, need to figure out why. When the input is at zero, there's no path for current except for LDR and capacitor leakage. Upgrade those parts, and the null impedance is enormous with a well-protected input.



Good upgrades that preserve the philosophy, and the very fine chassis, of the instrument would be:

        - Optocouplers for the chopper, with a clean drive
        - Better front-end transistor
        - Digital 86Hz generator
        - improved DC gain stage (precision op amp) (or maybe rengineer the whole chain)
         - maybe upgrade the input cap

Could also do a simple optical output, driven off of the output square wave. The on/off ratio would indicate the null point, and maybe even the percent of range.


The AC design; environmental and electrical isolation; precision divider; the isolated output; would all be preserved. 

[TiN]

Sounds like a great few year long project  .

[Kleinstein]

The 848 has good input protection through the dioden CR101 / CR102 and the 150 K resistor (R110).

The amplifier it not that unusual, its a chopper amplifier with DC feedback. The unusual part is having the output with an extra isolated stage that uses the same chopper frequency, but starting from the DC signal at TP9. For me, the only thing odd is they way the voltage for the DC offset adjustment are generated / taken from the circuit.

Leakage through the LDRs is attenuated by the loop gain of the amplifier. So it is reduced quite a lot.

There is an AC input resistance for the AC amplifier (1.2 M + 500 pF to GND and the input current to the transistor) that would act back on DC through the chopper. Like the LDR leakage it is also attenuated from the loop. However an loop internal offset could cause some bias current. If needed one should be able to add an adjustment for the offset of Q107/Q108 and thus the input bias. I might be interesting to know the input bias - from the circuit I would not expect it to be that exceptionally low (maybe 10s of pA range).

I am not so sure that building an chopper amplifier from discrete parts would be needed to day. At least one could do better today, especially by using an JFET instead of Q101, as this could give lower noise and higher input impedance (resulting in less bias).

[martinr33]

The Fluke opto upgrade added a circuit for DC offset into the low-side optofet (well, a pot connected to 5V). So they clearly agree with your diagnosis!

Although the "output" amp is DC coupled, the AC signal carries through the whole circuit. The use of the chopper signal in the last stage is interesting. They are effectively resquaring the signal. Still amazed by that use of the meter movement to average the AC output.

[Kleinstein]

I don't see the signal going all the way through as AC. As far as I understand the circuit, at TP9 the signal is DC (following the input) and the part around the transformer is just a kind of DC/DC converter or isolation amplifier with reasonable (good enough for an analog meter) accuracy. It is only a kind of convenience / saving parts that both parts use the same clock - today one would ideally use a something like 2 times the clock for the isolation stage (and the power supply), so that there would be less possible interaction.
So we can treat those two parts separately. If wanted one could replace those 2 parts separately with maybe an AZ OP and a different type of isolation amplifier. Just for the meter, I don't think one would actually need the isolation. The isolation is mainly for the recorder output.

There are two places for an offset adjustment. One is for a possible offset voltage at the input, like done with R117. However that can be a second place where an DC offset can occur: That is in the DC amplification part from Q107/Q108 on. A DC error in this area (e.g. from mismatch of Q107/Q108) would result in requiring a correction signal from the chopper stage and thus a not fully vanishing AC signal at the input when the meter reads 0 V. The DC error can be corrected via R117, however this would also cause an Bias current. So ideally one would have a second adjustment for the DC stage (e.g. make R148 or R150 adjustable), so that one could adjust the local error of Q107/Q108 here and this way have adjustment for offset voltage and bias current. I don't think the opto-upgrade added a separate trimmer for the DC stage. As far as I understood it they just replaced the adjustment via R117.

[Dr. Frank]

Dr. Frank:

I have to disagree with you, null meters was/are still are an important instrument, I still use them and I know others who also use them. 

Mr. Pettis, I did not make any statement about the importance of null meters.

I simply referred to the fragmentary specification, which lacks the bias current of this instrument.

This instrument is mystified in the metrology community, also by its 2nd hand price, like having 'infinite resistance', and that only THIS instrument can be used in the commonly known bridge arrangement, like with standard cells, or precision dividers as 752A, 720A, and also for the calibration of the 5450A.

There is a procedure in the Fluke manual, paragraph 4-22 which tells how to measure the leakage resistance and at null, a properly operating null meter does achieve an input resistance of near infinity at balance, no DVM or auto-zero Op Amp is going to get anywhere near that plus there is more bias currents from the DVM and auto-zero op amps than a null meter at balance.
 The null meter is the superior instrument for use with bridges, I use them in my lab all the time, all of my bridges use null meters, whether on board or external.

Sorry, Mr. Pettis, that is exactly this false mystification, overrating, or simply a misconception about the 845A, that I want to deflate.

I refer to the original manual of the 845AR from about 1975, as the online versions are incomplete.
The specification on page 1-1 only mentions :

"INPUT ISOLATION
Better than 10^12 ohms at less than 50% r.h. and 25°C regardless of line, chassis, or recorder grounding... with driven guard, isolation improves .. to 10^13 ohms."

This only parameter about high ohmic behavior, is further described on page 4-5, in paragraph 4-19 'Leakage Resistance Test', with Fig. 4-10.

It is evident, that this parameter and associated test describe the isolation resistance between the chassis ground, recorder ground and power ground, which are all shorted together, versus the Guard and Null- Input /  Common only.
It makes absolutely no statement about the bias current of the Null-amplifier-Input, nor its input resistance.

The null-amplifier also does not achieve 'near- infinity resistance at balance'.. that statement I've read very often, and it's wrong, or simply not appropriate.
Instead, the 845A input resistance is simply a fixed 10MOhm in the low voltage ranges, from 1mV down to 1µV.. and that's exactly so specified in the manual.

What you obviously mean instead, that in the case of perfect balance, there will not flow any current across the bridge-legs, as zero Volt divided by 10MOhm gives zero current..

Speaking about 'Infinite Resistance' is simply not the correct parameter to describe 'zero loading' of the bridge.
Instead the flowing cross-current, created by the possible bias current of the null amplifier, has to be considered.
Well, and this important parameter of all things is NOT specified.

Any other FET chopper amplifier, like in DVMs, or also in the 7650/7652A OpAmps create bias currents on the order of  several pA, up to 20..50pA.
This can be measured quite easily, also on the 845AR/AB.
Maybe, that this photo chopper principle is really superior, but this is to be proven, not to be assumed.

A 2nd remark :
The 845A draws current for the case of non-perfect balance, which is a used case in bridges, when you want to monitor differences and their drifts, like a standard cell versus a fixed 1.018V output.

If you measure 1mV difference, that would at least give 100pA, plus the possible bias current of the chopper.

In comparison, a 3458A draws 20pA at most,  for all voltages between 0V and 10V.
For this used case, of differential mode, modern DVM are superior over the 845A.

  ... They can achieve a voltage measurement of as little as 10nV resolution, a 3458A cannot do any better than 50nV resolution and the null meter is continuous unlike the DVMs.  I've tried using DVMs in place of null meters, not near as good, I've used auto-zero op amps for null detectors, good but still lacking a bit.  I even use a vacuum tube null detector for special purposes which actually outperforms the solid state units.

The 845A can't resolve 10nV, but the 3458A resolves very well 10nV, in its 1V and 100mV range.
 I think, you mix up here a different parameter, and that's noise.

Low noise nV - meter,  do not have necessarily very low input bias currents, no, in contrary!

There exists analogue NV meter, but also nV DVMs, like the HP34420A, and the Keithley 182A/2182A.
These are superior over the 3458A in aspect of noise, but they both have higher input currents.

I have several 845 units and two 419As, while the 419 is a bit easier to work on inside, I still prefer the 845, for a few years, ESI put the 419As in their 242 bridges (mine has one, it hasn't worked in some time), ESI dropped it (unknown reason), I'm using a Keithley 155 as an out board detector right now.  If you're wondering, I simply didn't have time to fix the 419s and 845s are all busy.

A null meter can make a world of difference when attempting to make such sensitive measurements, forget the DVMs and put the null meter back where the procedures call for them.

It's fine, that you own these old instruments, I also would like to have one, as these have some advantages over many DVMs, like high isolation, guarding, high common mode operation, and battery mode.

But anyhow, your appreciation for them is not appropriate.
Otherwise, these instruments would still be in production by FLUKE (and the other manufacturers), as FLUKE still sells bridge type instruments, like the 720A, 752A, and so on.

Fluke themselves instead promote their 8508A, and prove, that its performance as a null-meter in such bridge application can very well replace these old analogue instruments.

The 3458A also makes a very good null-meter.
It has a guard also, and the Isolation Resistance to case ground is also 10^12 Ohm, like the 845A.
It is specified having < 20pA bias current, which the 845A is lacking.
I determined this bias several times in my bridge applications really being less than 10pA, depending on the common mode voltage used.

It also has very low noise down to 0.01ppm @ 10V range, that is on the order of < 100nV rms. The 0.1V and 1V ranges deliver 20nV noise, only.
Please visit our DMM noise comparison test here in the forum.


These parameters are fully sufficient for the different bridges that I had to calibrate.

Maybe an even lower bias current, and the battery mode like the 845A would give an improvement.

So I would like to ask you, please really measure the bias current of your 845A, and present your results here.


Frank

[SZA263]

Instead, the 845A input resistance is simply a fixed 10MOhm in the low voltage ranges, from 1mV down to 1µV.. and that's exactly so specified in the manual.

Instead the flowing cross-current, created by the possible bias current of the null amplifier, has to be considered.
Well, and this important parameter of all things is NOT specified.
...
So I would like to ask you, please really measure the bias current of your 845A, and present your results here.

Frank

The input resistance of the '845' made after 02/1979 is 1M not 10M. Please see Fluke Change/Errata information issue No: 2 2/79 #6282.
Is easy to measure the bias current of the 845: with open  input, the voltage developed across the input resistance (R104-1M) is due to the bias current flowing in it.
Connect a polystirene 1nF cap (with very clean legs!) across input posts, common linked to guard, to reduce switching artifact.
Accurately zero the instrument on 1 µV range, and wait few minutes to thermal stabilization.
Switch to "OPR" and allow a little time for stabilization.
Now you can read the voltage on the meter: is a little hard to read the result but all my 3 '845' (2 AR & 1 AB) agrees that this is well under 100 nV.
So the input bias + leakage current should be well below 100 fA (yes, femtoampere!)

[Dr. Frank]

Instead, the 845A input resistance is simply a fixed 10MOhm in the low voltage ranges, from 1mV down to 1µV.. and that's exactly so specified in the manual.

Instead the flowing cross-current, created by the possible bias current of the null amplifier, has to be considered.
Well, and this important parameter of all things is NOT specified.
...
So I would like to ask you, please really measure the bias current of your 845A, and present your results here.

Frank

The input resistance of the '845' made after 02/1979 is 1M not 10M. Please see Fluke Change/Errata information issue No: 2 2/79 #6282.
Is easy to measure the bias current of the 845: with open  input, the voltage developed across the input resistance (R104-1M) is due to the bias current flowing in it.
Connect a polystirene 1nF cap (with very clean legs!) across input posts, common linked to guard, to reduce switching artifact.
Accurately zero the instrument on 1 µV range, and wait few minutes to thermal stabilization.
Switch to "OPR" and allow a little time for stabilization.
Now you can read the voltage on the meter: is a little hard to read the result but all my 3 '845' (2 AR & 1 AB) agrees that this is well under 100 nV.
So the input bias + leakage current should be well below 100 fA (yes, femtoampere!)

Great, thank you for these measurements..
Do you have the old neon choppers inside, or already the latest Opto-Fets?

How do you measure these 'switching artifacts', you obviously mean the switching spikes / glitches, and how big are these?

Frank

[SZA263]

Hi,
 I have a old neon, but I'm working to switch one unit to optofet, with a simpler circuit than the original Fluke (why use optocouplers on control side when they share the same input/output Ground connection?)
Yes I mean spikes. I have measured them with a Tek 7A22 amplifier, 10 µV/div, filtered DC to 1 kHz.
With inputs connected with a low capacitance cable, they have a peak value of about 10 µV (+-) with a fast risetime and an exponentially decay falling edge. The duty cycle is about 1-2%. The frequency is clearly that of the internal oscillator.
During measurement the two input impedance are in parallel so the result is 500 kohm//~100pF. After the transient the input return to 0 V, without any visible offset/histeresis (at least visible on a scope).

Adriano

[MisterDiodes]

Dr Frank:
You need to know that Fluke still does maintain 845's and even supply them refurbished-as-new when required when they have a very lucrative support contract - In a Fortune 100 cal room working on the latest laser diodes and quantum devices we use analog null meters for precise, quiet measurements all the time.  This is an upgraded 845 and not the old neon-bulb stuff on eBay, and that's why you don't find these for sale.  Keithley's will be maintained to new condition also when the price is right, it just depends on the customer.  The fact is that no DMM can replace a null meter in every case. 

When time is money:  With analog null meter you can get a valid measure in the some 10nV range or lower several times while you're waiting for a 3458a to Autocal, and no squiggly charts and spreadsheets required.  And the fact is 10nV on a 3458a is down in the bottom layer of mud anyway - not exactly high confidence or accuracy.  Many times we have to resort to an analog null meter to remove the noisy DMM from the measure - that's when a quiet null meter is the far superior tool.  Especially when you're trying to measure a few photons at a time on some exotic diode detector.  We do that all the time.

For a complete, accurate measurement, you use ALL the tools and techniques at your disposal to lower uncertainty and raise confidence.  We have 3458a's, null meters, 732's, 752s, 720a, RV722, automated switch gear all sorts of other precision tools - new and old equipment working side-by-side every day.  Use what is required to get the best possible measure.

When comparing null meter function:  yes a 3458a will work as a null meter, but a real analog null meter works best as a real analog null meter, and gets you an accurate, quieter result much faster if you know how to use one. 

IN essence:  A null meter is essentially a virtual near perfect two-terminal device at null, and at null most of the interior circuitry is balanced out, current offsets should be balanced out to zero, no energy is consumed driving a balanced analog meter needle at zero, its DC amp section should be doing absolutely nothing at zero, and at most there is a slight amount of energy coming from the input chopper switches gate capacitance (maybe a few pF driven by a fairly slow rise time chopper in the case of '155 - but that is mostly shunted in the input filter caps).  There shouldn't be many "meter" electrons wanting to flow anywhere at null, and virtually zero emitted EMI from the meter...unlike a relative VERY noisy DVM with running CPU's, oscillators,  input amps with bias current even at zero differential, noisy display drivers, power supplies and all that rot.  It's not the 3458a's fault, it's just always a consumer of energy and emitter of noise whenever it is powered up - null condition or not.

By the way - We can upgrade those analog null meter input chopper P-fets for even less leakage when they switch off - down to a few pA with newer parts from Linear Systems.  You really can get some great performance out of these rigs and they really hold their own against any DMM, even today.  It just takes a few upgrades.

For certain: Analog null meters in some shape or form are an absolute essential piece of gear in any serious cal room, and for some measure tasks where you need absolute quiet: For some applications they run circles around any DMM, but we love them all!

[Kleinstein]

The 845 circuit, both in the old neon and the newer photo-fet version does not look like it is low noise or super low bias. For the noise, already the 150 K resistor in series with the input gives a considerable unavoidable noise. Another likely significant noise source is the input transistor (BJTs don't get a very low noise figure). From these 2 sources alone the input noise is expected to be at least in the 50 nV/sqrt(Hz) range. So not a really low noise.

Due to the chopper operation there is some AC superimposed to the input - seems to be visible despite of filtering.

If the later stages of a chopper amplifier and the charge injection are not well adjusted, effects like an offset of the DC stage will cause an bias current - due to the not so high impedance AC amplifier this could be significant. At least in the circuit shown further up in this thread I see no adjustment for the input bias current. So far I have some doubt the 845 could compete in noise * bias current with modern Az OPs.

A null meter can have its justification. However the 845 circuit does not at all looks like a really good solution, definitely not noise/drift wise and likely also not with bias. Depending on the application low bias current or low offset / noise would be more important. So one can not expect a single meter type to fit all. I am not so sure an analog reading is better than a well made digital display. Analog filtering tends to be slow in settling - digital low pass filtering could be of FIR type and thus superior in settling. Modern µCs with everything inside the chip should make it relatively easy to avoid excessive EMI. The 845's chopper is also not that quite. Really quiet would be the old days mirror galvanometer - lower impedance, but no bias by design.

[doktor pyta]

Kleinstein,

I can make some measurements using 845A if You describe the setup, I have two in the lab.

P.S. the input transistor of 845A have really high hFE (selected part). I recall values like 500 or so.

[MisterDiodes]

So just for fun I took a Keithley '155, zero'ed it out and let it sit on zero check all night on 1uV scale...and noted a slow gentle  drift of around 50nV or so around zero, but just about no overall drift overnight.   The spec on this meter is a conservative 500nV per 24 hrs, but we've never seen one that extreme.

TEST:
So just now I took a metal film 1Meg 1/4W resistor and laid that on the '155 inputs, zero'd out the meter then switched to Operate - and once it settled down noted an apparent bias on the needle of maybe 600 nV, mostly resistor noise.  This particular meter could also use a touch up adjust on internal current cancel pot but let's just say it's in an "everyday" condition.  I think if we're conservative and called it 1pA bias current, the meter would -definitely- be below that.

Take the same resistor and attach directly to the inputs of an in-cal 3458a, DCV mode, Auto Zero ON, 100 NLPC and noted about 10uV apparent offset, very repeatable. So around 10pA apparent bias if my math is correct.  In other words the 3458a is throwing in an order of magnitude increased bias current at around zero volts input differential voltage.

Remind me again why a DMM is supposedly so much better as a null meter with less bias current & noise than a working analog unit? I'm looking at the two side by side as I write this.

Agreed - 3458a's make very good VOLT meters, but for detecting a near zero-volt condition, an analog null meter is hard to beat, especially if you need quiet and low bias current at null.  I'm partial to the '155's for being a bit more quiet in general, but some of the refurb 845's in the cal room are nothing to sneeze at.

[Echo88]

I also own a K155 and would like to ask which amount of noise is visible on yours MisterDiodes? On mine i see about 100nVpp in the 1µV-range after being on for about 2h. Ive never really cared about it, since im not really into bridge measurement techniques (yet). But since you guys mention visible 10nV resolution ive got the impression that my K155 is maybe faulty regarding the noise?

[MisterDiodes]

My unit B will usually stay right around +-50nV of zero, as a slow drift back and forth - generally very symmetrical around zero.  Leave it on for a while before zeroing out - at least a few hours to stabilize, like with any other piece of precision equipment.  Unit A will drift slow maybe around 60~75nV overnight but again no real overall drift.  You learn what the needle looks like as the terminals reach thermal equilibrium.

Applies to most Null Meters, not just '155:

Sounds like yours is somewhat in the ballpark...Make sure your batteries are good and you might go thru the cal procedure in the manual.  If your batteries have leaked at some point you want to get any gunk off the PCB.  Also replace the electrolytic caps, although these don't see much ripple current they will tend to get tired anyway.  Keep the PCB very clean at all times, and remember that some of these are light sensitive - so keep the lid on in between cal adjustments.

Keep your lab temp constant, and keep all air drafts away from the front terminals. Use the guard circuit.  Learn how to use that mirror scale, it is there for a reason!  Always make sure you have your eye positioned so the needle is lined up exactly over the mirror (the needle and its mirror image are as one), or if you have a slight offset the true needle position is halfway between the apparent needle position and the -image- of the needle in the mirror.  That mirror gets rid of the parallax error of needle position.  After a while you'll do that automatically whenever you see a mirrored analog meter.

You can also use your DMM to read the output of the null meter, if you like going digital - but usually we're after the really quiet aspect of these meters - one of the reasons we're going with a battery powered null meter is to avoid the noise a DMM introduces to the test.

Bear in mind on the 1uV scale just bringing your hand near will shift it.  If you leave these null meters "open circuit" on the 1 or 3 uV scale realize you may get an oscillation  and I don't like banging that needle at the hard stops - leave the power switch in Zero Check until your circuit is hooked up, then go to ON mode only at a higher range, and work your way down to smaller ranges while you maintain balance on your setup and get everything dialed in.  Don't just jump to 3uV or 1uV scale until you know you're within 3uV or 1uV of null balance.

Sometimes your circuit is too noisy to get to 3uV or lower, so you just stay on the scale that suits your measurement the best.  You won't be on the 3uV or 1uV scale unless you're looking at a very quiet circuit, so just make small adjustments and pull your hand back and be patient.  After a while you can detect where the center of the needle is, even if its slowly shifting slightly up and down - you have a good idea where the center of motion is.  That's your measure, and ideally that is right around zero or pretty close to it.

Null meters are best at reading a "null" or zero volts across a bridge with very good accuracy and precision.  If you're trying to read a non-zero voltage then your DMM is better suited.

Once you think you're at Null you swap the leads around on your test setup and check again.  If your making adjustments on a KVD + 732 to match an unknown voltage, you make a change on the KVD dials that is one-half the total difference you read on the null meter when you swapped leads around, and that should get you nailed right on the dot after about the second try.   

It takes a little practice but after a while it's easy and quick. 

[Echo88]

Interesting, thanks! Yes, those instruments really teaches you a thing or two about patience and nanovolt-sensitivity. There are other nV/Nullmeters out there, like the Keithley 147/148 or the Tegam AVM-2000, but i guess the Fluke 845 or K155 is the best compromise (availability, low input bias)? Maybe one of you guys own a AVM-2000 and can give further info/teardown (Tegam-Website isnt available at the moment)? 

[MisterDiodes]

Another thought:  When you're calibrating your 752 Hamon divider, if possible you want a real null meter (your '845 or '155 or similar) sitting on those "Null Detector" terminals, -not- a DMM.  As shown in the 752 manual.  Some pA or 10's of pA bias current @ zero volts differential you get with any DMM's input amp can sneak in and shift that apparent cal pot adjust point very slightly.  That may or may not make a difference to you. 

You can use a DMM (depending on how much bias current you've got you may have to find a slightly better -average- cal position when you change the + and - switch), but it's much easier with a proper null meter with much lower bias current (at null) if you're after top performance.

The same goes for resistance bridges, as Mr. Pettis has discussed above - in that case there is really no substitute for a real null meter for accurate resistance bridge measures.






 

[lukier]

Another thought:  When you're calibrating your 752 Hamon divider, if possible you want a real null meter (your '845 or '155 or similar) sitting on those "Null Detector" terminals, -not- a DMM.  As shown in the 752 manual.  Some pA or 10's of pA bias current @ zero volts differential you get with any DMM's input amp can sneak in and shift that apparent cal pot adjust point very slightly.  That may or may not make a difference to you. 

What about making a buffer from the null terminals to the DMM out of ADA4530-1 opamp? It has only 20fA of input bias and integrated guard.

[Edwin G. Pettis]

While the 20fA input current (check the specs to make sure that current is valid under a null bridge condition) solves the input bias current question (adds a bit more noise too), it still does nothing for the incremental LSD limitation of a DVM and its noise, you still can't 'see' absolute zero on a DVM any closer than ±1 digit or a half digit if it displays all zeros it isn't necessarily really zero.  A null meter (analog) is continuous and when adjusted correctly displays a true zero or how close you actually are to zero, so no, a DVM is no substitute when a null meter is called for no matter what you do on the front end.

[MisterDiodes]

Another thought:  When you're calibrating your 752 Hamon divider, if possible you want a real null meter (your '845 or '155 or similar) sitting on those "Null Detector" terminals, -not- a DMM.  As shown in the 752 manual.  Some pA or 10's of pA bias current @ zero volts differential you get with any DMM's input amp can sneak in and shift that apparent cal pot adjust point very slightly.  That may or may not make a difference to you. 

What about making a buffer from the null terminals to the DMM out of ADA4530-1 opamp? It has only 20fA of input bias and integrated guard.

As Edwin noted - you've got 40uV / rt Hz on that part, so it would be a challenge to get a quiet reading.  You want something in the nV/ rt Hz or much lower range for a useful sensitive null meter.  There is always a trade-off on op-amp features.

Again:  There is no substitute for a good working null meter when you need to detect the calm center of the "Eye of the Storm" at zero volts differential. 

This is an excellent example of where sometimes discrete transistors (or matched dual packages) are a good way to go for lowest noise and low input bias - Art of Electronics 3rd edition has a good read on designing discrete transistor amps for very low noise - this is an application where an IC amp might not be the best way when larger area and quieter, less stress discrete BJT's and Fets can do a very good job.  Analog methods are nice when you really want to look for the slightest twitch of the needle, and know you aren't blasting the test setup with noise...if you're working at low ppm DC volts or ohms measure.  As Edwin said - if you're looking at the very end of that digital display and hoping that right-most digit is going to help you find true zero - that might not work out very well.

That isn't to say people have been building various op-amp null detectors for decades with IC op-amps (Conrad here has a nice DIY article that was a great read!) - but the purpose made tools are the workhorses in the lab, and in the end they save time and money money if you don't have to re-invent a very good wheel.

[Kleinstein]

Those electrometer OPs are very good with bias, poor with low frequency noise.

Separate transistors / FETs have the advantage of better isolation between them. However the 845 circuit looks rather old in using a BJT for the demodulation and also for the AC amplifiers input. Both jobs are better done with FETs.
 
However, judging from the circuit, the Fluke 845 is also not really low noise. The resistors at the input alone give a noise level of about 70 nV/Sqrt(Hz) and the BJT based amplifier will likely about add as much noise. So it is not exceptionally low noise - the input stage on good DMMs can be lower in noise.

The high input impedance on the 845 is also limited to low input voltages. Once the voltage exceeds something like 0.5 V the protection diodes kick in and input impedance is down to 150 KOhms. The filter caps at the input can also lower the input impedance, once the signal is significant out of range - so when connecting the meter, there can be a considerable input current to charge the input caps. So the DC pA level in only reached after some time - it can start in the upper nA range.

A conventional digital display might not be that good for judging a zero. However, suitable scaled and with suitable filtering (e.g. running average) and additional statistical info it can work well, even better than the old style analog. This is especially true with a instrument that is slow anyway, like the 845. The slow analog filtering used here can be rather misleading. Averaging by eye only works well for a limited time.

[Edwin G. Pettis]

You apparently have no (or very little) hands on experience with null meters, the 845 is indeed quiet, noise is under 200nV on the 1uV scale, 250nV 1mV to 300mV and 300nV above that, granted that is with a shorted input but that is a standard method of noise measurement.  The Keithley 155 has very similar specs to the Fluke.  While it is possible that FETs may produce lower noise, that is not always the case, bipolars can produce very low noise levels as well.  What digital instrument that could possibly have a 1uV full scale range compete with a null meter for noise or accuracy with <200nV noise? 

Best practice is to use the lowest possible noise levels in instrumentation to begin with, it is poor practice to use higher noise levels and then attempt to mask them by filtering and math which has their own problems, you should know this hard and fast rule and if you don't, shame on you, if you do and don't practice it, shame on you, digital is not the end all fix or solution, never has been and never will be.  I have used null meters for decades, I'm quite familiar with their ins and outs, until you have some actual experience, don't knock them, they do their jobs exceedingly well.

The input diodes are for clamping overloads, they are inside the feedback network so no the input impedance does not fall to 150K, the lowest input impedance under 'normal' input conditions is either 10 Meg (older units) or 1 Meg (newer units), similar conditions exist in the 155 units.  Who cares, at null (after all this is primarily a null meter) there no current (or very little ) flowing in the input so your argument is again incorrect.  So an input overload causes a few seconds overload recovery, what happens in a ADC when its input is overloaded and needs to change scales?  That doesn't happen instantaneously now does it?  Correct use of a null meter is to start out on a higher voltage scale (when the input voltage isn't known closely) and adjust the scale downward so that a null can be achieved.  If you are silly enough to put the meter on a very low scale with significantly higher input voltage, guess what, the limiting diodes cuts in and shunts the current away from the input circuits, just what it is supposed to do.  Virtually all of your arguments start out with the misuse of the meter, what do you think happens in any other instrument where its input circuits are overloaded like you are describing?  The same recovery scenario takes place.

With all respect due, as to your "suitable scaled and suitable filtering and statistical info", baloney, you're already adding more noise, error and uncertainty to the mix, again you start out with the lowest possible noise, no amount of filtering and statistical dithering is going to improve on that, you claim a null meter is slow, I don't think all of your add ons are going to be all that fast nor as accurate, every addition to the signal chain introduces more noise, error and uncertainty, that is a fact of electronics .  Your assumptions are obviously not from actual use or practice, your arguments are mostly specious, and averaging by eye, I do it when needed, works just fine and is quite accurate, your digital is still incremental and has inherent limits, it is not best practice, just ask the labs who still use both old and new technology, there are some huge companies out there that rely on both technologies and know when to use the correct instruments.  I know primary cal labs that still use null meters, are they all possibly wrong in their conclusions?

Yes I have used DVMs as null detectors sometimes, their limits are quite obvious compared to an null meter in actual practice, within certain limits, a DVM can be used without problems but when you're in the hunt for PPMs or lower, a good null meter is exceedingly hard to beat.  Just check out the manuals for instruments that measure PPM or lower, their calibration setups require null meters, not DVMs.

Get a good null meter and learn how to use it properly then come back and talk about them, experience and use count for a lot when talking about an instrument, just looking at schematic and making observations is not best practice.  Try talking to others who have used null meters for years and get their opinions, it isn't just myself or MisterDiodes who share these opinions.

[MisterDiodes]

Just another thought on '845 / '155 / 419 etc. and why these are sometimes the very best tool for the job. 

For example you will see the Tegam avm-2000 advertised as a replacement for the 845 / '155 / 419, if you've got an extra $6k to spend:

https://www.tegam.com/shop/voltage-resistance/avm-2000/avm-2000-null-detectornanovoltmeter/

These can be OK if you don't have access to a real analog null meter - and this one does have a feature to trim down the bias current -  but some of the problems we found during eval a few years ago (might be upgraded in the last few years) made it not seem to be worth the price tag:

1.  A minor issue but still disappointing is that for $6k you get somewhat hobbyist grade Pomona binding posts.  They are "low thermal" but a step down in quality from the older beefier styles used on older meters.  The binding posts on old meters tend to have a much thicker gold flash and larger support insulator and seem to stand up to abuse better.  The newer style does have a slightly better thermal time lag, but we've seen the bases crack if they get bumped.

2.  The meter movement on these units isn't as well made as the older meters - the meter on the new unit has a plastic face that is a prone to storing static charge - see their app note on special procedures to overcome that problem.  If you actually test the taut-band meter movement itself on an older 1-3 scale Fluke / Keithley (usually made by API / LFE) or HP meter, you'll see that 2% mechanical accuracy spec is very conservative...we normally see more like 0.5% or 0.2% FS error on the movement itself (if that), very good linearity (<.05% error mid scale typical) and virtually no backlash or hysteresis. The newer meter movement doesn't live up to that manufacturing standard, at least not the one we looked at.

3.  The main problem is that on one test we've got a spec on the test jig setup that equipment in the local vicinity have virtually no detectable EMI over 10kHz  (That's why we also switch to battery powered halogen or LED lights during a finicky test).  And that's where this Tegam meter has its main problem:  They put in a noisy CPU where the quality PWW pot should go - and that splatters noise around the local area which might affect your DUT.

This is exactly the same type of issue that you run into when trying to run a DMM or digital anything as a null meter, regardless of relatively high input bias current on a DMM.  And the same reason why you have to keep an eye on any  AZ amp (like a '2057) very well shielded and bypassed around sensitive circuits.

This type of Tegam nullmeter itself (or DMM) could work out fine for you if your not worried about possible EMI in close proximity to your DUT.  A DMM can certainly work if you're not worried about pA or some 10's pA current flow.

So when you look at meter specs or design you're own, a pure analog (or very close to analog only even if it has a fairly slow chopper) design is really very useful.  You have to look at not only the noise of the meter's acquired signal, but you need to look closely at the meter's own clock EMI noise injected into the test system itself, and watch out for that.  Better to not have a clock at all beyond whatever the chopper is doing.

And that's where the 845 / 155 / 419a type meters really shine for the tasks we need to do. Yes they have choppers but the freq is low enough to not have a huge impact on what we're measuring.

ANOTHER TEST:  I did have a tech at the cal room quick-check some other 845's and '155s just for fun for bias current, like SZA263 suggested - and I can confirm '845 numbers around 150~250fA (slightly less with the FET switches but the Neon-driven LDR switches work well) or less bias current at null - one pimped out '155 with newer Fets in the chopper and first stage amp was running around ~75fA or less bias current at null.  So these meters are obviously -not quite- perfect, but are certainly -not- pumping out a large bias current at null if they are working and in proper adjustment.  At least an order of magnitude (or more) less bias current than a 3458a, and these older meters have none of the relatively high clock noise you get with CPU / FPGA emitters.

Could you build an even better 845 / 155 / 419 style meter with upgraded matched Fets arrays and BJT differential amps - and maybe an IC amp follower?  Probably. 

That's why in our cal rooms we have the older equipment working right along side the newer designs.  Use what is best for your application and test requirement.

[martinr33]

On the noise topic...

It would be straightforward to build a precision 86Hz oscillator using an 8-pin low-power microcontroller. SMT would keep the current loops smaller, and it would have good ground planes. An external crystal would be best, but the device can hit 2% with its internal oscillator. The trick with the 845 is that this modulator goes in the outguard section, driving the isolation transformer. The clocking can be recovered from the inguard isolation supply as it is today.

Any thoughts on whether the extra stability would affect the emitted noise enough to be a problem? The internal oscillator has a much wider bandwidth than a crystal, so the peak noise is much lower.

[MisterDiodes]

I wouldn't use an MCU if possible for our fussy application - even one single chip that is has many sync'd transistors switching fast on a clock is enough to mess up the test unless it is very well shielded - but you are on the right track about keeping the bandwidth spread out a bit.  A simpler BJT or FET oscillator works and keep currents very low.  The final freq around 85Hz (for '845) or 220Hz (for '155) is not super critical - those freqs are chosen so the 50/60 Mains power interference doesn't beat in the chopper section and make you new mystery noise.

For the '155 chopper fets and on newer / modified '845's: The chopper signal in this case doesn't want or need super-fast edges.   Slightly softer corners and not over-driving those chopper fets helps keep the system quiet. If you keep the switch gate drive rise / fall times well controlled on those switches then you're injecting less gate-charge noise onto the signal. 

There are filter caps on the inputs to deal with the leftovers but we try to keep all switching signals to an absolute minimum and as quiet as possible.  The '155 doesn't need the isolation transformer since it always used low voltage Fet choppers, and that's why we kind of lean towards that unit.  The '845 even with neons is pretty good though for most needs and the version with Fets is even quieter.  They will all get the job done with fairly low emitted HF switching noise compared to a DMM, and at a much lower bias current at null.

[zhtoor]

I wouldn't use an MCU if possible for our fussy application - even one single chip that is has many sync'd transistors switching fast on a clock is enough to mess up the test unless it is very well shielded - but you are on the right track about keeping the bandwidth spread out a bit.  A simpler BJT or FET oscillator works and keep currents very low.  The final freq around 85Hz (for '845) or 220Hz (for '155) is not super critical - those freqs are chosen so the 50/60 Mains power interference doesn't beat in the chopper section and make you new mystery noise.

hello,

what would be the effect of "dithering" the 85Hz chopper frequency from say 80Hz to 90Hz using some kind of a physical random noise source on
the performance of the null detector?

regards.

-zia

[SZA263]

hello,

what would be the effect of "dithering" the 85Hz chopper frequency from say 80Hz to 90Hz using some kind of a physical random noise source on
the performance of the null detector?

regards.

-zia

A complete disaster 

[martinr33]

That's an interesting question, because the whole design is a bit tricky.

Although the time window changes, the rise and fall times on the opto choppers do not. Therefore, as the frequency changes, the output waveform - and hence the final meter reading - will change out of proportion to the input voltage. So I would say less variation eliminates another source of uncertainty.

The whole design is a well-crafted balancing act, probably done by someone who previously had to use vacuum tubes. The photoelectric chopper was probably a huge deal.

[MisterDiodes]

"A Huge Disaster"... Yes maybe.

The problem with dithering the frequency is then becomes "how fast is the dither" and will the chopper amp front modulator and back end  demod get out of sync.  I would just enjoy the meter as it is, and don't change that unless you need to.

We have had on occasion on some test where say the '155 was showing a slight sine on the output (you learn the difference between sine error vs random noise), and so you can either tweak that multivibe freq a little, or find out what was left turned on that's beating in with the chopper, or switch meters to an '845 or vice versa.  Or find out who left their wrist watch /  cell phone / tablet /  PC / DMM / Switching power supply / Noisy LED light / USB port or whatever else left on in the lab, etc.  We've seen all of those things happen.

That's why it's handy to have a few different nullmeters and extra u-Metal foil, etc. in the lab for this situation - you can't have too many tools at your disposal.  Some tests require both a nullmeter and some DMM / PC / wafer prober in the same area, so you just have to work it out for the best setup. 

Really, these meters are nicely and thoughtfully designed - and for 1965-68 it is amazing tech that really stands up to this day.  If you take the time to really understand the schematics you'll learn a lot. Especially the clever technique of wiring the resistors around the range switch decks and how that works with the 1 - 3 scale meter to adjust input attenuators and amp gains. Neat stuff.  If you study that you can learn some tricks to make your modern circuit design better also.

By the way: Those nice PWW adjust pots and range switches ==> The original "Non Volatile" calibration memory, no noisy CPU required.  With a little care these meters are some of your best friends if you're working with high accuracy measures with 752s / 732's / KVD's / Resistance bridges, etc.

[Vtile]

uA741 did pop out 1968 about the same year as these "magical light isolated" chopping nullmeters (845). How leaky and noisy the uA741 were? 

[Kleinstein]

The circuit is really old style. The AC amp part is interesting to learn from. If available they could have used the µA741 for the DC amplification part (Q108-Q112).

The µA741 was not that good when it comes to noise leakage and similar. It was great because it was easy to use with internal compensation.

Noise wise the 741 is not that bad - 23 nV/Sqrt(Hz). The Fluke 845 should be somewhere at 100 nV/sqrt(Hz) - of this 70 nV/sqrt(Hz) are due to the resistors at the input alone.

[MisterDiodes]

The circuit is really old style. The AC amp part is interesting to learn from. If available they could have used the µA741 for the DC amplification part (Q108-Q112).

The µA741 was not that good when it comes to noise leakage and similar. It was great because it was easy to use with internal compensation.

Noise wise the 741 is not that bad - 23 nV/Sqrt(Hz). The Fluke 845 should be somewhere at 100 nV/sqrt(Hz) - of this 70 nV/sqrt(Hz) are due to the resistors at the input alone.

This is getting way off-topic - but a 741 is not considered low noise at low freq. (< 10Hz) and yes it still is/was not that great compared to lots of techniques available then and now.  Especially when they first became more widely available in the early 70's.  Even today you can build a discrete amp or hybrid that outperforms a lot of single IC amps especially for low noise at low freq.  The recurring problem you run into with single die op-amp packages is it's always a tradeoff between input voltage noise or low input current noise - and a null meter sort of wants both, even at the DC gain stage.

Real world measurement - and you'll see that even a bit more modern single 741 would generate a fair amount of noise compared to what you can do (and what is done in these meters) with discretes (This lab report done in 2005 and could be done with better amps today):

http://www.cs.ru.ac.za/research/g02c0108/files/noisehandin.pdf

The '845 / '155 / 419a will generally have less noise than the more simple and "trivial" op-amp attempts, but a modern high performance hybrid (discrete + IC) solution can probably work very well.  Part of the real magic occurs in the stable voltage zero and current offset systems built into these meters.

All of this is hard to do at 3uV ~ 1uV and lower ranges.  That's noisy territory.

It should also be noted that a lot of times with voltage bridge  measures (KVD + 732 for instance) you're maybe not working a lot at those very lower ranges - probably 10uV or 3Uv is a good range to be on if you're playing with LTZ's (10uV is 1ppm in the 10V world).  Resistance bridge null meter settings will depend on what you're after, and those lower ranges can help.

[Vtile]

I need to clarify my uA741 note. It were mostly meant to be kind of a historic sidenote, of the IC technology state back in the era when 845 did come out on the market. So please lets not took it too far away offtopic. This nullmeter discussion is most interesting as most of the discussions in here at metrology section.

As a technical side note (OT again since I do not have "big boys" instruments) I tested the bias current of my Meratronic/Marconi Meratester (Poland, 1970s IIRC) in 1.5mVdc range and did end up result of ~700fA over 10 meg resistor (60uV over 97 MOhms), the device have a null / center zeroed mode for bridge use. It does use HLY7006 hybrid amp. Keithley 197A did show wild readings as high as 1.5mV over 10 meg resistor. 

[classicTEK]

Hello All,

I figured this would be the place to keep the cumulative knowledge base for these tools, so the thread lives and breaths.

Mine is an early 80's vintage AR and will not fall into spec for Noise when ZERO'ing or "SHORTED" for operating stability. It drifts around null about 1uV back and forth and occasionally jumps more in the 1uV setting.  Setting ZERO is like a cat chasing its tail.

As I am reading the manual it EXPLICITLY says "Do NOT attempt to clean switches."

This is all well and good for all of the obvious reasons; however with an instrument as old as the 845 and an unknown history it worries me a bit. 

How have others made certain it is not dirty switches causing any unstable readings when ZERO'ing and operating the unit?  Did you throw caution to the wind and use any certain method to clean the switches with success or did you simply follow the manual to a tee and search elsewhere for the problem as is stipulated in the TS procedures?

I just don't want to jump to component level TS before doing the most common of problem solvers with old gear.........cleaning DIRTY SWITCHES.

Thoughts?

Cheers,

David

[ManateeMafia]

The safe bet would be they are referring to chemicals that can destroy the materials in the switch and/or reduce the isolation from the front panel to the guarded section. Contact cleaner would definitely not be recommended but IPA may also be bad. You can examine the switches with a magnifying glass on the lowest ranges. Caig DeoxIT would be my first choice in such a situation where alcohol might cause the plastic parts to break. Assume they are already brittle with age.

I would guess any powered device would need the power supply checked first. If the 845AR is using the optical chopper amp, then the neon bulbs would also be suspect.

Clean your binding posts. I find a lot of the older gear has hidden corrosion inside and could use a good cleaning.

Good luck.

[classicTEK]

GREAT!!!!!  That was my initial plan:

Clean all switches, binding posts, and grounding straps with a very deft hand, q-tips, moistened w/de-oxit.

Excersice the legs of all socketed transistors in place.

I will try this first and see if it helps.  If not, it is off to the races following the TS procedures in the manual.

Mine does have the chopper circuit, but my neons seem bright and strong.  HOPEFULLY!!!

Thank you and I will report back with my results to hopefully help others.

Cheers,

David

[wn1fju]

For what it's worth, the drift on my 845AR improved substantially after I scrubbed the range resistors with alcohol (and let them dry
thoroughly).

[classicTEK]

I have searched all over for a video of a fully functional and calibrated 845xx nulling on short or zeroing.

Could someone be so kind as to make a quick video and post it so those of us whom have never owned or used one can see the proper operation? Or more importantly the behavior of one that is functioning correctly.

I for one would truly appreciate that.  Especially for inspiration!!!

There zero videos of the 845 in action.

Cheers,

David

[Conrad Hoffman]

Sounds like watching grass grow. The needle just slowly drifts to where it's going, maybe wanders a bit and stops. At the most sensitive settings it will wander continuously due to noise. Alas, no way to make a video here- not a cell user!

[classicTEK]

Conrad,

Thank you for that !!!!

I must admit publicly that I should not have been working and posting last night.  I have found myself in the awkward position of discovering my tired mind an eyes were simply not connecting in the wee hours of the night.

I was actually on the 1uV scale and was reading it as if it were on the 10uV scale!!!   

Now that I am well rested and have LOOKED at my scales  and meter......I am experiencing about .1uV of noise when shorted on the 1uV scale.

I still think I am going to change out CR207 to see if that helps with what I "think" is a bit of a stability issue to get this dialed in if possible.

[FTM777]

Builder
Would love some more information on your conversion please.
Nick

[SeanB]

Just remember there are different grades of 741 ( or at least there used to be), and you could get somewhat more expensive ones that were selected for lower noise, and very low 1/f noise, though those typically were part of the military spec versions. You could get them as practically noise free for popcorn noise over the full military range, and they were at least good enough for 14 bit converters. Generally they were the TO100 cased versions, as those have a lot lower stress on the die, and the performance was quite good.

[velik_kazakov]

Hi,
last night I start to restore the second one Fluke 845AR. I score it from ebay and receive it in last week.
I find some bad caps and desoldered resistor on it. Someone is repaired the null detector in the past.
It start to work now, but still have some probelms. I find that the neon bulbs not fire every periond. Something else, when I turn off the light over the instrument, the neons stop to fire. When I turn on the light . they start to fire. I check all around it and all is OK ( I think). The miltivibrator wirk fine and stable. Have a HV on the pin 9 of the traf that drive neons. Other voltages (+15 and -15) from the same traf is OK. Maybe the neons need to be exchanged. Want to ask, am I right or need to check to find something else. Do you have the same problem with neons.
Here I post a video with the nepns. https://youtu.be/FcnWiiL_dJA

[doktor pyta]

Yes, neons need to be exchanged.

[velik_kazakov]

Thanks! Will look to order some neons or will do a leds mod to this one Null Detector. My other one 845 work well for now with original neons and I maybe will try led mod.

Edit:
Find some time now to do all. Make the mod with the leds from Epson inkjet priner. It is the same in R285, P50, L800, 1500,,, all use the same leds and I have lots of it. I look that the emitted light is very simmilar to the neon light. Do some mods and instal 2 diodes with 3,2 kohm resitor in series. Exchange the places of both to keep the oem diodes as before. Finaly all work well for me. I record the internal noice for some time and it looks good. Thank you again!

[branadic]

I bought a Fluke 845AR a few days ago, but I'm not very happy with it's performance. The needle is anything but stable and noise is in the order of 500nVpp in 3µV range.

First I adjusted full-scall deflection in the 10V range using one of my LTZ references. Next I adjusted the pot for the 84Hz (12ms) chopper frequency. I then used my reference to adjust for 1V output voltage and recorded the output voltage overnight using my R6581 (1V range and 10NPLC) in the 3µ range and with the input shorted. Within the first about two hours I readjusted the zero pot. As you can see the noise reading is rather high. I measured a few test points given in the manual and so far everything looks quite okay.
Will measure the capacitors next week, once I have the ESR meter at hand to see in which condition the electrolytic caps are. Any additional hint what to look for?

-branadic-

[Kleinstein]

The 845 is not a very low noise meter. It is more like very low bias. At the input there are some 300 K of series resistance and this alone gives quite some noise (some 70 nV/sqrt(hz)). The amplifier itself may also add noise. This could well possibly in the same range - otherwise they may have chosen smaller resistors.

For very low voltages it may need some additional averaging, e.g. in the digital domain. Digital filtering can by FIR type and thus need less settling time.

[branadic]

Can't agree on that, the manual says it should be 250nVpp in the 3µV range, 200nVpp in the 1µV range as maximum noise with input shorted.

-branadic-

[Dr. Frank]

Have a look at the neon drive waveforms.
Noise that high might indicate bad neons.
Noise is too high..
Frank

[branadic]

Thanks Frank, will look at that. Had already ordered a pack of 10 new neons, that will arrive soon. I know that there is this LED mod, but I want to keep the unit as original as possible.

-branadic-

[branadic]

Frank, seems you are absolutely right on the neons. While I can measure about 105V across one of them the other one is only half in amplitude. So need to wait for the neons to arrive and replace both of them.

-branadic-

[Conrad Hoffman]

Note that the trigger voltage of neons is light sensitive. That's normal. Long ago I worked on Ilex electronic shutters. We had a neon in the box that stayed on full time to illuminate the neon used for timing, to stabilize it.

[branadic]

I went through the dumpster of my colleague today and found a few neon bulbs. Unfortunately they turned out to be to long, though working. By accident I searched on the web if Conrad Electronics, we have a store here in town, has some neon bulbs and it turned out they have and they have them in stock in their store.

By the way: most of the time the stuff you need is available online only and can be ordered, which takes a few days.

So I went to the store and bought two packages, each containing 5 bulbs, 1.99€ each package. The neons are now installed and everything is adjusted. Indeed one of the original neons were already black on one side. Maybe I could have simply turned it around and use the clean side of it, but hey we have new ones
The needle is now stable down to the 1µV range, so we can call that a repair. Excellent
So if you need neon bulbs and need them fast, look for this one: Barthelme 00082305 Glimmlampe 115 V, 230 V 0.25 W T1 3/4 MG Transparent

-branadic-

EDIT: Picture of a measurement in the 3µV range before and after bulb replacement for comparison. Vertical scale is identical.

[guenthert]

  As far as I understand, there are different kind of Neon bulbs:  those meant for indication/illumination, which are still readily and cheaply available (but very limited on/off cycles) and those used for switching purposes in the sixties, which seem to be unobtainium today.  I got a HP419A with failing Neons and would love to be proven wrong.

[branadic]

The only difference I see is the distance and the length of the rods.
Anyway, I have ordered some bulbs that are sold as NEW Box of 10 Neon Bulb NE2U A3C Indicator 105 to 125 Volts AC/DC FREE RESISTORS. I will see how they differ from the original ones I desoldered once they arrive. But I couldn't wait any longer to see, if there is something else broken on my unit. If my current replacements fail early I'll let you know.
And I will change and measure behavior of the ebay items once they arrive, so we can see if there is a noticable difference. But I wouldn't expect any.

-branadic-

Edit: Can't see a real difference in the datasheets except that small difference in current and maximum breakdown voltage, which seems negligible:
1.9mA and 95Vac for AC3/NE2U
1.7mA and <=90Vac for the Barthelme 82305

[Dr. Frank]

If you really have found such a simple solution, I'll call you a genius.

I remember many discussions on volt-nuts / time-nuts, which neon replacement to use, as the ignition voltage and pairing was crucial, I think, something like 90V, or so..
The specialty was, as Conrad indicated, that these Neons probably had some radioactive substances inside, to lower this ignition voltage.
But anyhow, if it works flawlessly, then maybe I will repeat your stability tests and maybe replace my Neons also..

I've got an 845AR from late Singer Pharmacy, for a reasonable price (was the sell-out, maybe), and it's working quite ok for my purpose, i.e. autocal of my Hammon 100:1 / 10:1 divider.
Not the least quiet nullmeter, but with a bit of patience it works very well. Especially the extremely low bias gives good results.

Frank

[branadic]

Frank, it works at the moment. What i did today is I lifted the neons a bit, as they are a bit shorter. I'm currently monitoring if anything has changed in respect of noise. Therefor I had the leads kept long enough before to do that.
However, I call it a fix, but don't call me a genius though. You know, luck is with the stupid ones. Would be great to see how other 845ARs perform (DMM set to 10NPLC, AZERO on), so that I can compare my results.

I was looking for a Keithley 155, but they go for a lot of gold, which I wasn't willing to pay.

-branadic-

EDIT: Looks like lifting the bulb slightly improved noise, since more light can couple into the plastic rods. The small variation is due to open windows here to get temperature somewhat down in the flat.

[dietert1]

Your plot shows  a noise signal of about 200 nV. Could you explain a little what you are measuring? I assume the monitoring output of the 845A is connected to a DMM for logging and the 845A has some gain between input and monitoring output. Is the vertical scale of your plot for the input or the output of the 845A?

Regards, Dieter

[branadic]

It's the noise measured with my R6581D in 1V range, with the input of the F845AR shorted. Therefore the output of F845AR was adjusted to deliver 1V at the output for a fullscale deflection (see also manual).

-branadic-

[dietert1]

If i understand that right, the 845A gain in this measurement is about 300 000 and its input noise was measured to be 200 nV / 300 000 = 0,7 pV. Is that correct?
I guess this is into a bandwidth of 1 Hz or so, limited by sampling rate of DMM.

Regards, Dieter

[branadic]

No dieter1, this calculation were about right if this was the noise on the 1V output voltage, but the 1V is already scaled back to 3µV range.
However, I wonder for what observation period the given noise values for K155 (150nVpp in 1µV range) and F845AR (max. 200nV in 1µV range) are valid. If it is related to the response time only, then my F845 is far below the maximum value and equals a K155.

-branadic-

[dietert1]

Thanks, i was trying to get a comparison with a modern chopper amplifier like ADA4522. Its noise spec is 117 nV typ into 0.1 .. 10 Hz. So i guess into a bandwidth of 1 Hz this would be about 10 or 20 nV - roughly a factor 10 better than the FLuke 845A.

Where the Fluke null meter really shines (different chopper technology!) is the low input bias current. As far as i remember someone measured about 75 fA . Compared to the 50 pA typ spec of the AD4522 this is a factor 700 better. The difference matters whenever source impedance is bigger than about 0,1 uV / 50 pA = 2000 Ohm. In fact, with the ADA4522 a source impedance of 200 Ohm may already cause a DC bias error bigger than noise.

Regards, Dieter

[Kleinstein]

The F845 and K155 have chopper inputs and should thus have little 1/f noise. So the length of the observation period (lower frequency limit) should not be that important, unless extremely long so it includes drift (e.g. thermal effects). The upper frequency limit should have more effect - the logical choice here would be to have the build in filtering function. So the direct comparison should include that number, especially if the filter is adjustable (e.g. extra filter at the output).

The low noise AZ OPs have a much higher bias current - this alone would not be such a big problem, as there could be a compensation, as the current is relatively constant. The other problem is quite some current noise. For the ADA4522 it starts at some 50 K that current noise is more important than voltage noise.

Some of the F845 noise is from the input filter / protection part. Just those resistor of 300 K in total give quite some noise.
The input filter can also be tricky with really high impedance sources - they will slow down the filter and there can be quite some current pulse from a charged filter cap.

[Dr. Frank]

I also measured the 1V monitor output of my 845AR, for 1µV and 3µV range.
At first I found out, that this output was extremely slowly responding, not following the needle deflection instantly, at all.
Reason for that, in my instrument, FLUKE had implemented a strong filter capacitor of 400µF over 4k7, instead of these 10µF, which you'll find in the schematics.

So I removed this capacitor completely, filtering time constant being determined by the DMM, a 34465A, at NPLC 10, i.e. tau = 200ms.

This gave a StD =  128nV_rms or 360nV_pp for the 1µV range, and 234nV_rms or 663nV_pp for the 3µV range.
A simulated NPLC 100 filter gave an additional damping factor of about 1.5 for all of these values.

As expected, that is about 2 times noisier than specified, due to the age of the neons, which effect branadic demonstrated faithfully.
The 1µV diagram was corrected for thermal drift, whereas the 3µV diagram shows the uncorrected -160nV/hr. drift.

Frank

PS: It's necessary to check all electrolytics, as these may have high ESR, or even high leakage.. I've already had replaced many of these before performing this test.

[Kleinstein]

Even if using a DMM at the output, it could help to have some analog filtering. AFAIK most of the modern DMMs have no analog filtering before the ADC and are thus sensitive to some aliasing noise (e.g. 25 Hz range for 1 PLC AZ mode at the meter).
So it would make sense to have something like the 5 K *10 µF filter from the diagram. The minimum would be to dampen out residual signal at the chopper frequency. Some 50 ms time constant would not really hurt even if one want's to wait some 5-8 time constant. Another option could be to operate the DMM in non AZ mode.

[Dr. Frank]

Even if using a DMM at the output, it could help to have some analog filtering. AFAIK most of the modern DMMs have no analog filtering before the ADC and are thus sensitive to some aliasing noise (e.g. 25 Hz range for 1 PLC AZ mode at the meter).
So it would make sense to have something like the 5 K *10 µF filter from the diagram. The minimum would be to dampen out residual signal at the chopper frequency. Some 50 ms time constant would not really hurt even if one want's to wait some 5-8 time constant. Another option could be to operate the DMM in non AZ mode.

 My argumentation is different.
The 845AR noise is specified for a response time of 5 * tau = 3 sec for the 3µV range (250nV pp noise) and 5 sec for the 1µV range (200nV pp noise).
Obviously this is related to the needle response.

As this 400µF capacitor delivers a totally different response time for the monitor output, I tried to approach the specification by an appropriate integration time.
NPLC 10 should give about 5 tau = 1sec, better comparable to the spec.

But I don't know, how this compares to branadics measurements.

This 400µF, C3 sits directly on the monitor output jacks, so it's easy to identfy it, by opening the outer shield.

Frank

PS: technology of 1968..fascinating

[Kleinstein]

With the different time constants (.6 and 1 s) the different noise specs for the 3 µV and 1 µV make sense. The square root of .6 is close to 200/250.
For a first order filter the noise BW is  at 1 / (4 tau), while the integration over a time window gives a noise BW of 1/T (in non AZ mode). So 10 PLC integration (200 ms) without AZ would correspond to an analog 1st order filter with .8 seconds - just between the two ranges.
So the 10 PLC are indeed a good choice in non AZ mode. Brandic used this too.

The AZ mode would add some extra noise BW from aliasing. Thus my idea to keep a little analog filtering. Anyway just using non AZ at the DMM is the easier way, though it does not help that much with chopper residuals. There could be additional filters for this anyway.
The 845 provides so much gain, that extra noise / drift of the DMM in non AZ mode would not matter. So there is no need to use AZ mode at the DMM.

[branadic]

Thank you Frank for your results.

-branadic-

[lukier]

I did a similar test, but short on the input terminals (not zero function), covered with some foam, 1uV range. Meter is 3458A, 1V range, NPLC 10, AZ ON.

The plot is raw data, so 1V = 1uV. Pk-Pk noise seems to be ~440nV, but there was some upward drift (temperature?)

[Dr. Frank]

Lukier, Does your instrument have 10 or 400uF on the monitor out?

[lukier]

Don't remember off the top of my head. Next time I'll be opening it up I'll take a look.

[branadic]

It's a 10µF/25V capacitor in my unit.

BTW: There are some indications that my unit is from 1974.

-branadic-

[branadic]

So the neon bulbs arrived today. Here is a direct comparison of the new AC3/NE2U (left), the originals bulbs that were installed (middle) and the Barthelme 82305 (right). However, the Barthelme 82305 do work for me. Enough spare parts for the future though

-branadic-

[Dr. Frank]

I followed branadics hints, , and also replaced the neons inside my 845AR, with the Barthelme type.
The first picture shows the old neons, the one in front is only halfways lit, the rear one by 3/4 only.
Both evidently show random flickering, and the glass tubes are black inside.

You also see nicely, that due to the rectification of the 200V square wave, only the lower rods, adjacent to the plastic light guide, will glow.

The 2nd picture shows the new neons, the rods evenly lit, no flickering, and much brighter.
The apparent difference in brightness is caused by the cameras shutter speed.

You can also see two black markings on the wires of the front neon.
I first held the neons outside, in front of the hole of the metal blocks, to measure and mark, how high the glass bulbs had to be soldered, so that the light emission would be optimal.

The noise went down dramatically, back to specification, what can be observed in the comparison noise chart, new neons on the right.
The 845AR is quite good regarding noise performance, at NPLC10, it's 94nV_rms for 3µV range, and 48nV_rms for 1µV.

For NPLC100 @ 1µV, which reflects better the needle response time of 5sec, the pp noise is really < 200nV.
The movement is now rock stable, as good as new.

Frank

[Kleinstein]

From the pictures is looks that just turning the neons by 180 degrees and thus swapping the pins could give them some new life, using the other electrode and other half of the glass.

[Dr. Frank]

From the pictures is looks that just turning the neons by 180 degrees and thus swapping the pins could give them some new life, using the other electrode and other half of the glass.

I don't think, that just 'from looks ' the neons could get them new life.

You better may test on your own 845, if the gas inside, or only the rod is exhausted.

The blackening was inside the whole bulb, most prominent on the illuminated side, though.

[Martin.M]

you can be happy not to have the NE2 with front lense, (by example: used in the Philips PM2421),
they are really obsolete now.

I don`t like the repair of optical choppers   

Martin

(my nullmeter is the 883A)

[Dr. Frank]

Well, I'm sick and tired of assumed 'self-evident' conclusions, which on closer investigation turn out to be lukewarm only.
Such assumptions are currently quite popular here in Germany, e.g. concerning climate protests and effectiveness of CO₂ reduction measures.   

So instead of such speculations, I made some real physics experiments on the worn out and new neons.
Setup is simple, a variable 0.. +/-200V source, 100k current limiting/measuring resistor, across that the Dave-O-Meter, in series with the neon bulb, either exposed to the illuminated room, or inside a light-tight box.
I cranked up the voltage, until either the neon turned on, or the DMM showed a voltage drop.

The old and new neons were both specified to have an ignition voltage of < 135V_dc. In addition, the old NE2U contained some radio active substances to mitigate the dark effect, that is to let the neon ignite even if these are used in total darkness, like in this chopper application.

The old neon at its worn out bar ignited at about 120V, likewise inside the box, but the emission was really bad and irregular.

Using the pristine bar, by reversing the voltage, it also ignited at about 117V with light shining on it, and it really showed a very regular glow around the bar, but at > 150V / 1mA only.

Inside the box, the neon did not ignite before 170V, so very probably not inside the 845AR any more.
Therefore, these bulbs were really worn out, probably the neon gas, and/or the radioactive substance.


Now I tried that same experiment on the Barthelme neons.

With lights on, they ignite already at ~ 65V, independent from polarity, and sample. Only 500µA @ 110V were needed to get a nice and full glow.

To my big surprise, they even ignite inside the box at exactly the same voltage as outside, so they show absolutely no dark effect.
That's maybe due to new gas mixtures, like adding some Krypton gas.

So that's the reason, why this replacement really works, and why I now really love such optical choppers with neons inside.

Frank

[Pipelie]

hello,
I ran some test today.

but instead of short on the input terminals I used the zero function,  1uV range. Meter is 3458A, 1V range, NPLC 10/100, AZ ON.

The plot is raw data, so 1V = 1uV. Pk-Pk noise seems to be ~100nV when using 10 NPLC.

edit:
20190921-3458_k155_ADC#3_100NPLC.csv.zip  is a CSV format file, delete the ".zip" before use.

[Dr. Frank]

Hi Pipelie,
thanks for the measurements!
By specification, the K155 has about half the noise of the 845AR, and that's obviously the ballpark what we both do measure, only 18nV_rms for the K155, and ~41 nV_rms for the 845AR, both at NPLC100.

Would it be possible, that you also determine the bias current of the K155?

It's not specified for neither instruments, but for the 845AR, we measured < 100fA, I think that can be found far upwards in this thread.

Thanks. Frank.

[TiN]

For sake of pointless comparison, adding 20uV range datalog with EM A10 head.
Input is measuring zero, with copper wire extension, soldered by cadmium solder to copper version LEMO input, terminated with 34420-81603 short.
DSV raw data file. 2048 samples from raw data used to plot first attached graph.

And yes, cadmium solder have huge difference on stability.
With normal 60/40 solder I had poor thermometer instead of nanovoltmeter. See second graph. 

I have also ran log on Keithley 182-M, 3mV range with Keithley 1488 short at the input.
All filters OFF.



P.S. /me about to get banned to mentioning RoHS prohibited cadmium solder. Toxic members beware! 

[Kosmic]

And yes, cadmium solder have huge difference on stability.
With normal 60/40 solder I had poor thermometer instead of nanovoltmeter. See second graph. 

So did you used the Sn50Pb33Cd17 solder bar you got recently ? I was wondering if a little bit of cadmium would help.

Nevertheless, If Wikipedia is right, Cd70Sn30 is the real thermal free solder. I don't think it's still made though.

[TheSteve]

TiN any testing done with silver solder?

[antintedo]

TiN any testing done with silver solder?

See this message: https://www.febo.com/pipermail/volt-nuts/2016-October/005023.html. "Silver solder" described by Keithley is Sn96Ag4, they use it in their short for 2182A. According to that post it is not much different from PbSn alloys. I tested samples of numerous alloys at lower dT with unsatisfactory results.

[e61_phil]

Does anyone know if any cadmium solder is used in the Keithly 182?

@TiN: Thank you very much for the measurement!! I was searching exactly that to compare it to my Keithley 182. Unfortunately I don't have a short for my meter.

[notfaded1]

Actually what happens to the neon tubes is called cathode poisoning.  Over time the cathodes sputter material and it collects inside on the glass.  This is VERY common for nixie tubes and other cold cathode tubes.  You're assessment is 100% correct... the tubes were worn out.  Replacing them is easiest and best solution here (they're cheap).  For nixie tubes with cathode poisoning we often crank up the current to exercise the cathodes and burn off built up material sputtered from one cathode to another... in these little NE-2 tubes that won't help much because the glass inside is covered with sputtered material.  Nice to hear the new NE-2 tubes made a significant difference... you're right 1968 technology... pretty neat!  I feel the same way about Nixie tubes, Pixie tubes, Dekatrons and other cold cathode display tubes.  You can totally see from your pictures how the difference is visually obvious.

Bill

[Dave Wise]

READ TOGETHER WITH LATER UPDATES

Here is my take on Fluke's chopper replacement board.

So it doesn't get lost, I'll start with my biggest doubt.
In an original-construction instrument using neons, when T202 terminal 9 goes positive, DS101 lights, turning on V101.  (When it goes negative DS102 lights turning on V102.)
But in an upgraded instrument with Fluke's H11F1 board, when terminal 9 goes positive, I calculate that U5, the H11F1 replacing V101, turns OFF instead of on.
How can this be?  The meter would read downscale for positive input and vice versa.  Did they modify other places?  Why not just make the H11F1 board do the right thing?

It makes my head spin.

Here is my writeup, attached and inline.  It refers to posts by TiN, Jay_Diddy_B, and picburner, especially post #7 and #14, which for me show up on page 1 of this thread.
For easy reference, I also attach pictures and schematics from the above posts.

I'm looking forward to your analysis and comments.
Thanks,
Dave Wise

FLUKE 845A CHOPPER REPLACEMENT BOARD
David Wise
September 2019

The Fluke 845A, 845AB, and 845AR Null Detector was designed with
a photocell chopper.  But see the NOTE in the Change/Errata pages
of 845A_AB_imeng0000.pdf .  (This is pdf page 3.)  Transcribed below.
See https://https://xdevs.com/fix/f845ab/
See https://www.eevblog.com/forum/metrology/teardown-fluke-845aabar-tweaks-and-mods-(and-repairs)
Photos at https://doc.xdevs.com/doc/Fluke/845AB/img/f845_chopt.jpg and
https://www.eevblog.com/forum/metrology/teardown-fluke-845aabar-tweaks-and-mods-(and-repairs)/
reply #7 March 3 2016 by Jay_Diddy_B, attachment "green board.jpg" .

NOTE

A major circuit modification has been made to the Fluke
Model 845AB.  This change entails the complete redesign
of the Photo Modulator, previously based on
photo-conductive resistors.  The new circuit now employs
optically-isolated Bilateral Analog FETs, and is
fabricated on an additional circuit board.

Updates to the relevant sections of this manual are now
in process and will be available for shipment
approximately december 1993. This manual is completely
accurate with regard to specifications and operating
instructions. The circuit modifications are totally
internal and are completely transparent to the user.

End quote.

Nobody has the new manual.
One guy (TiN at xdevs and eevblog) has an updated instrument.
Another guy ("picburner" at eevblog) drew up a partial schematic,
https://www.eevblog.com/forum/metrology/teardown-fluke-845aabar-tweaks-and-mods-(and-repairs)/
reply #14 March 7 2016 by picburner, attachment "845A_324_RevB.pdf" .

He got it almost right.

ERRATA

1. U2 pins 3 and 4 are K2 and A2, respectively.  Therefore,
section 1 closes on positive and section 2 closes on negative.

2. For Q1 and Q2, the base resistor is R10 and R11 (1K); emitter
resistor is R12 and R13; pulldown resistor is R16 and R17 (10K).

3. The two H11F1's are U4 and U5.  U5 pin 6 goes to CN8 BLUE.
U4 pin 4 goes to R15 (220 ohms) to CN7 ORANGE to R14 (20K)
to U5 pin 4.

4. U4 pin 6 goes to R7 (1Meg variable) to R6 (562K).
This is offset trim.

5. U3 is misdrawn.  The input (marked pin 1) is actually pin 3.
The output (marked pin 3) is actually pin 1.  R4 goes from pin 1
to ground, not pin 1 to pin 3.  The junction of R5 and R6 goes
to pin 3.

6. I believe U3 is an LT1009 or LM336-2.5, wired as a negative
shunt reference.
Pin 3 is minus, pin 2 is plus, and pin 1 is trim.
It's not an LM385 whose pin 1 is a no-connect.
It's not a TL431 whose pin 2 is minus.

7. R3 is 2K not 1.6K .

PRINCIPLES OF OPERATION (speculative reconstruction)

DS101 and DS102 are removed.
R154 is changed from 33K to 6.5K.
C119 is changed from 0.22uF to 1uF.

Positive pulses from transformer T202 light transistor-output
HCPL2731 optoisolator U2 section one (A1/K1/C1) via R154, C119,
CR106, and ORANGE wire CN1 where DS101 used to go.  Negative
pulses light section two via R154, C119, CR107, and YELLOW wire
CN2 where DS102 used to go.  Where C119 used to create dead time
between DS101 and DS102 lit periods, now it ensures equal timing
even though T202's negative and positive peak voltage may be
different.  The increased load on T202 reduces its peak voltage
to less than U1's max rating of 36V.

Positive pulses also power U1 (an LT1120A regulator)
via RED wire CN5, rectifier CR1, and reservoir/bypass C1 and C7.

We can't read the printing on R1, so we don't know Vcc for sure,
but it ought to be less than 18V, HCPL2731 optoisolator U2's max
rating.  It's probably just 15V, being a separate supply to
keep current pulses off the original +15V unregulated rail.
(It was okay to use the old -15V rail to power U3, because
it's a constant DC load.  +15V means that R1 is 1Meg.

Reducing R154 increases the current driving the LEDs inside U2.
The new R * C product is the same as before, but this is misleading;
with the low start voltage of the LEDs, each LED is lit for the
entire half-cycle.  All the dead time comes from C4 and C5,
which also give a soft turn-on.  (I don't know if that's
important.)

When T202 goes positive, U2 section one is lit, shorting
C1 (pin 7) to ground, discharging C4 and extinguishing U5.
Meanwhile, C2 (pin 6) opens, allowing C5 to charge and Q2
to light U4.  The same action occurs on the opposite
components on the other half cycle.

Note that U5 is turned off (and U4 turned on) when DS101
used to illuminate V101, so the modulator and demodulator
have the opposite phase relationship to original construction.
This would reverse the meter polarity, so I am obviously
missing something.

C4 * R9 = 0.93ms .
With 15V, at phase begin, C4 or C5 charge at 16V/ms or
about 15mA/ms, but that doesn't continue.  After 1ms we're
up to about 10V or 7mA which is still way past U4/U5 turnon.
Factoring in transistor and LED diode drop, we probably get
an actual dead time on the order of 100us, which is plenty
for the H11F1.

U3, an LT1009 or LM336-2.5 -2.5V shunt reference, provides
offset trim to U4 and U5.  Minus pin 3 is powered with 6mA
from the -15V rail via R3.  Plus pin 2 is ground.  Trim
pin 1 is set by voltage divider R4/R5 which is wired across
Minus and Plus.  I presume this zeroes out U3's temperature
coefficient, since they are two different resistor types.

U4 and U5 act as an SPDT switch with dead time in the
middle.  U5 pin 6 or CN8 BLUE connects to the input
voltage on C103 in place of V101; U4 pin 6 or CN6 RED
connects to return/feedback in place of V102; their
junction, CN7 ORANGE, goes to the AC amp at C104.

When lit, U4 and U5 produce on the order of 100uV of
offset voltage, with source negative and drain positive.
Asymmetrical resistors R14 and R15 forming the common
point between U4 and U5 transmit the respective offsets
unequally, resulting in a net positive, which current
from U3 via R6 and R7 trims back to zero.  The trim
current is adjustable from 1.6uA to 4.4uA, developing
16 to 44uV across R114.

[Kleinstein]

Swapping the switches at the chopper would do more than just change the polarity of the reading. This part is inside a feedback look - so a reverserd polarity would cause wrong polarity feedback and thus not work at all. Chances are the H11F1 and neons would use a different transformer as they also need a much lower voltage.

[Dave Wise]

Oops, you're quite right, it would turn negative feedback into positive feedback.  That never ends well!

This board was intended to be an in-place retrofit.  I suppose they could have changed things elsewhere - after all, they changed R154 and C119.
I don't know if this was only applied to new production, but similar things were done as an upgrade performed on instruments brought in for service.  My 887A is an example of the latter, 1960's date codes everywhere, except one board from the 1990's.  (Different board but the same idea, ditch the old chopper.)

I don't think they changed T202.  The daughterboard has design elements that suggest it was working around characteristics of existing parts.  The only thing I know they did - and as far as I know, the only things they had to do - was
(a) Fit the board instead of DS101/DS102/V101/V102, and
(b) Change R154 and C119.
If you gloss over the phasing issue, that's all they would have had to do.

It would have been trivial to lay out the board with opposite phasing, so why do it wrong and then correct it elsewhere?  I just want to know if I've lost it.  Can you trace through and check me?
Here's a trace on the original circuit.

Let's say the input is positive.
V101 passes the positive when DS101 is lit, so the square wave phase is positive when DS101 is lit, negative when DS102 is lit.
The AC amp inverts, so it presents a square wave to Q106 that's negative when DS101 is lit and positive when DS102 is lit.
When T202 is positive at terminal 9, lighting DS101, it's negative at terminal 7, grounding Q106.
Therefore, negative peaks are grounded and positive peaks are passed.
Therefore, positive input to the 845A means positive input to the DC amp.
The DC amp is noninverting, so the meter reads upscale.

Thanks,
Dave Wise

[Dave Wise]

Here's my marked-up schematic, goes with my writeup and reflects the errata there.

[picburner]

Thanks for the corrections, I post the updated pdf.

R5 is connected right.

From an pix posted by TiN, even if blurred, I can read for U3 "LM" but I can't see the numbers.
Only a few unknown values remain that only those physically own the board can say what they are.

[Dave Wise]

Thanks for the updated schematic, picburner.  The only thing missing is annotation on the input/power leads.

ORANGE replaces DS101
YELLOW replaces DS102
GREEN is ground (at DS102)
BLUE is -15V
RED is T202 terminal 9 i.e. red wire in cable

I looked at every voltage reference Mouser knows about that begins with LM.  Rejecting parts where pin 3 is NC or pin 2 is minus (i.e. LM329, LM285/385, LM431, LM4040/4041), we are left with LM336 only.  It has 2.5V and 5V versions.

I still think the phase has gotten reversed, please review and help me figure out where I've gone wrong.

Dave

[picburner]

It could be that, instead of connecting the "HV" side of T202 intended for neons, they connected the other side of the secondary of T202, the one at low voltage, since the optocouplers require a lower voltage.
Someone on "volt-nuts" made a similar change.
This would explain the phase reversal but I can't see if it is so only look at the pics.

I have a doubt about the R3 value, maybe it's too low, it would load the -15V line very much.
Looking at the pic better the colors would suggest that it could be 20K or 200K.

[Dave Wise]

Pictures clearly show the orange and yellow wires soldered in place of DS101 and DS102.

The LM336 wants between 600uA and 10mA.  20K is not enough when you factor in the load current of R6.  200K no way.  2K draws a comfortable 5mA or 7.5mA, what's wrong with that?

[TiN]

I can take few more pics of chopper board, as I found 845AB guts while cleaning up the place.

[Fred_47]

The LM336 wants between 600uA and 10mA.  20K is not enough when you factor in the load current of R6.  200K no way.  2K draws a comfortable 5mA or 7.5mA, what's wrong with that?

I greatly appreciate the work you and picburner have done to document this 'secret' Fluke mod.

The 845 ckt is practically identical to the null detector used in the 335A/D and suffers the same problem with aged neon lamps in the input chopper.

I've been working all summer on a replacement for the neons or the entire photochopper in my 335D.

Initially I looked at MOSFETs similar to the Keithley 155 but I couldn't get rid of the switching spikes. Then replacing the neons with LEDs. I actually made a new block out of Al to replace the neon blocks and designed a constant current ckt to drive the LEDs. That sort of worked but there was an offset beyond the null zero capacity. I blamed that on the difference in the photo resistor on/off resistances. V1 had a much larger swing than V2.
 
Right now I have H11F3 optocouplers with the diodes driven thru a constant current circuit powered by terminal 7 of T1 (T202 in the 845). The offset is beyond the range of the zero null control in the positive direction on ranges 1mV and higher and in the negative direction for the 2 lowest ranges on the 335D, 100uV & 10uV.

I've attached my latest schematic.

Could Fluke have hand selected LM336's that work at low currents, or just ignored the minimum current?

As part of exploring alternate chopper lamps (LEDs) I measured the ±15v supplies at different loads. Winding 7-8 of T202 (T1 on the 335A/D null det PS board) has a high DC resistance, ≈600 and terrible regulation. Several posts upthread comment on the low voltage. Winding 8-9, which powers the neons, is about 4k .

My testing indicated a voltage droop of 1-2V per mA of load on each of the 15V supplies.

Other testing indicated that the meter ckt takes a about 1.25mA for full scale deflection. The meter in the 845 is different from the one in the 335A/D so I don't know if this carries over.
 
Since these transformers (and the meter) are the ultimate unobtainium, I've been very leery of overloading them.

Can others measure the DC resistance of their transformer to confirm, or deny, my measurement?

edit to correct T202 was T201.

[dietert1]

When i experimented with optofets i made the mistake of giving the LEDs currents of 5 to 10 mA, like other opto couplers. It works much better at 1 or 2 mA. Probably you also want to have separate adjustment of the two LED currents in order to avoid selecting optofets.

Regards, Dieter

[Fred_47]

When i experimented with optofets i made the mistake of giving the LEDs currents of 5 to 10 mA, like other opto couplers. It works much better at 1 or 2 mA. Probably you also want to have separate adjustment of the two LED currents in order to avoid selecting optofets.

Regards, Dieter

At this time I'm controlling the LED current with a jFET in a bridge rectifier, similar to the preregulator ckt used in the 335/332/A/D. I've set it to 2mA. Being able to individually control the LEDs may be a necessary change.

I've been resisting the use of brute force, i.e., using a nulling ckt to overwhelm the offset, in favor of fixing the source of the offset (if I can find it). Apparently Fluke didn't think that is the way to go. Maybe there are so many different ways that the ckt could be off that they decided to use a one fix fixes all ills approach.

Looking at Fluke's optoFET ckt., I'm wondering what is the purpose of R14 & R15 between the FETs?

[Dave Wise]

R14 and R15 intentionally generate an offset, so that the trim circuit can be unipolar and still cover a range from left of zero to right of zero.

[notfaded1]

So the neon bulbs arrived today. Here is a direct comparison of the new AC3/NE2U (left), the originals bulbs that were installed (middle) and the Barthelme 82305 (right). However, the Barthelme 82305 do work for me. Enough spare parts for the future though

-branadic-

banadic-

Did you ever try the new longer NE2U spec'd in the manual in the 845AR?  I'm curious because I have the same exact box of 10 tubes coming and I would think, from my experience with neon, that they would produce more light.  BTW with neon they're called tubes not bulbs FYI.  That's a querk for neon lovers and common error for neon newbs.  https://en.wikipedia.org/wiki/Neon_lighting

Thanks for all your work on this... it saves me a LOT of time reinventing the wheel.  Also I appreciate you EE's putting up with me.  I don't have the EE experience you all have but... I love this stuff and it takes my mind off the computer science, networks, and security I have to deal with during the day at my real job.  I know I've irritated Dr. Frank a couple times.  I hope you guys can forgive me.  I love all the graphs, pictures, and videos people post but most especially the measurement data and tips on how to fix stuff. 

Bill

[Dr. Frank]

I'm not aware that you irritated me, so don't worry...

Maybe you can check these NE2U on your own, first if the dimensions are ok, and second, whether they show that Dark Effect. This test is easy to do, like I have described.

If they show no Dark Effect, just plug them in.

Frank

[branadic]

notfaded1

I bought AC3/NE2U, but haven't tested them in circuit, as the Barthelme 82305 already worked for me. See also this post https://www.eevblog.com/forum/metrology/teardown-fluke-845aabar-tweaks-and-mods-(and-repairs)/msg2597721/#msg2597721

-branadic-

[notfaded1]

Thanks I figured that's what you did... well I'm going to try them when they come so we'll see what happens.  Also a maybe dumb question.  I think I know the answer.  Does (on the low ranges) the needle jump up if you get you hand near the inputs while in OP mode?  I printed out the full manual... it seems to really really help on this device.  I knew exactly what you were talking about testing with shorted inputs and the full deflection... it's all in the manual.

Best Regards,

Bill

[branadic]

It does, though it depends on how you put your hands on the inputs. If you put it symmetrical to both input connectors (high and low), the needle stays almost where it was.

-branadic-

[guenthert]

Does (on the low ranges) the needle jump up if you get you hand near the inputs while in OP mode? 

  That would be either due to thermal EMF (if nothing is connected to the inputs that shouldn't happen, as the junctions which could then contribute would be inside the case, protected from the air movement) or electrcal fields.  The Fluke is shipped in a configuration with 1MOhm input impedance, but can be reconfigured to high impedance (>100MOhm iirc).  The manual probably has a performance evaluation section, which might have a subsection determining the input impedance.  If the instrument is new to you, it might be worthwhile going through that in any case.  There should also be a section with grounding/shielding recommendations.

  I always wondered why Null meters have their inputs next to the operational elements ...   I think Keithley got it eventually right with a 'remote' Nanovoltmeter, separating pre-amplifier and user interface.

[notfaded1]

Here's the datasheet for the various NE-2 type neon tubes.  The F845AR manual calls for NE2U neon tubes.  The stats vary on these tubes.  An interesting thing with these, which makes sense, if you feed them DC then one of the metal bars inside (acting as the cathode) lights up.  If you feed it AC both sides light up because the bars take turns being the cathode lighting on the alternating current.

Bill

[dietert1]

Yesterday i worked on a FLuke 845AR. It is working very well now and shows small movements up to about +/- 50 to +/- 100 nV in "ZERO" mode, which seems to be normal or lower than normal.

Renewed both neon tubes and changed the 10.5 V shunt regulator (meant as a battery charger) to a regulated supply using a LM7810 plus one diode offset. R202 became 10R and C201 gets 24 V now. I also replaced all carbon resistors as proposed by TIN. Most of them were off tolerance, especially those of 100K and above. One 680K had arrived at 790K. Removed the transistor sockets and renewed all caps, replacing 6.8 uF, 2x 10uF and 15 uF by film capacitors. Didn't do any adjustments yet.

I already noticed extra noise from mains/guard connections, so maybe the device needs batteries and/or some mains filter.
Should we replace R127 and CR103 by an AD587? What did others try to reduce noise?

Regards, Dieter

[Kleinstein]

I don't think CR103 is a major noise source and thus not much gained from replacing it with a lower noise but expensive AD587. Zener noise can be quite different between types.  A TL431 + resistors could be an option too if the old zener should turn out too noisy. There is a change to get lower noise from 2x5 V zeners instead of 1 x 10 V. However this likely also comes with a different TC ( usually negative with 5 V and positive with 10 V).

Chances are other parts will dominate the noise: The resistors in the input filter, Q101 as a BJT is not the best choice for a high impedance signal and the neon lamps even if new can cause jitter when starting. The new neons and replacing some resistors could have already helped with slightly lower noise.

So to get really lower noise is would be more like a new design.

[dietert1]

Today i looked at the AC amplifier output (TP7) and found a tremendous noise level of about 12 Vpp with occasional limiting clearly visible on the scope. Limiting hurts. When running the device from an external power supply noise level went down to less than 50 %, no longer limiting. Output noise went down to about +/- 25 nV level, which is almost perfect considering the visual resolution of about 10 nV in the 1 uV range.

Will put two 10 KOhm resistors in front of primary side of mains transformer and see what happens. Such resistors are very effective RF blockers, too. Power consumption on the 10.5 V is 30 to 40 mA.

The photo resistors with their light guides are extremely sensitive to the slightest scatters/knocks, maybe even microphony. Don't know yet how to improve that.

Regards, Dieter

PS:
With R202 replaced by 10 Ohm, my LM7810 based 10.5 V supply runs perfect with 2x 4K7 in the mains connection. This is a 230V configuration and the instrument gets almost a factor 2 more quiet with this change. So there is magnetic coupling from the mains transformer into the guarded section.

The +/- 15 V guarded supply is one weak point of the design. The low voltage secondary of T202 has a resistance of 630 Ohms. The saturation driven 84 Hz oscillator shows strange changes of frequency and duty cycle depending on what the meter outputs. Apparently the isolated output circuit takes a good fraction of power, especially in out of range condition. For the time being i just put 15V zeners parallel to C117 and C118. This works well as long as there is no out of range condition.

I think what is causing noise is the missing 84 Hz filter before the sampler. In this type of synchronous detector one either uses a linear mixer driven with a sine wave or a narrow band filter followed by the sampler. Sampling into wide band noise like in the Fluke 845AR isn't the best design. Maybe changing some caps can make the existing AC amplifier more narrow band.

Also i don't yet understand the amplifier gains. The AC amplifier Q101..Q103 makes about 4700x (47K/10R). The AC amplifier Q104..Q105 makes about 220x (22K/100). The DC output amplifier makes 260x (3M9/15K). So in total there are 2.69 E8, which is 269x more than we need to get from 1 uV to 1 V. You want some negative feedback reserve, but maybe the instrument can work better at lower gain. Will try that, too.

And the sensitivity to mechanical touch/shock is from the light guides, which are free to move about 1 mm in my instrument, with terrible effects on the measurement.

[dietert1]

Circuit revision revealed another problem. The operational amplifier formed by Q107 .. Q112 has the isolation drivers Q113 and Q114 as load. This load generates 200 to 300 mV peaks on the opamp output that go directly to the inverting input (base of Q108) via the integrator cap C116. The opamp can't perform well like that. The problem is easy to cure though. I used a bipolar 22 uF cap as buffer mounted close to Q113 and Q114. Another mod is a transistor gate limiting the integrator output to about +/- 1.3 V in order to avoid +/- 15 V supply breakdown.

I also put LEDs instead of neon tubes and did some other mods mentioned in this thread, like tuning the capacitors for limited bandwidth. I also put a p-channel FET as sampler (replacement of Q104). The instrument is slowly becoming something useful.

Regards, Dieter

[Dr. Frank]

Hello Dieter T.,
please explain your use case for your improved 845AR.
I mean, it was intended for bridge nulling, with drift and noise parameters as is, so giving 200nV..500nV safe zero indication.
The undisclosed parameter was extremely low bias current in the fA range.
This instrument was built for alignment of 335D, 720A, 752A, but not much more, I think.

So I'm extremely curious what application you have in mind for an even improved 845A.

I fully agree with you, that this 1967 design has lots of disadvantages, which could easily improved, as you have described. Maybe you could more precisely describe, including a quick schematic, which components you have replaced, and how.
In particular, I didn't fully understand how you changed the 10.5V circuit, and where you placed this bipolar 22µF capacitor.

Thank you!

Frank

[Kleinstein]

I would not expect leakage much below a few pA. Still quite good, but AFAIK not much information on this.

The 10.5 V part should be relatively easy: just replace the shunt type 10.5 V regulation with CR207, C201 and R202 with a filter cap and linear regulator. This should be easy and may reduce the heat generated.

Getting some pulses, interference from the chopper for the isolated output also makes sense. The simple buffer with Q111 and Q112 just has quite some cross over distortion and thus can not react fast to load changes. Not sure if just a cap at the output is such a good idea - I would at least include some series resistance.

The 845 circuit is old in many aspects, so not sure if local fixes can really improve the performance so much.

By modern standards there are several not so ideal points in the circuit: the neons (flickering, aging), a BJT input AC amplifier, the rather large caps at the input (and thus even with PP caps some input current from DA), the 10.5 V shunt regulator, the Q111/Q112 buffer with large dead zone, BJTs for chopping (need special types with high reverse UBE tolerance), C114 +C116 as electrolytic caps, rather high power consumption, combined power and sampling clock (this also has a few good points).

So it may be more logical to use a new design and keep the 845 close to original as an antique example of it's time.

P.S:  the gain of the amplifier without the final feedback needs to be high so that the final feedback has something to compensate errors. Only a factor of 260 extra is not that much. So less gain is not a good idea. With a better DC amplifier and demodulator, there could be less AC gain and more DC gain.

[dietert1]

A very similar modern circuit can be found here, see Fig. 5:

http://www.janascard.cz/PDF/Design%20of%20ultra%20low%20noise%20amplifiers.pdf

The author claims low noise (8 nVpp).
Or one can buy a Keithley nanovoltmeter for several thousand.
Or one uses a ADS1256 that includes a chopper stabilized amplifier.

My intention is to update the Fluke 845AR to a similar low noise level. Using the old Fluke there is a cabinet with multiple shields, a meter, a power supply, a range switch, low thermal EMF connectors, output isolation. I don't see any reason why it should not perform to todays standards when using modern parts. Power consumption isn't that bad, the whole device needs about 300 mW, so it can be run with batteries. And i think someone in this forum measured input current to be 15 fA (MisterDiodes?). Changing the input stage from bipolar to FET in order to reduce current noise should be a simple exercise. I got some 2SK369 FETs from another project, where 8 matched FETs where  selected out of 20. Yes, i can show my schematic when i'm ready.

Today i found that reducing AC gain just reduces the speed of the instrument (integrator gain). The needle will creep, no real improvement.

Regards, Dieter

[valley001]

Hi all, great thread here.  Still absorbing much of it.  I have also been working on an 845AR, I hope it is ok to join in on this. 

I have recapped mine and installed new neon's that seem to be stable for the moment at least.  The unit seems to function correctly. 

As I work with the unit I notice it is sometimes difficult to zero the 1uv range, the zero pot is too sensitive and getting a good absolute zero is difficult due to under and over shooting the spot.  I have checked it against the specified "at least" +/- 5uv and see that in practice mine gives +8 and -6 or so.  I realize this is technically in spec but I would like to to be slightly less sensitive.  Is there a component I can change to slightly reduce the sensitivity of the zero pot?

[dietert1]

If you increase R116, for example to 2M2, sensitivity will be less. I did this on my instrument and i think i will increase it even more. You loose the capability to compensate some microvolts of thermal EMF, but that doesn't help anyway. Thermal EMF isn't a stable offset to compensate.

The adjustment potentiometer R117 itself is a little suspect concerning noise. It has 5M, probably a carbon resistor and probably increased to 7 or 8M after all that time. Will probably renew it with a much smaller resistor. As recommended above, one should change the bias generation +/- 0.5 V to something stable, instead of taking it from the AC amplifiers. Currently i am using two resistor dividers to get the +/- 0.5 V from the +/- 15 V.

Regards, Dieter

PS: On my Instrument R117 just measured 8.26 Meg.

[valley001]

Interesting so R117 is a problem. Original specification is 5 meg +/- 30% so well out of tolerance.  I hesitate to start measuring all the carbon resistors for fear of going through and replacing them all!   This brings up another question I have, I read back somewhere in this thread that using non carbon resistors (metal film etc) for replacements may not be recommended because carbon resistors are non inductive.  Is this correct?

So, 2.2 meg ohm replace for R116 will reduce how many uv sensitivity of the zero range pot, approximately? 

If you increase R116, for example to 2M2, sensitivity will be less. I did this on my instrument and i think i will increase it even more. You loose the capability to compensate some microvolts of thermal EMF, but that doesn't help anyway. Thermal EMF isn't a stable offset to compensate.

The adjustment potentiometer R117 itself is a little suspect concerning noise. It has 5M, probably a carbon resistor and probably increased to 7 or 8M after all that time. Will probably renew it with a much smaller resistor. As recommended above, one should change the bias generation +/- 0.5 V to something stable, instead of taking it from the AC amplifiers. Currently i am using two resistor dividers to get the +/- 0.5 V from the +/- 15 V.

Regards, Dieter

PS: On my Instrument R117 just measured 8.26 Meg.

[dietert1]

As far as i understand several people inserted metal film resistors and there was no report about oscillations. So i wouldn't think a lot about that remark of 2N3055.
The zero pot inserts current into the feedback circuit at R114. Near the center of the potentiometer its nominal output resistance is about 2.5 MOhm, so increasing R116 to 2M2 will change the total output resistance from 680K+2M5=3M18 to 2M2+2M5=4M7, so sensitivity in the middle will be down to roughly 0.66. The maximum + and - offsets will be down to about 0.3 with respect to the original circuit.

My instrument gets nulled in the +/- 1 uV range when the potentiometer outputs 1.27 mV. So the chopper has almost perfect balance all by itself and the bias of R117 could be +/- 20 mV - a factor 25 less than the original design.

Regards, Dieter

[Dr. Frank]

Obviously, TiN has blown up his unit by or during replacement of all carbon resistors:

Doh, I replaced carbon (which were 5-10% high) resistors with modern ones in my 845AB, and now unit is dead
It was working before, just noisy, but now it's just overflow either way and just stays there, on any range.
Battery is ok +9.4V, +15V and -15V are ok too.

My own 845AR sometimes also shows signs of instability.
When I'm in the lowest ranges, i.e. 1 or 3µV, and have a probably noisy input signal, the 845AR just jumps into overload, and does not get back to zero, even if the bridge I'm measuring is probably nearly balanced.
So I have to range to 100µV or higher, and after the 845AR is displaying zero there, carefully go back to full sensitivity, where it really shows a few 100nV only.
I always assumed, that my unit is having problems with oscillations.
The story told by dietert1 now supports my assumption, maybe some carbon resistors also drifted too much.

Up to now, I only replaced several electrolytics and the neons, and didn't want to touch a running device, because I only want to get the alignment of my bridges done, at ~ 200nV_pp noise level, and <<1pA bias, which is fully sufficient for me.

Maybe I will have a closer look now, how the different circuit parts are behaving.

Frank   

[dietert1]

To solve the instability under overload conditions
- i increased the power supply from 10.5 V to 12 V. I am using LEDs, neons will get higher voltage, too.
- put 15 V zener diodes over C117 and C118
- made a limiter for the integrator that keeps the output of Q111/Q112 within +/- 1.3 V.

Attached sketch also shows the buffer cap close to C113/C114 i mentioned before.

Regards, Dieter

[Kleinstein]

Taking the reference for offset compensation or some  +-0.5 V from the amplifiers is not that bad. The voltage is essentially a diode drop in both cases. If the needed offset compensation is so small the value does not need to be that stable.

How much offset there really is to compensate, can be different between units (e.g with the balance in the chipper and neons). If the range is sufficient I see no problem increasing R116 to make the range smaller. One could alternatively use series resistors and a lower resistance pot (may be easier to get than a 5 M pot).

The suggested modifications on the supply make absolute sense.  The divider for the base voltage reduces the need for transistors that can tolerate a high reverse base emitter voltage.

If overload to the output stage that effects the supply voltage, it is possible that the circuit gets a kind of second stable point in overload. So the suggested modifications of the supply could fix the problem of the amplifier getting stuck in overload.

One weak point could still be the buffer with Q111/Q112 with a large dead zone and C115 in front.

[Dave Wise]

I am reminded of my Fluke 887A AC/DC Differential Voltmeter.  Some years ago I noticed that if the instrument saw a transient of the right size/shape, the amplifier in the AC converter would break into oscillation, which would not stop except by turning off the power.  In troubleshooting, I discovered that several transistors were saturating during the oscillation cycle, and their charge storage time was so long that it introduced more than 90 degrees of phase shift.  This turned the overall negative feedback into positive feedback.

I fixed it by preventing saturation, by means of a "Baker Clamp".  Just a 1N5711 Schottky diode antiparallel to the base-collector junction.  When the transistor approaches saturation, the diode absorbs current that would otherwise become excess base charge.

I don't know if your 845 is performing the same perverse trick, but I recommend you check it.

HTH,
Dave Wise

[dietert1]

The 845AR does not have an AC mode, only DC. Its AC amplifier sits behind the chopper and is well protected from fast transients.

Kleinstein's hint concerning C115 was correct. The DC amplifier (Q107 .. Q112) remains stable when removing that capacitor (at least with my 22 uF buffer cap in place) and of course it works much better. To be safe i put in a 1 nF in series with a 100 Ohm resistor.
Also i added two 22 uF buffer caps for the collectors of Q111 and Q112. The spikes on the output of the amplifier vanished (< 1 mV).
When nulled i measure 0.25 V across each of the 1 KOhm resistors R158 and R159. The whole amp takes about 1 mA.

What worries me are the mains cable that "tunnels" the guarded section and the Null switch located near the mains switch without shield. I think the null switch mount lacks proper grounding. Also i would like to check R114 for thermal EMF, maybe i can find something better. Amazing how well the instrument worked with all these little issues.

Another thing i will try is doubling the chopper frequency, the photo resistors are fast enough for that. The modern circuit i mentioned before supposed to reach 8 nVpp has a 1 KHz chopper.

Regards, Dieter

[Kleinstein]

A higher chopper frequency is mainly needed if the AC amplifier has 1/f noise. The old BJT based version should have very low 1/f noise. The other point would be filtering at the input and possible aliasing. Here a higher frequency helps to filter with smaller caps.
The down side of a higher chopper frequency is that one gets mode charge flow from Charge pumping of the chopper. So the input bias tends to go up. 

The main idea with such a chopper amplifier made from separate parts instead of a ready made chopper stabilized OP is that one can use slow chopping, as one is not limited to small capacitors and a CMOS amplifier with high 1/f noise.

The 845 chopper can never reach the very low noise levels of he circuit here:
http://www.janascard.cz/PDF/Design%20of%20ultra%20low%20noise%20amplifiers.pdf
as there is already way more noise from the 300 K of resistance at the input.
The noise comparison off some 200 nV for the 845AR  and 8 nV for the CMOS chip chopper is in addition using a larger BW for the modern circuit. So the ratio is even larger. The two circuits are kind of different ballparks: the 845 is low bias (though I still have some doubt in the < 1 pA number), but high noise. The low noise circuit should have a higher bias current (more like 20-200 pA), but a much lower noise (e.g. 50 times less at the same BW).

[valley001]

Dieter, did placing the resistors on the primary side of the mains transformer have any effect?  I am interested in some of these less invasive modifications for somewhat improved performance. 

[dietert1]

Yes, soldering in the guarded section is a bit tricky. The board is two-sided and has several vias not plated-through. For example when you unsolder C115 and insert nothing, the instrument no longer works because one of the wires of that capacitor is needed as a via. The board does not have solder masks. Quite easy to rip off traces when unsoldering parts or to produce shorts.

When i installed the resistors into the mains connection (on the other board), noise at the output of the AC amplifier came down from 1.5 Vrms to 0.8 Vrms and it stopped clipping. This was with the two upper covers open. Low frequency noise on the DC output should be down proportionally.

@ Kleinstein: I was proposing exactly that: choosing a chopper frequency. Above 100 Hz the device may work better, with some distance from the 100 Hz fundamental present in the power supply currents (magnetic coupling).

Regards, Dieter

[Kleinstein]

Dieter, did placing the resistors on the primary side of the mains transformer have any effect?  I am interested in some of these less invasive modifications for somewhat improved performance.

Resistance at the transformer primary acts a little like R202. So one may have to/can reduce R202 to compensate. Resistance at the primary side could reduce the magnetization peaks in the transformer a little and this way slightly reduce the stray field. However otherwise it is rather similar to R202.  If used, the resistors should be well behaved on overload, e.g. fusible resistors.

[Dr. Frank]

Yes, soldering in the guarded section is a bit tricky. The board is two-sided and has several vias not plated-through. For example when you unsolder C115 and insert nothing, the instrument no longer works because one of the wires of that capacitor is needed as a via. The board does not have solder masks. It is quite easy to rip off traces when unsoldering parts or to produce shorts.

..

Regards, Dieter

@ TiN: I suspect that's why your 845 does not work anymore (since 2016) after replacement of all carbon resistors!
Maybe you take the wrecked instrument out of your repository and solder on the upper side, also.

Frank

[dietert1]

Maybe in a lab with multiple 3458As you can use one of them as a null meter in case you need one. Even our 3456As perform almost as well a native 845AR.

Regards, Dieter

[Kleinstein]

The 3458/3456 should be lower noise (over the same waiting time), but chances are it would have a higher input current and more current noise.
Also battery operation is not really practical with the 3458/3456  .  AFAIK battery operation is not included with the 845AR  , but it sounds very doable and a point I would really like with a null-meter.
About the only really good point of the 845 is that is can have very low bias and low current noise (not so sure about this - it would need to increase the 1 M resistor originally present at the input, as this alone causes some 100 fA/Sqrt(Hz)).

[valley001]

I notice my unit will drift 300-500nv over night, either to negative or positive.  The specification for "stability of zero" (drift?) is less than .15uv/hour and less than .3uv/day, so I think I may be slightly out of that specification.   At first I thought this may be temperature related but as the house warms up into the afternoon the needle does not travel back to zero.  Where would I look to improve this? ...or perhaps I am too picky. 

[Dave Wise]

The 845AR does not have an AC mode, only DC.
<snip>
Regards, Dieter

I'm quite aware of that.  My point is that any amplifier that has feedback and whose transistors can saturate is susceptible to the phenomenon I described.

Can you support your claim that it's possible to run the 845A successfully with 2x chopping frequency?  I have two HP 740B DC Voltage Standard / Differential Voltmeters, and both instruments' photocells are insufficient at 1x frequency.  Yes, the neon lamps are good.  I even tried driving the cells with high-brightness LEDs, tuning intensity and pulse width individually for each cell.  It was not possible to pass the tests for Input Resistance and Efficiency simultaneously; making one property adequate made the other unacceptable.

Regards,
Dave Wise

[dietert1]

Removed the low pass filter and recorded the AC amplifier output TP7 averaging 1024 tracks. The instrument has 1 uV input in the +/- 1uV range with the integrator shorted, so the chopper really sees 1 uV difference. Channel 1 shows the 84 Hz signal used for trigger. Without low pass the signal is very noisy and my measurement on channel 2 is small. So i made a math track to produce a 4x vertical zoom. I think the measurement demonstrates that the photo resistors in my instrument have a risetime of about 1 msec and are good enough for 168 Hz.

Regards, Dieter

[Kleinstein]

The AC signal looks rather odd, far from what one would expect from a simple circuit.

There quite some fast transients during switching, that look like they are no coming from the LDR, but more like direct capacitive coupling or possibly an effect of Q106 effecting TP7. But also the slow part look not that ideal, as if the is some coupling capacitor or similar (e.g. C106, C110) too small, so that the gain goes down towards low frequency.
The rise time part looks not that slow, but it still is quite some fraction of the period. So I don't think a higher frequency would really help, more like make things worse.

[dietert1]

Are you trying to troll my contributions? Do you want to do the work of tuning the filters for the new chopper frequency? Sorry, your comments are good for nothing.

[Simon]

Are you trying to troll my contributions? Do you want to do the work of tuning the filters for the new chopper frequency? Sorry, your comments are good for nothing.

If you cannot stand comment then why post? If you don't like the fact that others can say things then: leave!

[Dave Wise]

You will lose some efficiency as the speed increases.  Make sure you still have enough open-loop gain.

[dietert1]

Your advice is based on your experience with HP gear. Why do you think Fluke was using the same photo resistors? Do you have a part id or datasheet of the photo resistors? The ones in my 845AR are sealed in glas tubes somewhat similar to the neon tubes, except with a flat window on the front.

Yes, i understood that changing the sampler frequency requires a lot more than just changing C203. Just wanted to provide some answer and did not have time to finish everything. Of course i will do more measurements.
I also read your remarks concerning the LED driving pulse shape. So today i put back a capacitor similar to C119, except somewhat bigger than before. What was important before for fast neon tube ignition (this trick missing in HP 740B circuit), now provides stronger LED pulses with some decay. Maybe the photo resistor turn-off time is even more critical than the turn-on time.

Regards, Dieter

[Dave Wise]

I'm familiar with the HP instruments from first-hand experience.  I don't own an 845A.  (Hey TiN, want to ship yours to me?)

I expect similar outcome because I believe the effort was similar.  HP made their own cells, using a proprietary blend of Cadmium Selenide and Cadmium Sulfide.  (They're molded in some kind of clear plastic, and they have individual serial numbers.)  The CdSe must have been included to achieve acceptable speed, at the expense of short service life.  I presume Fluke encountered similar obstacles and employed similar techniques to overcome them.  Part of the reason their instruments have continued working longer is that they used photocells only where necessary, and used transistor switches for demodulation, where microvolt accuracy isn't needed.

I will be interested to hear of your ongoing progress.

[Kleinstein]

Usually LDRs are faster turning on than they are turning off. There is a superposition of several time constants and the faster part is sufficient to get a reasonable on, while one has to wait for the slower parts to really get an off.
So chances are there is a time when both LDRs a conducting somewhat.
It may be a good thing that there can be a dead time when non of the neons  is lid. There is a resistor at the input anyway so some cross conduction is not that bad.

I don't see a real advantage of increasing the modulation frequency, as I don't expect the BJT based amplifier to have a lot of 1/f noise. The filter frequency at the input is already well lower than 80 Hz, so no problem there either.

[dietert1]

Dave Wise, if i understood correctly, the life span of those LDRs  in the chopper of our Fluke 845ARs may be over. Is this what you meant? Is there any reference explaining the higher speed and the shorter lifetime of CdSe elements in comparison to CdS? Couldn't find anything on the web.

Received an offer for FF627 fotofets, maybe they are a good replacement and pave the way to higher chopper frequency.

Regards, Dieter

[Kleinstein]

The Photo Fets tend to have some offset-voltage, at least the H11F1 used in some newer version. Newer Photo-mos might be an alternative, though usually in a lower impedance range.

The main real advantage of a higher frequency is that filtering against aliasing is easier. However the present filter for the 845AR is way below 80 Hz.  The way to lower noise would be with lower resistance in the filter, maybe lower resistance in the chopper (i don't know the LDR resistance range), lower noise of the AC amplifier, more optimized waveform for demodulation (e.g. dead zone) and possibly going from a simple 2 switch chopper to a full polarity reversing chopper with 4 switches (this would also reduce the need for an anti aliasing filter).

Looking for photoconduction aging and CdSe, there seem to quite a bit of research done. From a first fast look things that happen may be oxide coming in and changes in the crystal structure (the thin films are far from equilibrium). At least for the oxigen effect there would be a slight advantage for the glass encapsulated version over epoxy. Increasing dark conductivity could be an issue. So the dark resistance might be used as a first indication of aging.

[Dave Wise]

Deiter: I would be somewhat suspicious of the photocells, but I don't assert that every one out there is dead.  But they may be dying.

Kleinstein, I believe the average resistance of one of these photocells while driven in a continuous on-off cycle is on the order of 10K-100K.

I can't remember where I read about CdSe, but it was faster than CdS but less stable over temperature.  And, I think, time.
What I observed in my old HP cells is loss of sensitivity but more importantly, loss of speed.  Most of their low efficiency as modulators is their slowness in turning off.  There is significant overlap between the turned-off one ceasing to conduct, and the turned-on one beginning to conduct.

H11F PhotoFETs generate quite a large offset!  I've seen everything from 0 to 200uV.  It varies with temperature and with LED drive current.  Some are positive (voltage on pin 4 with respect to pin 6), some are negative.  When I put two in a "totem-pole" topology for a modulator, I orient them so their offsets oppose so they at least partly cancel.  But after studying Fluke's H11F1 replacement board for the 845A modulator, I have concluded that I must manually cancel the remaining offset, and it's more important to match offset TC than offset amplitude.

Recently I played with a PhotoMOS relay, type ASSR-4114.  (Because I had one.)  It is unusable - it generates a large spike at turn-on, on the order of 20mV.  Yes, mV not uV.  I think it's due to gate capacitance imbalance, which means all PhotoMOS relays are disqualified.

The FF627/LS627 data sheets confuse me.  How do you operate one of the parts?  I'll be on the edge of my seat as you experiment with them.

[dietert1]

Meanwhile i checked the photocells (resistance measurements with static illumination) and they agree to each other within a factor two at various largely different illumination levels. With the orange LEDs i used the LDRs had 5,5 KOhm and 27 KOhm, dark resistance was above 20 MOhm.

One error i had made was selecting orange LEDs (611 nm). Their light appeared similar in color to the neons, but i learned that LDRs are rather selective at certain wavelengths, so they pick certain strong lines from the Neon spectrum.
I tried to use a white LED with monochromator to find the wavelength of maximum sensitivity, but without clear result after one or two hours. Trying a three color LED i found the LDRs were more sensitive to green and blue light, which means the LDRs are probably rather pure CdS. With those really bright green LEDs (e.g. HLMP-CM3A-Z10DD) the LDRs both reach below 1 KOhm.
In the Keithley 148 nanovoltmeter manual they explain a little what source resistance means for noise in the nanovolt region and the LDRs add to source resistance. Somehow i learned what we knew before from the Neon tube repairs: Instruments with low light intensity will be noisy.

FotoFets have been ordered, but will take some time.

Regards, Dieter

[Dave Wise]

When I experimented with LED illumination of HP photocells, I used surface-mount LumiLeds parts in the Luxeon product line.  I believe these are blue driving a red-green phosphor mix.  They are super bright compared to neon at the spec-sheet Typical drive current, and in designs where modulator input resistance is important, you have to decrease the current or the duty cycle to avoid excessive overlap.

Most of my cells had similar steady-state lit resistance but various different resistances at the normal duty cycle, I presume because they reacted at different speeds.

Slightly off-topic, the 740B front-end contains a special high-voltage photocell intended to isolate the input during overload or in STD mode.  It exhibited large photovoltaic response, 10-100 microvolts when lit.  I had to do away with it.  I installed current limiters based on the Supertex LND150 depletion-mode MOSFET, and an Ixys CPC1981Y SSR. </ot>

[notfaded1]

Somehow i learned what we knew before from the Neon tube repairs: Instruments with low light intensity will be noisy.

I have the box of new Fluke manual spec'd NE2U tubes for my 845AR now... I may try swapping them in this weekend.  I had to read that sentence a couple times but I get it now... you mean when our old neon tubes are sputtered with cathode material inside the glass envelope they don't produce as much light output and thus the photocells output more noisy resistance levels.  I think you're probably on the right track with the 3 color LED's and finding the wavelength that works best... I'll be interested to hear if you find an ideal solution.  It sounds like you're on the path to rebuilding the 845AR into a whole new animal now o.O!  Why not improve on 1969 technology in the year 2020 right?  God forbid it might work better than a 3458A?

Bill

[Kleinstein]

For the LED color chances are high that the green ones are good, yellow ones may also work well, but usually the efficiency is relatively low (about the worst wavelength in the visible range) at yellow and yellowish green.

I don't think the higher noise is about to little light. It is more about flickering light as not the whole electrode is lit evenly every time.

In principle the LDRs can also show some photovoltaic effect from the contracts, especially if unevenly illuminated. At some point more light would do more harm from PV effects and maybe thermal effects and even lower resistance when on would not help very much. So there is an optimum intensity. Some 5 -10 K resistance may be low enough.

[dietert1]

One problem with LEDs in the Fluke 845AR is its low power design, so the LEDs need to run at average currents of 1 or 2 mA. Higher LED currents can ruin once more the stability of the +/- 15V supplies. Currently i have the two LEDs anti parallel with 1 uF plus 5.6 KOhms in series, driven from the low voltage secondary (+/- 15.7 V at 84 Hz). That seems to work very well. Since i didn't want to drill the original neon mounts, i used some aluminum i found in the workshop plus black shrink-on tube.

Regards, Dieter

[notfaded1]

Another symptom of sputtered cathode material is it can also build up on the cathodes themselves and cause part of the cathode to not form the plasma completely (cathode poisoning) at least that's how I've always thought of it.  I'm a big collector of nixie tubes so I have a long history with cathodes in neon gas.  You can't do anything about the built up material inside the envelope (they're trash at this point) but for some of the larger more expensive nixie tubes we often run extra voltage through them for a while and it'll burn off the built up material off the cathodes.  I've rejuvenated many nixie tubes this way... it'll never be the same as NOS but for say a 300+$ tube it's nice to bring them back to seeing the entire digits on the cathodes.  Sometimes it takes running them hot for a day or two even for each cathode with cathode poisoning on it.

Bill

[Dave Wise]

Fluke's own LED retrofit (using H11F1's) appears to have specifically avoided using the +/-15V rails to drive the LEDs.  Instead, they repurposed the transformer winding that previously drove the neons.  A schematic sketch is in an earlier post in this thread, along with a couple of errata.

[dietert1]

When i look into the work that has been done to re-engineer that retrofit i am wondering who invented that circuit. I can hardly believe that they drive the HP2731 LEDs from the high voltage secondary of the 84 Hz converter. The risetime of that signal is about 20 usec in my  instrument, so there is no different timing than on the low voltage secondary. By the way: the resistors R14 and R15 are 1K as far as i have seen. 200R and 20K as in 845A_324_RevB_2-1.pdf don't make a lot of sense.

I checked that the LDRs do not generate significant photo voltage under the more intense LED illumination i proposed above. Interesting enough LDRs are best used with AC. If you let a LDR conduct DC current for some time, it becomes directional. It will generate a small photo voltage afterwards and needs some minutes to calm down again. Currently cross conduction absorbs about half of the input voltage in my instrument (test with short on integrator).

Meanwhile i studied a little bit thermal voltages on the input resistors. R111 and R112 were carbon resistors that generated about 20 uV each in a simple test. I replaced them with UPF50 that measured about 3 or 4 times lower. In the same test R114 (wire wound) generated only 1.8 uV, so it will stay. Will have to check those huge resistors R110 yet (2x 300K).

Regards, Dieter

[Kleinstein]

Possible thermal EMF from the resistors is a good point. under normal conditions there should not be much thermal gradient at the resistors though. Replacing them with other types one has to keep in mind the maximum voltage. R110 is the main series resistance to limit current from ESD. There should be no need for precision / low noise resistors, but low thermal EMF may help a little. So the cheap thin film ones should be well good enough.

The rise / fall times for the LEDs should not be a big deal, as the LDR is way slower anyway.  A more important factor / possible point for improvement could be a dead time with both LEDs off. For a speed up one could also consider to drive the LEDs initially with more current and less current later. The series capacitor already does this to some degree. Because of the cross conduction / delay I see no real advantage in a high chopper frequency. The AZ OPs use a higher chopper frequency, as there CMOS amplifiers have a low of 1/f noise and one can not include large capacitors on chip - so fast chopping makes sense there. With discrete parts one could even consider going lower to reduce the cross conduction and charge pumping effects. However the transformer would make this difficult.

I had a quick though about the input current. As far as I understand the circuit there could be some input bias if the DC amplifier (and demodulator) has an offset and thus significant ADC voltage is needed to compensate. The AC current though R115/C105 would than cause DC bias. So it may help to trim the offset of the DC amplifier.

Are the LDRs used very special (extra fast) parts ?
I have 2 old equal LDRs in glass case - would it make sense to build an LDR chopper with these ?

P.s. to answer the question myself: the common LDRs are slow and hardly useful. 20-30 ms seem to be normal and the ones I have are not much faster.  So the LDRs for the chopper are special faster types.

[Dave Wise]

Partial quote.

By the way: the resistors R14 and R15 are 1K as far as i have seen. 200R and 20K as in 845A_324_RevB_2-1.pdf don't make a lot of sense.

Dieter, after studying TiN's photographs again, I think I have to agree.  R14 and R15 are marked 10011 and 11001 respectively, with one simply facing backwards.  Not 20022 and 22002.  If those stripes are brown not red, 1K makes more sense than 110 ohms, the alternate interpretation.

In that case, my explanation of the offset trim is bunkum.  I don't know why R14 and R15 exist, unless it's to swamp out U4 and U5's small but variable on-resistance.
Fluke took the trouble to install them, along with their teflon terminals - it must have been worth it.
And if R14 == R15, then U4 and U5 must be selected so their mostly-canceled offsets result in the correct net polarity for R7 to trim back to zero.

DISCREDITED PARAGRAPH
When lit, U4 and U5 produce on the order of 100uV of
offset voltage, with source negative and drain positive.
Asymmetrical resistors R14 and R15 forming the common
point between U4 and U5 transmit the respective offsets
unequally, resulting in a net positive, which current
from U3 via R6 and R7 trims back to zero.  The trim
current is adjustable from 1.6uA to 4.4uA, developing
16 to 44uV across R114.

NEW PARAGRAPH - February 2020 - ALSO DISCREDITED - January 2021
When lit, U4 and U5 produce on the order of 100uV of
offset voltage, with source negative and drain positive.
U4 and U5 are selected so their mostly cancelling offsets
result in a net positive, which current
from U3 via R6 and R7 trims back to zero.  The trim
current is adjustable from 1.6uA to 4.4uA, developing
16 to 44uV across R114.

Oh, and one more mistake I made.  R3 is marked 10031, not 20011.  Must be 10K, since 130 ohms is absurd.  That means current into U3 is (15-2.5) or 1.25mA .
I don't know what I was thinking.

I sure wish I knew U3's part number.  I want to say LM336, but 162K for R5 sounds absurd.

Thanks,
Dave Wise

UPDATE 2021-Jan-25

When I built the H11F1 modulator for my 740B #1, I matched offset.  This turned out to be a mistake, and I rebuilt it with a new round of selection and matching.
Using a toaster oven environmental chamber, I recorded the offset voltage of nine H11F2, all the same production lot.
I recorded V at 15mA, my chosen drive current, and 20C, 30C, 40C, 50C, and 60C, then calculated TC.  I also varied the drive current and calculated drive current coefficient IC.

Offset was all over the place, some positive some negative, but all had positive (and fairly constant) TC and IC.  I found two good matches.
TC was approximately +0.5uV/C and IC +0.5uV/mA.  (Pin 6 with respect to pin 4.)  In my chosen pair, one had positive offset, one negative.  The sum when wired in series pin 4 to pin 4, i.e. cancelling TC and IC, was around 50uV.
This is more than the zero adjust range, so like Fluke I added a coarse trim.  In the 740B this is easy, just insert a resistor at the appropriate end of the control.
In the oven, my chosen pair drifts only a couple uV from 20C to 60C.

Note from TiN's picture in post #21 that U5 has a paint dot.  I believe this reflects selection and matching, and dictates install order so the net offset is positive.  Which is then cancelled by Fluke's negative-only coarse trim.

NEW PARAGRAPH - January 2021

When lit, H11F optoisolators produce significant offset voltage.
U4 and U5 are selected so their temperature coefficients cancel.
Secondary selection rejects pairs with excessive total offset.
Then the pair is installed such that the offset - U4 pin 6 minus
U5 pin 6 - is positive.  Current from U3 via R6 and R7 trims this
back to zero.  The trim current is adjustable from 1.6uA to 4.4uA,
developing 16uV to 44uV across R114.

[dietert1]

Today i tried once more to measure the thermal EMF of the large 150 KOhm input protection resistor of the Fluke 845A (2x 300K DALE parallel). This time i put a capacitor in parallel (6.8 uF foil) and shielded the whole thing. Then a HP 3456A read:
low thermal short: -0.3 uV
Fluke resistor: -1.2 uV
Fluke resistor reversed: -0.5 uV

These readings are with 100 PLC with Autozero and reproducible +/- 0,1 uV. So reversal of the resistor generated 0.7 uV and i think it has a thermal EMF of 0.35 uV. Then there seems to be some offset current into the HP3456A input that generates -0.55 uV over the resistor, gives 3.7 pA.

So it means that Fluke 845A resistor generates a fraction of a uV, which could be disturbing in the +/- 1 uV range. Thermal EMF will cause very low frequency noise. On the other hand when i see the input protection resistors of that HP 3456A (a chain of 8x 7.5 KOhm carbon resistors) and how stable they work, maybe it doesn't matter and the thermals in the Fluke 845A stem from the LDRs. Maybe the LDRs need to be inside a metal block, similar to a HP 419A.

Regards, Dieter

[Kleinstein]

Thermal EMF at the resistors only comes up when there is a temperature gradient. The main point is the 845 is likely that the resistors will likely see essentially no temperature difference. The same is true for the resistors at the input of the 3456.
Reversing the resistor would likely also change the temperature gradient - so while it is a good idea to measure both directs, one not easily combine the 3 readings. I would take the 0.35 µV only as an order of magnitude estimate. The terminals of the relatively high power DMMs tend to be quite warm and thus susceptible to produce  thermal EMF problems.

To really check resistors one should expose them to a defined temperature gradient / difference, with pure copper wire as the return path. So more like intentionally one side some 10 K hotter than the other side.

[dietert1]

To answer your questions: The measurements i reported were taken observing low thermal EMF methods, i.e. using pure copper wires and lugs. Otherwise an accuracy of +/- 0,1 uV is not possible, of course. In the Fluke 845A that resistor is not mounted on a board, so it should get packed into some thermal insulation, even better with some heat spreader around it.

Regards, Dieter

[Dave Wise]

I have updated my post #245 to point out my previous errors and add a new, more plausible description of the H11F offset trim.

[dietert1]

As far as i understood, the problem was rectified charge injection, as it also happens when using a separate LED driven gate voltage generator as in those solid state relays. Or is the selection in DC mode?
Recently i found Mosfet arrays with fine tuned gate thresholds: ALD810021. Maybe those can balance injected charge/current without further selection. Just an idea.

Regards, Dieter

[Dave Wise]

My SSR offset was charge injection, in the form of a 20mV(!) pulse each time the relay was energized.  (Turnoff produced a smaller pulse, which is the problem.)  When it was steady-state on, I couldn't detect any offset whatsoever.

Gate threshold has been loosely specified prior to the ALD810021.  But its intended use is at DC, not switching, so the gate charge may not be so matched.  If someone wants to try them, good.