Null voltmeter

[doktor pyta]

I'm starting a topic which might be interesting for volt nuts.
I'm going to share some thoughts on null voltmeters as I own three units.
For me there are a beautiful example how the engineers in the 60's managed to overcome the lack of high precision components.

Repairs:
In meters that use neon bulbs most common failure are the neon bulbs They get metalized inside and their flash threshold voltage gets much higher. They start to flicker. The flickering can be seen off course however You should remember about the 'dark effect' in neon bulbs (dim the lights in the room while observing).
Off course electrolytic capacitors and leaking batteries are a common issue.


Fluke 845AB- difficult access to the components because of it's mechanical construction . Batteries inside can cause problems due to leaking electrolyte. Goes down to 1uV full scale.

Fluke 845AR- easier to repair, no batteries inside. Goes down to 1uV full scale.

For both Flukes, notice two 47nF capacitors between LO- Guard and between Guard- mains earth. If You connect directly a output of an OPAMP to LO terminal it is highly probable that it will start to oscillate at high frequency. I warned You!

HP419A- very good design. Goes down to 3uV full scale. Seems to have lower noise than Fluke models. Watch out when disassembling and assembling because the screws have different lengths (about 3mm and 5mm ) and if You put the longer one in place where it shouldn't be you will have a short between LO terminal and mains earth (it will disturb floating measurements). The meter utilizes a Mercury battery as a floating voltage source, but You can use the meter without this feature as well.
If the meter is operated without batteries, two Zener diodes should be soldered in place of the batteries to create a low impedance ground path.

Keithley 155- I heard that it is a very good design. If there is any owner, please write an opinion.

Let's also consider using modern good digital multimeters with the autozero function at 100mV range. What are the pros and cons?

[HAL-42b]

With 1uV full scale you obtain about 2nV resolution.

Makes me wonder, how cheaply can we obtain  2nV resolution with modern off the shelf equipment.

[ManateeMafia]

The Keithley 155 is based on a solid-state chopper. Some parts are hard to track down. I have one functional and one non-functional Keithley. Another plus is that it can run from four 9-volt batteries. If you are in the market for one, try and find one with the external power pack. It allows the meter to run from mains. If not, an external ps will need to be attached so it is easier to replace the batteries as needed.

Edwin Pettis pointed out to me that ESI used the HP and Keithley null meters in the 242's. The 801 came mostly with the HP but it appears on the last model (801C) it was replaced with the K155. It looks like Fluke's design may have been too big to fit.

I also have the 845AR and 845AB. I recommend any model that will null after warming up. Any that are unstable most likely need new neon bulbs. A member of the volt-nuts group uploaded an LED mod for the 845 to replace the neons. It located on the ko4bb website .

[doktor pyta]

With 1uV full scale you obtain about 2nV resolution.

Nope. Usable resolution is something about 50nV due to the low frequency noise.

[TiN]

I'm new to null-meters (was under false impression that K2002 or 3458A can replace nullmeter), but already have F845AB in the mail my way.
Should be working, but we know how it goes.

Reason why I hit that evil "bid" button, is the fact that nullmeter can run from battery, floating from rest of circuit, which none of long-scale DMMs are able to do. With proper setup, I'm hoping to very low leakages.
And since I have Keithley 182-M with lowest 3mV range and battery powered front-end amp, would be interesting comparison.

[pelule]

Fluke, HP/Agilent/Keysight amd Keithley stated in artikels, a modern DMM is able to replace a null-voltmeter.
In many cases this is true - but not, if an ultra high impedance is required.
Typical modern DMM have input impedance of >10G, but null-voltmeter (when correctly nulled) impedance is "infinite".
I have two HP419A
- The latest build one is fully operating (2 years ago I complete refurbised, replaced the accumulators, cleaned and made performance verification).
- The oldest one is planned to replace the NEON-Chopper circuit with a moden chopper opamp circuitry.
I use the HP419A mainly for measurements of isolation/GOhm  resistances and for no-load measures of my  standard-cells.
BR
PeLuLe

[branadic]

There is this little circuit based on LTC1050 chopper as a DVM front-end available: http://www.tradeofic.com/Circuit/8273-NULL_DETECTOR.html

I'm sure with LTC2057 something similar would be possible.

[pelule]

Yes, that's a part of the circuit of the "mini-metrology-lab" from Conrad Hofman
http://conradhoffman.com/mini_metro_lab.html
BR
PeLuLe

[acbern]

After trying many meters for use as null meters (keithely 155, 182, 2182, 2002, 34220, 3458A) I ended up with the Keithely 155 and the 3458A. They deliver (in my setups) the most stable results. The 3458A actully is my preferred one for ths purpose. Never tested the Flukes and 419, pretty old design and I heard not really reliable and also noisy (true or not).

[uncle_bob]

Hi

You *can* use a ballistic galvanometer as a null instrument and get "infinite" leakage resistance to ground. By the same token, anything that is solely powered by 9V batteries and has only a + and - terminal to the outside world can do fine in that respect.

Past that it becomes a question of what sort of calibration you need. The spot on the wall from the galvo will tell you high / low. It's not a real good way to come up with +3.25184 nV high. The galvo is only a zero current device at null, so not a big limit as normally used.

Bob

[Squantor]

What about the keithley 148/149?
It is a nanovolt meter and uses cadmium based solder internally to keep the thermal EMF's low.

It uses an interesting scheme to achieve its performance. It uses a vibrator to convert the DC into AC and feeds that into a shielded transformer. This transforms the impedance from Kohm to Megohm and feeds that to a low noise JFET. The rest resembles the chopped amplifier scheme used in the Fluke 845 and friends.

I have the PM2434 and Fluke 845.
The Fluke is being completely revised, all the resistors and electrolytics have been replaced, neons replaced by LED's and the Zeroing circuit is now based on 1.25V bandgap references. All those carbon composition resistors have all drifted high, varying from 10% to 30% sometimes. Quite a chore but hopefully worth it.

[manganin]

What about the keithley 148/149? It is a nanovolt meter and uses cadmium based solder internally to keep the thermal EMF's low.

Fortunately there is no cadmium solder. The whole front end is made of pure copper with crimped connections.

The performance of the 148 is quite impressive, even today. Especially the stability.

I still have two of these in fine condition.

[enut11]

I'm starting a topic which might be interesting for volt nuts.
I'm going to share some thoughts on null voltmeters as I own three units.
For me there are a beautiful example how the engineers in the 60's managed to overcome the lack of high precision components.

Let's also consider using modern good digital multimeters with the autozero function at 100mV range. What are the pros and cons?

Hi
I am experimenting with a wheatstone bridge as a better way of calibrating a Hamon Divider. I am using 22K 0.2% low TC wirewound resistors and a 10v source and have tried the following methods of null detection:
1) Yokogawa 7552 5.5 digit bench meter set to lowest DC range
2) Conrad Hoffman's metrology chopper circuit feeding a 4.5 digit DMM set to 200mV
3) Conrad Hoffman's metrology chopper circuit feeding an analogue 100uA moving coil meter
4) Levell TM10 DC Microvoltmeter with a low range of 30uV.
All seem to have limitations around the null point, mainly with widely varying readings when set to the most sensitive range.
Maybe this is normal but I would appreciate feedback on your experience and possible solutions on how to achieve a more stable null detection.
I might add that the bridge was built on veroboard and is currently not in a box. It is susceptible to "hand capacitance" near the null point.
enut11

[enut11]

A bit more experimenting. I have solved the "hand capacitance" problem by rebuilding the bridge circuit.
The original bridge was wired on veroboard using the parallel copper tracks.
The new bridge was built on a board using "island" connectors and does not cause reading problems when I place my hand near the board.
I can only assume that the copper tracks on the original bridge were responsible.
enut11

[lowimpedance]

What about the keithley 148/149? It is a nanovolt meter and uses cadmium based solder internally to keep the thermal EMF's low.

Fortunately there is no cadmium solder. The whole front end is made of pure copper with crimped connections.

The performance of the 148 is quite impressive, even today. Especially the stability.

I still have two of these in fine condition.

The manual does mention Cadmium solder but this was for the DIY test cable/jig setup by the user and not for the instrument itself, and here is a pic of the front end goodness on a K 147 itself with the muMetal sheild removed for your veiwing pleasure. (salvaged 4 units that were destined for landfill a number of years back, had forgotten about them till this thread  ).
The manual also states that  no parts inside would need to be serviced/replaced except the mechanical chopper and the electrolytic caps !!.

[zlymex]

Good topic.
There is this AVM-2000 ad advising to throw away 845A, 155 and 419A
(I'm not related to this company in anyway)

[mmagin]

I got an HP 419A cheaply, and the 40 year old NiCd pack has leaked.  I'm not so happy about how HP buried them between the case and inner shield.  Besides the KOH electrolyte being corrosive to metals, it seems like certain plastics become brittle when exposed to it.  Also it seems to get sticky, yuck.

Anyway, I'm in the process of repairing it for fun, and most of the damaged stuff seems fairly straightforward (IEC connector, 115/230 volt switch, rear binding posts, dirty case pieces, maybe some trickery with the front binding posts), I'm left thinking about whether to replace the batteries (and if so, to maybe put them externally via some 3-pole non-shorting plug, I hate the original placement) or to just substitute a TLE2426 rail splitter for AC-only operation.

I would also like to replace the 1.35V mercury cell (old one was dead) to restore it to full function.  Any opinion on using a BR-2/3A lithium and a TPS709135DBVR (fixed 1.35V regulator) with appropriate MLCC caps (lower leakage than tantalum)?  I know I could use an alkaline cell, but I'd prefer something with really long life (hence the BR lithium) since it's buried in there.

[ManateeMafia]

The Fluke 540B also used the mercury cell. Fluke eventually created a mod to do away with the mercury cell and replaced it with a lithium and a regulator pcb.

Maybe this will be useful? You will need to tie a switch in to prevent the battery from discharging while it is not in use.

http://www.ko4bb.com/manuals/97.76.44.130/Fluke%20540B%20Mercury%20Battery%20Rep.pdf

[mmagin]

The Fluke 540B also used the mercury cell. Fluke eventually created a mod to do away with the mercury cell and replaced it with a lithium and a regulator pcb.

Maybe this will be useful? You will need to tie a switch in to prevent the battery from discharging while it is not in use.

http://www.ko4bb.com/manuals/97.76.44.130/Fluke%20540B%20Mercury%20Battery%20Rep.pdf

Thanks.
Yeah, the main reason for my approach is that the battery is 1.2 Ah and the quiescent current is around 1.2 uA at room temperature, meaning I could get away without a switch   It's only about 2% stability over temperature, but I'm not sure thats much worse than the garden variety mercury cell in practice.

[amspire]

I just had a look at your meter and I saw the battery is connected between polarity reversing switches, so you cannot easily separate the battery from the regulator unless you add an extra switch.

Sounds like you plan is a good one.

The battery will go flat in a million hours so you need to add a sticker to remind someone to change the battery in June in the year 2130.

[mmagin]

I just had a look at your meter and I saw the battery is connected between polarity reversing switches, so you cannot easily separate the battery from the regulator unless you add an extra switch.

Sounds like you plan is a good one.

The battery will go flat in a million hours so you need to add a sticker to remind someone to change the battery in June in the year 2130.

What got me thinking Panasonic BR series lithium is that the BR-series battery in my HP 3478A somehow lasted 30 years powering a static RAM.  I'll be content if I get 10 years

[sfox7076]

Do any of you know anything about the Calibration Standards Corp. DC-100A?  Cannot seem to find much info on it. There is one in this listing ( which has ended). http://m.ebay.com/itm/262522718669?_trksid=p2057872.m2749.l2649&_mwBanner=1

[Vgkid]

Looks like a differential voltmeter.

[sfox7076]

You are correct.  Sorry to sidetrack the thread. 

[VintageNut]

With regard to the OP, I use a KE2000 as a null meter for measuring an Eppley Model 100 voltage cell.

I have had the Eppley cell for about a year now. It required a week for the cell to stabilize. The cell degrades around 2 microvolts per month.

The rig I use to measure the Eppley cell is a Fluke 731B, a General Resistance 7-digit Kelvin Varley Divider and the KE2000.

I divide the 10V DC down to match the Eppley Model 100 cell. I measure the diff with the KE2000. I can see individual PPM changes when I adjust the KVD.

The first stable measurement of the Eppley cell was 1.018757 on 7/26/2015. The latest measurement of the Eppley cell is 1.018731V on 6/11/2016. At the apparent degradation rate, the Eppley cell should not fall below 1.018V for another 30 years. At that point, I will probably not care about the Eppley cell.