HP 3400a, LDR's (photo cell) replacement

[r6502]

Hello,

a few months I've bought a 3400a, that had sometimes the problem, that the indicator needle sows a strange behaviour, but only in the 3 / 1 mV range. Meanwhile, the Instrument show this effect in all ranges, but only after about 30 to 45 minutes after power was turned on.

I did some research in the internal parts, and found out, that the LDR's (photo cells) from the optical chopper amplifier where bad. As soon, as they have been warmed up, this behaviour becomes more visible.
I have locked in the net, but the LDR's you can get today, are much to slow for the application. So I thought about the problem and came to the conclusion,  to replace the LDR's with electronic switches.

Yesterday, I've drawn a schematic, that shows how I'd like to do the replacement of the LDR's from opto chopper assembly.  Please have a look to the schematic  and the original schematic of ths amplifier, that I also uploaded.

In general, I'd like replace the LDR's, but if no more available - what should I do ...

I hope, I can assemble the parts on a bare board tomorrow. Then I will send additional information of how it works.

Kind regards Guido
====
Edit: with  LDR I mean  the photo cell in the copper amplifier

[garrettm]

What do you mean by "LDR"?

There's really only two parts that could go bad in the optical chopper circuit: the photocells or the NE2 neon lamps.

If the cadmium sulfide photocells are bad, you should be able to get replacements. Also, the neon lamps can be subsituted with LEDs which are significantly more reliable.

The newer units eventually used an Intersil chopper stabilized OPA and got rid of the nuvistor input buffer.

[r6502]

Hello,

With LDR I mean the photo cells inside the chopper amplifier.

LDR -> Light Depending Restor

In my device the Neon bulbs are fine, but not the LDR's / photocells. The nuvistor in the first stage is fine as well, and I got some spare's as well.

It would be interesting, if somebody else has made an update like I planed.

Guido

[garrettm]

Can you upload some photos of your photocells? I don't think the chopper circuit works at much more than 100Hz. So really any photocell with the correct resistance should work okay.

I attached some photos of an HP photocell I have from a junked piece of gear along with an RCA unit for comparison.

If worst comes to worst, and mine matches your unit's photocells, I could sell you mine for next to nothing. I was going to make a solar tracker with them for fun, but if someone could actually use them to repair some gear, then that's probably a better use.

[garrettm]

Or better yet, use Fairchild H11F photo FETs as replacements for both the photocells and neon lamps.

[factory]

There are two different versions of the chopper block used in the 3400A, the early ones use a larger assembly, then a smaller assembly for the later 1960's ones, the LDR's are two different sizes to suit.
I've had a dead neon in one & another where someone had bodged the wires back on the LDR's (that one was intermittent), a parts donor fixed both of these.

Also check the PSU outputs are good if you haven't already and if it uses the pin connectors for the wiring, check those aren't loose.

David

[r6502]

Hey  garrettm,

thank's for the reply

I have made two photos, sowing one of the HP photocells:
" alt="" class="bbc_img" />
The front diameter of the photocell is 8.2mm.

I also made tests, on the sensivity of the Photocells:
1st the original HP photocell:
" alt="" class="bbc_img" />
2nd a CDS photocell:
" alt="" class="bbc_img" />
channel 1 signal that was used to power the LED, illuminating the photocell
channel 2 HP or CDS photocell response. This is the test schematic:
" alt="" class="bbc_img" />

The HP photocell is compared to all others the most sensitive photocell. Please have a look on the on the y-scale of the ch2 in both photocells.

When you have a  look at data sheets of photo cells, they specify rise and fall times in the range of 30ms. The chopper of the 3400a operates with about 120hz what  results in about 4 ms pulse time. so the photocells in the HP react really fast. This was the Idea, to use analogue switches instead of new photo cells.

The Idea with the photo FET's is also good, and I will test this as well.

Guido

[r6502]

Hello all,

Here is just a photo of the optical chopper assembly with one of the lid and one  photocell removed:
" alt="" class="bbc_img" />

Guido

[garrettm]

Thanks for the photos Guido. Those cells are significantly smaller than mine! I pulled mine out of a very old HP 419A null voltmeter.

Looking at the configuration of the optical chopper assembly, the H11Fs would be a breeze to swap in. Just place the diodes in series where a neon illuminated two cells at the same time. Then figure out the min and max resistance of the photocells (under dark / illuminated conditions). The minimum photocell resistance then sets the maximum drive current for the diodes of the opto-FET and the maximum photocell resistance can be had with a parallel resistor across the opto-FET (the H11F is about 300Mohms in its off-state).

It might be possible to improve the noise of the chopper by playing with the on/off resistance values.

[r6502]

Hello all,

Meanwhile, I have tested a bit with analog swiches and with one opto MOSFET relay, I had in my stock.

The analog swich I used, was the AD7510
The Opto photo relay  I Used was a PVG612s

I just took the the screen shots from my scope AD7510:
" alt="" class="bbc_img" />

I just took the the screen shots from my scope PVG612S:
" alt="" class="bbc_img" />

The wafeform with the analog switches looks much better. The PVG612s shows large spikes whent swiches are turned on / off.

I'm going to order the H11F as well, to see how they will perform.

I keep you informed ...

have a nice time

Guido

[Kleinstein]

The chopper part looks similar to the HP419 Null meter or the Fluke 845 Nullmeter.
The H11F.. Photofets are used in the newer Fluke versions. However they do have some offset and may need mode adjustment. There may also be more bias current, as they do show charge injection.

The LDR are usually special types that are relatively fast. So normal cheap ones for normal light detection usually don't work, as they are slow turning all the way off.

Quite often it is the neons or the excitation circuit that go bad, not so much the LDRs. So one could replace the neons with new ones (not so easy) or change to LEDs.

The demodulation part is less critical. Here it would be no big issue using CMOS switches like 4066, 4053 or the like. So to keep the input the same, one only needs 2 working LDRs.

[r6502]

Hello All,

meanwhile I did some progress in repair / replacement of the opto chopper assemby of the 3400a.

I assembled the schematic i posted in the first post on a proto board. I removed the opto chopper assembly,  and placed the proto board there and connected the analogue switches to the cosponsoring pads  of the A5 photochopper board.

see 1st attachment.

To operate the new board I needed a positive voltage. Therefore I modified the power supply board. I removed C711, R701 and R702. In the open holes from R701 / R702 connecting to the contacts 17 and 19 i placed a cap 100µF 50V. The transformer output for the 17,5V was used to generate the positive voltage by adding 2 rectifier side by side to CR711 / CR712. the cathodes have been soldered together  and with help an isolated copper wire connected to the positive terminal of the new cap, see schematic / photo.


With this modification the instrument is working very well now without a readjustment. I checked the lower frequency's against my HP34401a and the higher frequency's against the RMS value of my Rigol scope. All measurements are within the specification of the 3400a.

The next step is, to design a small PCB holding the components and that fits perfect to the position where the opto chopper assembly was positioned on the A5 board. When this is ready, I keep you unformed and will upload the design files.

see last 2 attachments.

Have a nice Time ... Guido
Edit: I updated incorrect photos and corrected spelling ...

[r6502]

Hello all,

I'm currently exchanging all electrolytic capacitors.

So I have one more question regarding the 3400a, especial the output state of the video amplifier.

The output of the video amplifier is connected with help of C413 / C415 to the thermocouple input. I marked the caps green in the attached section of the schematic. What I'm wondering about is the use of electrolytic capacitors here. The 3400a is specified to work up to to 10MHz and in that range the electrolytic's are really poor.

Do you think, it makes sense, to use ceramic caps here rather than electrolytic caps?

Thank's for your Answers.

Guido

[boblevy4321@sbcglobal.net]

Hello All
I just did a retrofit of the Photo chopper in my hp3400.
I used 4 photo fet isolators H11F1 type. The trick is to use a less than 50% square wave to drive the photo fets. My meter uses a 90 -100 hz chop frequency.
I set my oscillator to 95hz and the led on time to 4.8msec .2 msec off time.
I did this because it was fast, cheap (under $10 USA money), and simple.
Because this modification is several magnitudes better than the original , I had to change the 1/10 scale offset voltage to compensate .
I lifted the ground end of R8 and and added a 5500 ohm resistor and connected the other end to the - 17.5 supply.
when you replace the capacitors ,use high quality low leakage  caps in the signal path.

this modification should work in many of the H.P. photo chopper Instruments
hope this helps someone

[Kleinstein]

AFAIK the HP3400 uses the photo-choppers only internally for the thermocouple converters. So they don't have to be very high impedance. If needed one could likely just replace the whole Chopper amplifier with a modern AZ OP. However this would be quite far from the original.

[garrettm]

AFAIK the HP3400 uses the photo-choppers only internally for the thermocouple converters. So they don't have to be very high impedance. If needed one could likely just replace the whole Chopper amplifier with a modern AZ OP. However this would be quite far from the original.

Depends on the SN of the instrument. In late model units from the mid-80s HP switched to the Intersil ICL7650 chopper (non "S" model, as it wasn't available at the time).

I've owned most all of the revisions except units with the old HP logo. It's pretty neat to see the changes made to the instrument over time as new technology became available and old obsoleted. I finally ended up with my coveted 3400B model, but found it to be slower and less accurate than my 3400A that was a few years older than it. Sadly, I sold the 3400A while the 3400B was being shipped out and now have to live with my mistake.

[r6502]

Hello All,

it's a while ago, that I did the last work on my 3400a. I have moved,created a new lab bench, and now its time to continue with this unit.

As garrettm wrote in his last post, HP changes to an ICL7650 in later models, and Kleinstein suggested for a modern AZ OP, this is my idea, to replace the whole board. My 1st Idea is, to use a LT2057 for this job, and design a more modern version of the HP Board with the ICL7650.

I added the schematic of the A6 board with the ICL7650. Interesting for me, that HP generated the positive supply for the ICL7650 out of the 77 / 75 V. Could this newer bord  be a direct replacement for the older A6 boards?

Guido

[factory]

I've got several 3400A of different ages and when diagnosing an older unit, I did use the newer A6 board to confirm the rest of the unit was working.
The newer A2 board has some parts removed & values changed, some parts are not required with the IC based A6 board & some change to update the PSU from Ge to Si transistors, the manual notes that the rear panel in some older units prevents the newer A2 from fitting.

David

[Qmavam]

From the HP IO Group.
"A fellow going by the handle br4av01 posted on replacing the photochopper with a 555 and a DG403 quad analog switch. The switch is pretty critical; you need something with low charge injection. Unfortunately the full details seem to have been lost from the HP-Agilent group file and photo archives when they moved from Yahoo to groups.io :-("

                       Mikek

[r6502]

Hello all,

meanwhile I've changed all caps in the unit and had to change a transistor in the power supply, and the readings are not so stable, as expected. It depends on the signal and ...

I decided to put in a new A6 board, with more modern stuff.  The last version of this Meter had a integrated chopper ICL7650 instead of the photo-chopper, that has very little drift, but it is an old design and it is getting obsolete.

I'd like to go to more modern components.

1st thought was to use a LTC2057, followed by a LT1012 for driving the meter.

2nd thought was, using a double OP with low drift and Vos so the LT1013 as replacement for both OP's. I think, the specifications of the LT1013 are good enough, to be used in both stages - what do you think?

I attached the circuit diagram of the HP 3400a with ICL7650 as chopper amplifier.

Guido

[TheDefpom]

I designed a board to replace this, it is on PCBway here: https://www.pcbway.com/project/shareproject/HP_3400A_A6_Chopper_Board_Replacement_06465eb7.html

I did videos on the repair of a 3400A from 1965, I installed the board I designed and it worked fine: https://www.youtube.com/playlist?list=PL2z3V9RkHQE12roTSKiF60icefxviRo72

[Kleinstein]

Using a LT1013 for both stages could work, though there may be a little longer term drift. I don't see a need for super low bias for the 2nd OP-amp and the first one does not look like it needs to work with a high voltage. So a complete new design could use a cheap 5 V AZ op-amp (e.g. MCP6V66) and the 2nd OP-amp should be relatively non critical, but could be the same as the 1st. Low power parts could be a good idea. Chances are the LT1012 is choosen for low power precision and not for super low bias.

[factory]

I've got several 3400As in my collection of vintage TE, here are couple of the different boards from mine, the only one I haven't got working is the earliest version, seems very hard to find one that hasn't had one of the two newer replacements at some point it's life. Both types of optical chopper are used in some other HP voltmeters & DVMs from the 1960s, with some slight differences.

Oldest version with large optical chopper assembly, 03400-66504 OR.



Two versions of the smaller optical chopper assembly, 03400-66508 Rev B & Rev D.





The newer IC based version, 03400-66152 Rev C.
Microchip seem to still make the 14 pin TC7650CPD (ICL7650 1826-0968), the other is 1826-0310 (might be MC3476G according to a cross ref found on the web, not 100% sure this is correct as it doesn't appear in the original HP cross ref charts).



Attached below is 03400-66512 Rev D, found on the web.

David

[r6502]

Hello All,

Thank's for your replay.

Currently, I will go with LTCD2057 / LT1012 solution. Please have a look on the dimension of the board, that I took from my original sample and on the schematic.

For the schematic, I have to say, that I generate a positive supply from the windings that are used to generate the -17.5V rail. therefore I updated my power supply board and it is connected to pin 3, as the voltage for the neon bulbs is no longer used. I will use this rail, generating approx. 35V DC unregulated, to generate the positive rail for my OP-Amps.  I also used film caps instead of the electrolytic caps HP used in the version with the ICL7650 for the 3rd order low pass filter. I use the same  C in different quantity's, as 0.47µF is easier to get than 2.2µ and there is plenty of space on the PCB.

I mainly replaced the ICL7650 by the LTC2057 (in the HP schematic  U601) and the U602 with LT1012, that will do the job not to bad I think.

When I'm finished with the Layout, I will publish it here as well. It is my 1st KiCad Project.

Guido

[r6502]

I designed a board to replace this, it is on PCBway here: https://www.pcbway.com/project/shareproject/HP_3400A_A6_Chopper_Board_Replacement_06465eb7.html

I did videos on the repair of a 3400A from 1965, I installed the board I designed and it worked fine: https://www.youtube.com/playlist?list=PL2z3V9RkHQE12roTSKiF60icefxviRo72

I did not saw your video before I did my work, but very interesting and close to the original solution from HP for the last devices with Nuvistor and ICL7650.

I got my HP 3400a very cheap - I's difficult to get it for a price of less then 200€ here in Germany nowadays, but I'm always looking to get more devices of this type

Guido