HP 3456A-10V Range Stability

[gilbenl]

Preface-I think that I have fallen into one of those traps, but...

I recently picked up a late generation 3456A and I'm finally getting around to tuning up and calibrating if need be. I'll be using my EDC 501J to do the performance checks up to the 10V range, but hooked up my crappy 10V standard just to see how it fared. As the video below will reveal 10V on the 10V range takes a dip every few readings. However, 1V on the 10V range is fairly solid. Non-ideal conditions, 30 minute warm up, 3 or 4 pilsners in, etc, etc. See the video.

Is this a PEBCAK or is the 10V range buggered? Noise from the 10V range switching FET, perhaps?

https://youtu.be/CMuM-qa5iEs

[Vgkid]

That is rather interesting, have you done any work to your 3456? Since you will need to troubleshoot it, just go ahead and replace the psu caps. I may edit this post, not at pc.
EDIT
Alright go to pg 315, 8B8 which is used to troubleshoot the inguard portions, which will lead you to 8B20. All of this is in the 3456 service manual, troubleshooting section. Guessing a range fet.

[gilbenl]

That is rather interesting, have you done any work to your 3456? Since you will need to troubleshoot it, just go ahead and replace the psu caps. I may edit this post, not at pc.

Aside from opening the cover to inspect/take pictures upon arrival, the 3456 is untouched. I don't think this is a power supply issue, but we'll see. I'm hoping I'm making a bonehead error somewhere.

[Vgkid]

Since the updated portion of my message did not update:
"
Alright go to pg 315, 8B8-8B10 which is used to troubleshoot the inguard portions, which will lead you to 8B24(Noise/Innaccurate readings,pg319). All of this is in the 3456 service manual, troubleshooting section. Start with 8B24, as it ddeals with jumpy/noise readings with the digital, then proceed to 8B25 for the analog.Check U402(pg8B6//pg320)""
The reason for replacing or atleast checking the PSU rails/ripple is the caps are known to give varying errors when they go bad(failling self tests), so do it out of habit.

[gilbenl]

Ah. Will do. Still odd how it only occurs on the 10V range with 10V applied and not 1V. Seems section C lumps 100mV, 1V and 10V range failures, and 100V and 1000V separately.

[MadTux]

First, check all PSU rails for ripple with oscilloscope. That's the far most common failure in 3456A.

If that's fine, go ahead, grab the DCV input schematics and work your way trough the FET switches (apply 10V to input and follow that voltage until it reaches the ADC input pin, gets attenuated 10x IIRC). I had one bad 3456A in which the big round reed relay had failed (reed relay was high resistance when on), yours is looking somewhat similar. So check relays first, before continuing to FET switches.

PS. looks like Auto-Zero has something to do with it, disable that and see if it makes any difference. Also run the build in test function and see which error it produces. Then enter that failed test number, press store and key 5 => store and see what reading that failed test gives you.

[gilbenl]

First, check all PSU rails for ripple with oscilloscope. That's the far most common failure in 3456A.

If that's fine, go ahead, grab the DCV input schematics and work your way trough the FET switches (apply 10V to input and follow that voltage until it reaches the ADC input pin, gets attenuated 10x IIRC). I had one bad 3456A in which the big round reed relay had failed (reed relay was high resistance when on), yours is looking somewhat similar. So check relays first, before continuing to FET switches.

PS. looks like Auto-Zero has something to do with it, disable that and see if it makes any difference. Also run the build in test function and see which error it produces. Then enter that failed test number, press store and key 5 => store and see what reading that failed test gives you.

Will start trudging through the input section today. No errors on self test, but the HP service manual seems like its going to spoon feed me compared to the ambiguity of the Datrons. Oh well, what's an ebay score if you don't have to fix it?

Will post back some results when available. Thanks

[gilbenl]

+33V rail reads 26.3V with both analog board in and out, and waveform is virtually AC, so thats obviously a (the) problem. I'm still a bit perplexed as to why this isn't a much more global failure. Maybe someone could enlighten me?

The +33V rail supplies U303 providing the 30V regulated supply for the input amplifier (schematics attached). If I understand correctly, U307 is setup a non-inverting opamp with a series of voltage dividers on the output (U200) providing negative feedback.

As all ranges pass through the opamp, could someone summarize the consequence of having this voltage issue on the attached circuit and why it only seems to appear on the 10V range?

[gilbenl]

I'm dumb. Scratch my question, I think I've figured it out. Spent too much time going "wtf" and not enough time thinking about how an opamp/comparator works Please fill in any non-sensical BS with correctness.

The low voltage is not the primary issue, but instead the tremendous p-p (virtually AC) variation of V+. At low input voltages (ie 1V), there's ample voltage throughout the V+ curve for the opamp to work as it should and output whatever gain level (range) is selected. However, when you throw 10V on the input, the output of the opamp is saturated at all V+ values less than what is needed for the output to match the input gain.

Recall that 10V was stable on the 100V range. If my theory is correct, then increasing the input voltage on the 100V range to lets say, 20 or 30V, should elicit the same behavior exhibited by the 10V on the 10V range. OOH a chance to use the 6920B!

So this makes sense in my head, but then I went and played with buttons and confused myself. At 10V, I flipped the polarity of the EDC 501J and lo, -10.0000V--steady as could be. So why on earth would it work for a negative input, but not a positive?

[Kleinstein]

The poor positive supply to the inut amplifier will limit the working range of the ampl to something like -12 V (as original) to + 5 V (depending how bad the supply). So a negative input will be no problem. It's just the large positive voltages where early "clipping" starts.

So chances are good that it will work when the +30 V are OK again.

[gilbenl]

Pulled out the PS tonight to begin repair. Happened to have A10C8 (100uF 50V) on hand and replaced it despite neither the sides nor top bulging. Cut the leads and found a big round bottom. Good on woman, bad on caps. Never would've seen it unless I had removed it. Sneaky bugger. +33V rail now measures +43 (spec range +33 - +46)Very pleased to have my 3456A back in business. Next up will be a calibration on the 1 and 10V ranges. Happy to post details if anyone is interested.

Thanks for the help folks

[Vgkid]

Good job on the fix.