Data Precision 8200 calibrator - repair and other fun stuff

[Conrad Hoffman]

FWIW, other than the intermittent switches, mine's been as reliable as a cast iron boat anchor. Long ago it needed a single 1k precision resistor, but other than that, nothing.

[Rax]

FWIW, other than the intermittent switches, mine's been as reliable as a cast iron boat anchor. Long ago it needed a single 1k precision resistor, but other than that, nothing.

This far, within the limits of the instrument, though pending further checks and empirical data around some suspicious datapoints, I feel this has been doing really well.

But thank you for confirming your good track with it.

[Rax]

One thing I'm seeing as I've adjusted and now I'm observing it relatively short term is that the 100V range seems to need quite a bit of time to settle down. Assuming I'd shoot for 100.0009V with my 100V field adjustment (which is my absolute characterization point per my 8502A long form cal), it'd start at about 99.9999V or so and it'd need maybe 5-10 minutes to get to the 100.0009V.

I'm considering adjusting it a little high to shoot for a reasonable amount of time before I would expect it to be accurate (I'd ideally want it to be pretty quickly ready after the 100V range turn on, in a day-to-day switching between ranges situation, which doesn't easily accommodate 10 minutes for switching the 100V range).

Is this something others have seen, or is it an issue in mine indicating a need to repair?

[Le_Bassiste]

might be worth checking as to whether extra settling time for ranges 100V and above are partly caused (or even cancelled) by the DMM's input divider.

[Kleinstein]

My suspicion is with a thermal effect. 10 minutes souds like a reasonable time frame for this.
The feedback divider in teh calibrator with 10K and 90 K is quite low value and thus quite some heating expected.
Compared to this the divider in the DMM is likely less of an issue as it is likely made also for higher voltage (e.g. 1000 V range).

For a test one could check what happens at a low voltage like 70 V: this would be about half the effect of heating, but a comparable fraction of the voltage for an effect due to amplfier settling or dielectric absorbtion.
The heating effect in nonlinear and not just a settling problem.

[Rax]

My suspicion is with a thermal effect. 10 minutes souds like a reasonable time frame for this.
The feedback divider in teh calibrator with 10K and 90 K is quite low value and thus quite some heating expected.
Compared to this the divider in the DMM is likely less of an issue as it is likely made also for higher voltage (e.g. 1000 V range).

For a test one could check what happens at a low voltage like 70 V: this would be about half the effect of heating, but a comparable fraction of the voltage for an effect due to amplfier settling or dielectric absorbtion.
The heating effect in nonlinear and not just a settling problem.

This makes sense.

[branadic]

I have upgraded AN3200 with an over 8000 h pre-aged ADR1399. I had to adjust R7 as well as R8+R9 from original 1.14kΩ to 1.7kΩ to have enough of margin left for adjustment on the 20Ω pot. Attached a 1h measurement taken with S7081 in 7.5-digit mode.

Edit: Repeated the measurement with S7081 in 8.5-digit mode and eventually verified the measurement with R6581D and indeed noise improved.

-branadic-

[Rax]

I have upgraded AN3200 with an over 8000 h pre-aged ADR1399. I had to adjust R7+R8 from original 1.14kΩ  to 1.7kΩ to have enough of margin left for adjustment on the 20Ω pot.
-branadic-

I'm not sure I follow - R7=3.01k, R8=5.3896k - how exactly did you have to replace these? Specifically, what were their replacement values?
Thank you for sharing these results and work.

[Rax]

-branadic-

branadic - this is completely unrelated, but I see all these cool graphs and charts on EEVBlog - what's the high level approach here? Automated readings over GPIB, logged some way on the computer and plotted through another something? I candidly am at this clueless level with this kind of data visualization, but I'd love to explore it.

[branadic]

It's actually a low level approach, some Raspberry Pi and GPIB magic in the background aquiring and storing data on a local drive, I still work completely script based, no fancy packages and libraries involved and the rest is simply LibreOffice visualization if I'm lazy re GNU Octave for the more interesting data, that I analyze each evening on the couch

-branadic-

[Rax]

I have upgraded AN3200 with an over 8000 h pre-aged ADR1399. I had to adjust R7+R8 from original 1.14kΩ  to 1.7kΩ to have enough of margin left for adjustment on the 20Ω pot.
-branadic-

I'm still interested in details on the resistors values revisions.

Have you seen any ill effects with doing this on the linearity of the instrument? Put differently, have you had to tweak the octade adjustments (within the "metrology calibration" procedure - R33, R30, etc.)? If not, this seems to be a pretty effective and relatively "low impact" upgrade to this unit.

R12 is part of the field calibration procedure, so I don't think that has bearing on the linearity of the instrument, and is obviously an expected adjustment to do given the change in the reference (in addition to the other resistors changes you mention).

[Kleinstein]

Changing the reference and accordingly adjust R7,R8 (or R9) and R12 should not have  an effect on the linearity. The change of R8,R9,R12 is to bring the ref. voltage seem by the divider back to the old value. So the octal divider chain would not see if the reference voltage changes (e.g. 7 to 6.9 V). The main change would be a slight higher current to the reference.

It is still a question if this would really be an upgrade worth while: the ADR1399 is lower noise, but the old LM299/LM399 is usually well aged, depending on how long the unit was powered on.
The lower reference noise would mainly make sense together with a change in the OP-amps (Z6,Z14,Z19) for lower noise.

[branadic]

R7 = 3.01 kΩ was replaced by 1 kΩ so roughly the same 3 mA through the zener as before.
R8 and R9 were replaced to the value given above to get back the adjustment range of R12. The noise went down significantly, even though Z6 wasn't replaced yet. I have an OP177 sitting on the shelf, but am not tempted to install it at the moment, also due to missing TO to DIP adapter and I don't like a dirty bodge to fit it in.
And as stated too, the ADR1399 saw a burn-in for over 8000 h before.

-branadic-

[Rax]

After having two units here to play with and assessing different things, I think it's very important for any of these to have its offset adjusted (4.3.3 in the manual).

It's technically part of the "metrology standard adjustment," but I don't think it affects anything on the linearity etc. of the unit. I may be wrong, but I think they should have included this in the "field adjustment." My standard tackle on these is currently the offset and the field adjustments. Once these steps are done, I'm finding this very well dialed in.

I've found the offset to be very high on these when they get to me - hundreds of uV - due to aging, lack of maintenance, etc. And this residual can throw all sorts of mayhem in the measurements, adjustments, etc.

[branadic]

I was asked to share some noise measurements with replaced reference, outputting 10 V.
Notice, the opamps are still original and have not been replaced. Please replace mV by µV and you end up with 3 - 4 µVpp.

-branadic-

[Rax]

I was asked to share some noise measurements with replaced reference, outputting 10 V.
Notice, the opamps are still original and have not been replaced. Please replace mV by µV and you end up with 3 - 4 µVpp.

-branadic-

branadic,
I may have missed it, but did you intend these to be compared with the "before" measurements? If yes, can you please point me to where they are available?
Than you.

[branadic]

It was way larger before, more in the 10 µVpp range, but I don't have an image of a measurement at hand.

-branadic-

[Rax]

It's been a while, y'all.

I have a second 8200, and got around to adjustments this weekend. It's a much cleaner unit and it seems to be working well overall. One issue I'm finding, though, is that I'm unable to get it to 70V when adjusting it per 4.2.3-9. The adjusting effect is very small, and I'm pretty sure I'll run out of range before I get to within 100uV of value. I'm typically finding an adjustment like this to be very close to the original pot position - more like a trim than a finger-spinning workout. This isn't acting like that, it barely moves in output value with vigorous spins of the pot adjustment.

I posted on this on the Repair group: https://www.eevblog.com/forum/repair/data-precision-8200-100v-range-out-of-pot-range-during-adjustment/.

Thank you for all input!

[branadic]

he lower reference noise would mainly make sense together with a change in the OP-amps (Z6,Z14,Z19) for lower noise.

As far as I understand you suggest OP177 for Z6, Z14 and Z19 and to remove C19 at Z6, is that right?

-branadic-

[Kleinstein]

Yes these 3 OP-amps will contribute to the noise. The original AD517 was a good OP-amp at it't time, but is still pretty noisey (2 µV_pp for 0.1 - 10 Hz). This may be acceptable for a LM399 or similar reference with some 5 µV_pp on its own, but an upgrade would make sense with an ADR1399 reference. With a change in the reference one would need to readjustst the voltage anyway and the different offsets should be no issue.
C19 is odd in the circuit, it reduces the phase reserve and not all OP-amps may like it - here it may be worth checking at least to check if it works better without or possibly oscillates with C19.

An OP177 would be a candidate that is also awailable in DIP.

[branadic]

I today received some DIP-TO99-adapters and desoldered the components on the board (Z6, Z14, Z19, C19).
Z6 was an AD517, while Z14 was an opamp marked AD40932, same as Z19, except it had a plastic cap/cover and was potted inside with some sort of silicone. When I removed the silicone I found that the leads on all pins were extended with 10R (brown, black, black, gold) carbon composite resistors, probably to reduce heat flow?, which I found pretty interesting. The three OP177 are now in place, we will see how that turns out.

-branadic-

[branadic]

After some zero point adjustments, I still use the original 50k at the offset trim pins of the opamps, I took some noise measurement data on the 10 V ouput. It seems like there is an improvement. 

rms = 388 nV
std = 387 nV
pp = 2.7 µV

rms = 500 nV
std = 500 nV
pp = 3.2 µV

rms = 401 nV
std = 401 nV
pp = 2.8 µV

rms = 382 nV
std = 382 nV
pp = 2.9 µV

rms = 399 nV
std = 399 nV
pp = 2.5 µV

-branadic-