HP 34401A reads a little low on AC

[bdunham7]

I have a recently acquired HP 34401A that appears to be accurate as far as I can determine on DCV and R, but reads a little low on AC.  I've read about issues with the power supply to the TRMS converter circuit, but this is so slight I'm wondering how I can test for it.

The unit looks minty fresh and clean with a bright, even display and very few marks or dings.  I disassembled it and removed the inner shields and found no burned components, no corrosion and not even any dust.  I couldn't tell it apart from brand new.  It has a CAL count of 35 (one calibration) with a CAL date of 24 AUG 1999.  The firmware is 10-05-02.

It reads about 0.2 to 0.25% low on all AC ranges, tested on three different sources at various combinations from 1VAC to 120VAC and 50Hz to 1kHz.  I don't have a source or reference accurate enough to legitimately calibrate the unit, but I do have several that are in the same range of 0.06-0.08% basic accuracy and states of calibration ranging from just back from Father Fluke to factory calibrated 10+ years ago.  The important thing is that all of those references uniformly agree to well within their specs--0.01 to 0.02%, while the HP is consistently below that by 3X the spec or so. 

I'm not terribly concerned about the actual calibration right now--I could tweak the gain constants to match up with my 8846A or actually have it calibrated--but I'm thinking there is some other single issue that is affecting the entire TRMS conversion process.  It's not yet broken enough to make it easy to troubleshoot, but I don't want to wait until something lets go. 

Any ideas?

[mawyatt]

I have a couple old 34401A's, one is an HP and the other an Agilent. I recently purchased a Keysight 34465A, and will later do a home-brew cal and use the 34465A as a "reference".

I developed a few simple calibration PCBs to help, one includes a reasonably accurate low frequency Square-Wave reference which is generated from an accurate reference voltage (5.000VDC). The duty cycle of the SW is very closed to 50%, and guaranteed by design from a CMOS FF. The FF output is buffered with a discrete CMOS inverter (low Ron PFET and NFETs) for a lower output impedance (you could just parallel a bunch of CMOS inverters for the same effect).

So with a 5.000VDC reference, the 50% duty cycle should produce 2.500VAC rms, and the 34465A confirms this, and the 34401A's agree as well. Also have some resistor references, and testing a reference with 10, 1 and 0.1VDC outputs that's based upon an LM399 (also a REF01) and another that includes a LTZ1000 & LM399. All these are just for my home use as I've retired so don't need NIST traceability.

Edit: Just did a pair of quick readings on the 34465A and 34401A, 5.00020VDC/2.499983VAC and 5.00023VDC/2.50016VAC respectively.

Best,

[mawyatt]

Here's the reference PCBs using the LM399 and the LTZ1000 & LM399.

Best,

[guenthert]

     Sorry, when I read 34401A and low AC readings, I thought of the obnoxious "feature" of that unit displaying a "satisfying" 0, when the voltage becomes less than 100uV as Joseph Geller published (I knew Geller Labs sadly closed shop, but I just learned that their web site went off-line too.  A copy of his findings is preserved at https://pearl-hifi.com/06_Lit_Archive/15_Mfrs_Publications/20_HP_Agilent/01_HP_3400B_20MHZ_True_RMS_Voltmeter/Geller%20Labs/34401A%20AC%20zero.pdf). 

    Alas, that has nothing to do with the issue you're reporting.  To me it sounds like that unit simply drifted out of spec (3 times max. spec'ed error in 20 years wouldn't concern me too much).

[nfmax]

Send it out to a competent cal lab for calibration & adjustment, then you will know. If the DC ranges are as good as they look to be, its probably a good meter, and worth spending the money on calibrating. It's all closed-case, software based calibration, but the AC ranges need some high-voltage, high-frequency signals which require specialised calibration equipment.

[bdunham7]

Send it out to a competent cal lab for calibration & adjustment, then you will know. If the DC ranges are as good as they look to be, its probably a good meter, and worth spending the money on calibrating. It's all closed-case, software based calibration, but the AC ranges need some high-voltage, high-frequency signals which require specialised calibration equipment.

The DC and ohms ranges appear well within specs.  One of the sources I have available to use is a Fluke calibrator, so I can feed it anything it needs except that 750VAC 50kHz input--that would require a newer, better calibrator.  I'm currently not needing to calibrate it--I want to know how it works, or doesn't work, as the case may be.  If it is in the beginning stages of a breakdown, then calibrating it wouldn't be all that helpful in the long run.

[bdunham7]

To me it sounds like that unit simply drifted out of spec (3 times max. spec'ed error in 20 years wouldn't concern me too much).

That sounds reasonable at first, but my experience with top-tier modern (relative term here) closed-case calibration meters is that they maintain their calibration pretty well over the years unless they are damaged or something breaks.  I have a number of similar class meters that are well within their 90-day specs, or even 24-hour specs, more than a decade after their last calibration adjustment. If there is something in the TRMS circuit that is prone to drift, I'd like to find it. 

[mawyatt]

To me it sounds like that unit simply drifted out of spec (3 times max. spec'ed error in 20 years wouldn't concern me too much).

That sounds reasonable at first, but my experience with top-tier modern (relative term here) closed-case calibration meters is that they maintain their calibration pretty well over the years unless they are damaged or something breaks.  I have a number of similar class meters that are well within their 90-day specs, or even 24-hour specs, more than a decade after their last calibration adjustment. If there is something in the TRMS circuit that is prone to drift, I'd like to find it.

One reason I got the Keysight 34465A was because how well these old 34401As have held up, one (Agilent) after 15 years since cal, the other (HP) even longer.

The KS34465A computes the AC RMS rather than with specialized circuits like the 34401A uses I believe, maybe someone that actually knows how these AC RMS values are created in both DMMs will comment, I certainly would like to know.

Think that if these DMMs are to be used for work rather than a hobby, then a full certified calibration is in order.

Best,

[RoadRunner]

Hello,

Nothing to add to this post but i have question

Hello mawyatt , In your images there is a nice looking tweezers, Is this possible to know make and model of that?

Regards

[Dr. Frank]

Hello,
I also see a problem that AC accuracy is off, but the other modes are fine.
It should be < 0.1% accurate in the ranges you described, even after many years.
Basically, the conversion is done by a TRMS chip, which has a fixed conversion factor, and the AC-attenuator resistors probably do not drift that much over time.
The DC-output from the TRMS chip (AD636 or so) is measured by the DCV part of the 34401A, and that does not drift that much either, especially as you found out, that DCV is still quite accurate.

You mentioned that you have no appropriate tools to calibrate ACV properly, so how would you know that you measure (compare) ACV properly?

As Guenthert already mentioned, there's the 'Geller-problem' with ACV on the 34401A, but also on many other DMMs with TRMS chip.
This simply means that their AC specification is valid only for signals between 5% and 120% of full scale, see footnote [4] in the specification.
If you measure consistently on 1% to 5% F.S. level, i.e. 0.5V_ac on the 10V range, then you might already encounter systematically low readings (additional -0.1%) , worsening at even lower levels, and exact Null at I think < 0.1% F.S. signals. That your reading is 0.2% LOW, but not 0.2% HIGH, already indicates that this very probably explains the effect.
Make sure that you make your measurements at 100% F.S.

Distorted AC signals and similar might also be the root cause for worse accuracy, e.g. additional 0.15% for Crest Factor between 2-3.

If you can rule these effects out, then there might have happened an overload condition on the AC input attenuator network, i.e. especially on this 1MOhm divider resistors, R301, R302, see schematic, which led to an excessive drift. Maybe an overload with > 1kV occurred, which might also have damaged input circuit, C301, C302, U301. I would try to check these components.

Anyhow, if no damage can be found, but only a drift occurred, usually a new calibration will solve the problem.

Frank

[mawyatt]

Hello,

Nothing to add to this post but i have question

Hello mawyatt , In your images there is a nice looking tweezers, Is this possible to know make and model of that?

Regards

These are from Techni-Tool, they are very old with no model number but have Anti-Acid and Non-Magnetic indicated.

Best,

[bdunham7]

You mentioned that you have no appropriate tools to calibrate ACV properly, so how would you know that you measure (compare) ACV properly?

My definition of proper tools would be currently certified NIST-traceable standards with uncertainties 5 to 10X less than the tolerances of the meter being calibrated, or whatever is called for in the HP calibration manual.  That would typically be a Fluke high-performance calibrator and the corresponding power amplifier.  I don't have anything at that level.  I also don't have anything to produce the 750V 50kHz test signal for the last step of the AC calibration sequence, so even if I had an appropriate reference meter, I wouldn't be able to do a 'proper' calibration.

I do have stable sources and several DMMs that I know are at least as accurate as the HP 34401 specs, all of which agree to well within their published tolerances.  The sources at the moment include a Fluke 5100B (repaired, not current cal), a Fluke 5200A (repaired, not current cal), a Siglent SDG2042X and for DC, a Fluke 731B and a Power Designs 5020.    Meters include an in-cal Fluke 8846A, an expired but fairly recent cal HP3455A (which also is surprisingly good) and a trio of old 8842A models that were last calibrated from 2 years to 20 years ago.  I can get 5 meters to agree on an AC signal to better than 0.02% and on DC to better than 20ppm.  The HP 34401 is also well within 20ppm of the others on DC, but consistently sits at -0.2% or so from the others on AC.

As Guenthert already mentioned, there's the 'Geller-problem' with ACV on the 34401A, but also on many other DMMs with TRMS chip.
This simply means that their AC specification is valid only for signals between 5% and 120% of full scale, see footnote [4] in the specification.

Distorted AC signals and similar might also the the root cause for worse accuracy, e.g. additional 0.15% for Crest Factor between 2-3.

I've tried this with 1V, 1.414V, 10V, 50V, 100V, sine waves from three sources, square waves, etc.  All the other meters agree and the HP34401A is very consistent, but low.  The 8842A model uses the same or similar AD converter chip, so if there were a distortion, noise or crest factor issue, I would think they would be similarly affected.  So it seems to me that the circuit is basically working well, but something is causing a scaling error somewhere. 

If you can rule these effects out, then there might have happened an overload condition on the AC input attenuator network, i.e. especially on this 1MOhm divider resistors, R301, R302, see schematic, which led to an excessive drift. Maybe an overload with > 1kV occurred, which might also have damaged input circuit, C301, C302, U301. I would try to check these components.

Anyhow, if no damage can be found, but only a drift occurred, usually a new calibration will solve the problem.

Frank

I'll check those if I can figure out exactly how that first section works or how it could fail.  It is labelled '50 kHz Flatness' and my error is the same at 50Hz and 50kHz, so I sort of ignored it.  I've attached a schematic in case anyone wants to follow along.  I also ruled out the middle gain sections because the same error occurs at all different levels.  Perhaps I should reconsider those areas.  However, I'm pretty sure the problem lies somewhere in this AC schematic.  People with more severe AC measuring problems on this model have reported leakage issues with tantalum caps and power supply issues of various sorts.  I'm reluctant to calibrate a drifting meter until I know what it is doing.  The real challenge will be tracing a signal through the circuits looking for an 0.2% error without causing that error by circuit loading. 


[/quote]

[bdunham7]

If you can rule these effects out, then there might have happened an overload condition on the AC input attenuator network, i.e. especially on this 1MOhm divider resistors, R301, R302, see schematic, which led to an excessive drift. Maybe an overload with > 1kV occurred, which might also have damaged input circuit, C301, C302, U301. I would try to check these components.

After looking at the circuit and how the ranges work (see attached) I took the following measurements:

Compared 5VAC, 500mVAC, 50mVAC 1kHz nominal output from SDG2042X, both set to 20Hz filter

Fluke 8846A: 4.9974, 0.49782, .049594
HP34401A: 4.9880, 0.49687, .049507

Error from 8846A -- 5V: -0.188%, 500mV: -0.191%, 50mV -.175%

So I have a pretty consistent -0.19% error.  The maximum error at these test levels should be about +/- 0.08%, and of course we usually see better than that. I tested the 100V range separately and got a similar result.  This seems to rule out anything in the 1st stage other than R301 and R302, as well as the second stages since the error is there regardless if one, both or none of the amplifiers is switched in.  I thought carefully about U307-A and the RMS converter section, but I just don't see anything there that would behave with such a constant error over both voltage and frequency, other than perhaps the AD637 itself.  So, I went back to R301 and R301 and measured them:

Measured with Fluke 8846A:      

R301: 500954R
R302: 501307R
R301+R302 1002263R

Now I have a conundrum.  If these resistors together were exactly 1M, the meter would have an apparent error against the 8846A of less than the allowable tolerances.  It would be off by perhaps +0.03%.  I actually calculated that at 1000359R, the two should match, excepting noise.  The trouble is, I don't know what the original resistances of these were, so I've no way of knowing if they've drifted somehow.  The resistors are Vishay PTF65500K00CT16, which is an epoxy-coated 500K 0.25W 0.25% 5ppm/C 500V mil-spec resistor that is still in production.

https://www.mouser.com/datasheet/2/427/ptf-1763051.pdf

So by the datasheet, they are still within their listed tolerances.  If they were those values originally, the closed-case calibration method would have just stored the appropriate gain constants and the meter would be accurate--the nominal value doesn't matter much within limits.  Just to make sure my numbers were right, I bodged on 2 parallel 1G resistors.  Sure enough, the readings now agree with the others well within listed uncertainties--although the frequency response became poor with a +10% error at 50kHz due to the parasitic capacitance of the 1G resistors at 0.375pF each.  I can get replacements of the same type with a 0.05% tolerance, which should bring it into spec.  I'm just wondering if anyone can think of anything else to look at in this circuit.

One other thing--the series input capacitor C301 is rated for 400VDC.  That doesn't seem appropriate.  Any thoughts?

[bdunham7]

I finally obtained some new resistors.  I had my choice between 0.1% and 5ppm/C tempco, or 0.05% and 10ppm/C.  The original were 5ppm/C, so I bought 5 of the former.  I picked 2 out of the 5 to get as close to my desired value as possible and put them in.  The AC range now reads spot on at 60Hz and very close at 1kHz, but 50kHz is out of spec.  The metal shield that sandwiches the AC part of the board is critical--removing it causes gross errors and the manual warns that if you even remove and reinstall it you may be out of cal at 50kHz.  I looked at it and realized that the press-in standoff on the bottom half was cracked and allowing the bottom half to wiggle around.  This has nothing to do with the original problem as the overall error is much lower, just barely enough to be out of spec.

I finally found the standoffs today at McMaster-Carr and although I had to firm it up with epoxy, the standoff is now tight and the shield secure.  I also have 9 spares in case anyone else has this issue. The meter is now accurate and well within spec at 10Hz, 1kHz and 50kHz on the ranges I've tested so far. 

The moral of the story is "don't calibrate a broken meter".  It's not a cheap-out, I'm just very wary of 'recalibrating' a device that typically doesn't go out of calibration without understanding what the issues are.  I'm going to call it fixed.

[guenthert]

     Did I read this right?  Instead of calibrating the DMM by adjusting the NVRAM values holding the calibration coefficients as intended, you replaced internal components which drifted over the years (as such do) out of spec with those causing the DMM to show measurements agreeing with your other (non calibrated) DMMs?  And you recommend that practice?

[bdunham7]

     Did I read this right?

No, you appear to have glossed over the details.

Instead of calibrating the DMM by adjusting the NVRAM values holding the calibration coefficients as intended, you replaced internal components which drifted over the years (as such do) out of spec with those causing the DMM to show measurements agreeing with your other (non calibrated) DMMs?  And you recommend that practice?

I didn't say 'instead' of calibration.  I'm recommending repairing broken instruments before calibration as well as understanding how they go wrong when they do.  This recommendation is based on my recent experiences in repairing various test equipment.  Are you suggesting that given a meter with 10 separate AC voltage gain constants that is way out of spec on all ranges by about the same amount, the proper course of action would be to adjust all 10 constants without any further investigation?  b/t/w, I'm sure that is actually a common result, but I can't agree that it is proper.  Can you support your assertion that these meters drift in the manner you have suggested?  I see absolutely no evidence that they do.  In this case Dr. Frank suggested that R301 and R302 could have been damaged by an overvoltage and I could not find any other component that would affect all the ranges.  It is now ready for calibration if needed or for use if not.  As far as my other instruments, I have an in-cal 6.5 digit DMM and a bunch of others that agree with it (without having been adjusted) so while I wouldn't claim good enough accuracy to calibrate another precision DMM (as I've posted above) I can be pretty sure about when another meter is off or not.

As far as recommending the practice, this is the first time I have replaced components that directly affect calibration like that.  I have repaired other broken units and had them be perfectly accurate afterwards, but those repairs were to non-calibration-critical parts.  However you want to think about the resistor replacement, surely you would agree that repairing a loose shield and making sure there are no failed or out-of-tolerance components would be a prerequisite to a proper calibration? 

I'm still curious though whether a momentary overvoltage would cause an increase in resistance.  I think I'll fry them a little bit and see how they behave. 

[bdunham7]

An experiment demonstrates that the resistors will increase in value given an overload, but it took quite a lot.  I don't have the means to provide a controlled transient, so I just used DC.  I started at 500V for 10sec and worked my way up.  The resistor in question was removed from the DMM and had an initial value of 0.50132M.  It seemed to be completely unfazed by even 1000VDC for 10 seconds, so I tried 800V for 30 seconds, then a minute, then 900V.  Well, 900V for one minute raised the resistance to 0.50145M without any noticeable overheating or discoloration.  I was able to touch it a few seconds after shutting the power off and it wasn't hot.  A second round of 900V for one minute raised it to 0.50153M and it seems to be a permanent change.  So that theory is perfectly plausible.

Now I wonder if I should replace the GDTs.  I ordered two but I didn't install them yet.  It seems that if the 'drift' was caused by an overload, it wasn't a small overload!

[guenthert]

     Did I read this right?

No, you appear to have glossed over the details.

     Apparently.  I did see however "So by the datasheet, they are still within their listed tolerances. ", which made me wonder why you decided to replace those resistors.

Instead of calibrating the DMM by adjusting the NVRAM values holding the calibration coefficients as intended, you replaced internal components which drifted over the years (as such do) out of spec with those causing the DMM to show measurements agreeing with your other (non calibrated) DMMs?  And you recommend that practice?

I didn't say 'instead' of calibration.  I'm recommending repairing broken instruments before calibration as well as understanding how they go wrong when they do. 

     Sorry for the mis-quote.  And yes, we do agree that broken instruments need to be fixed before they are calibrated.  We do seem to disagree what makes an instrument 'broken'.

This recommendation is based on my recent experiences in repairing various test equipment.  Are you suggesting that given a meter with 10 separate AC voltage gain constants that is way out of spec on all ranges by about the same amount, the proper course of action would be to adjust all 10 constants without any further investigation? 

      I'm no expert, far from it really, in either calibration or repair, so don't take my word for it, but I wouldn't consider it implausible that a 20+ year old instruments drifts out of spec and that it does on multiple ranges by a similar amount.  If it isn't obviously damaged then I'd would attempt calibration and if it can be calibrated, i.e. its component didn't drift too far out of spec, then I'd leave it at that.

[bdunham7]

     Apparently.  I did see however "So by the datasheet, they are still within their listed tolerances. ", which made me wonder why you decided to replace those resistors.

     Sorry for the mis-quote.  And yes, we do agree that broken instruments need to be fixed before they are calibrated.  We do seem to disagree what makes an instrument 'broken'.

      I'm no expert, far from it really, in either calibration or repair, so don't take my word for it, but I wouldn't consider it implausible that a 20+ year old instruments drifts out of spec and that it does on multiple ranges by a similar amount.  If it isn't obviously damaged then I'd would attempt calibration and if it can be calibrated, i.e. its component didn't drift too far out of spec, then I'd leave it at that.

It does seem a borderline case, which is why I thought it was interesting enough to post.  My previous statement on the resistors wasn't quite right, the 0.25% tolerance works out to 1.25Kohms and the worst of the two was off by a little more than that.  I replaced them because my only working theory was that they had changed from their original resistance for some reason--and Dr. Frank suggested an overload as a possible cause.  My experiment has shown that it is at least plausible.  I don't discount the possibility of instruments normally drifting in general--and many do-- but I would find it troublesome and unusual in this instrument and not all that plausible for that resistor.

I wasn't looking to calibrate the meter, but I wanted to know what had happened.  This specific type of instrument generally doesn't drift much according to legend, and this specific example looks lightly used and not abused.  If all the ranges including DC had been off, I would understand that the LM399 had drifted.  The resistor, however, is a mil-spec high stability model that is quite explicitly made not to drift.  You can look at the datasheet and do the math and you'll find that according to the specs, even if the meter were connected to 700VAC for 1000 hours straight, the maximum drift of the resistors would be 0.04%--and that is at an ambient temp of 70C, with less and less drift at lower temps.  So if it did drift, it wasn't supposed to.  Which brings up calibration.

A single calibration certificate, especially the budget no-data version, tells  you very little about how your meter is going to perform over the next year or two.  Ideally you would have years of data points showing either no drift, or acceptably low drift that is quantified.  Being out of spec by 3X (which, b/t/w is a big deal IMO, not a shrug and send it to the cal lab deal) might mean that it drifted slowly over 20 years or that it just started happening a year ago and will continue (bad) or happened due to an event like an overload or being dropped.  The danger here is that you don't know which you have.  In this case, you are correct in that it is almost certainly within adjustment range, so you could have sent it to the cal lab and gotten it back with a notation that it had been adjusted and was now within specs.  And, if the overload theory is correct here, it likely would have been good to go after that.  That just wasn't my goal.

Here is the datasheet for the resistors:

https://www.mouser.com/datasheet/2/427/ptf-1763051.pdf

You can read for yourself and see that these really are intended not to drift.  Another example is the rectangular white ceramic laser-trimmed units found inside Fluke and other high-end DMMs.  They are sometimes installed at jaunty angles and it is tempting for some to want to straighten them, but you shouldn't.  AFAIK, that type of resistor will never drift unless it gets hairline cracks in it and moisture gets in--and then it is worthless.  The bottom line for these is that if they drift or change, they are defective or damaged.

You said that we disagree on what a broken meter is vs one needing recalibration.  IMO, any device that drifts more than its siblings, or more than can be explained by a known process such as LM399 burn-in and aging, is suspect and should be evaluated before considering calibration.  I don't have enough personal experience with the 34401A meters to know how they perform over time, but perhaps someone who does could chime in.  I've seen enough other meters with decades of cal stickers stacked over the calibration switch hole to know that there are many of them out there that just don't go out of spec.

Now, I guess I'll have to keep the meter around and see how it performs.  The one other thing that could have drifted is the TRMS converter chip.  It is possible that the resistors I replaced have been that value all along, that I've fixed nothing and that the meter will continue to drift because the issue is elsewhere.  We'll see.  I'm willing to give 2:1 odds if anyone wants to bet.

[manupthehills]

I also got an  HP34401A calibrated once in its lifetime in the 1998 (35 count). Perfect conditions, in and out, to the point of looking brand new. It was probably used in a sealed rack to monitor something. The display was uneven though.

Well, replaced the electrolytic capacitors (that were still good!) and the VFD with one coming from China (was lucky to get a good one), and after one month burn in, I sent it to Keysight for calibration.

It failed several VAC measures by ~0.11-0.12% (all measures below the real value). The VDC was good, and in particular, the 10VDC (strictly related to the internal reference) was only 2ppm off!
It came back completely readjusted (cal count 66) and ready for many more years of good service (a perfectly aged Barolo wine).
Hopefully, it will drift less than its newer sibling 34401A branded Keysight made in the 2015...

[bdunham7]

I also got an  HP34401A calibrated once in its lifetime in the 1998 (35 count). Perfect conditions, in and out, to the point of looking brand new. It was probably used in a sealed rack to monitor something. The display was uneven though.

Well, replaced the electrolytic capacitors (that were still good!) and the VFD with one coming from China (was lucky to get a good one), and after one month burn in, I sent it to Keysight for calibration.

It failed several VAC measures by ~0.11-0.12% (all measures below the real value). The VDC was good, and in particular, the 10VDC (strictly related to the internal reference) was only 2ppm off!
It came back completely readjusted (cal count 66) and ready for many more years of good service (a perfectly aged Barolo wine).
Hopefully, it will drift less than its newer sibling 34401A branded Keysight made in the 2015...

Thank you for an excellent data point.  Do you happen to have the calibration information and can you share it? 

I suspect that despite the huge number of 34401A models in the wild, there isn't much data on the AC calibration.  I suppose HPAK has some, but unless you go that route and get a with-data calibration after a 20 year gap, most people won't know that their AC readings are off by 0.1-0.2%.  If the meter is calibrated annually with adjustment and without data, the user will never know about any drift and the meter will never actually be out of spec. This could be a common issue that I have mistaken for a rare event.

[manupthehills]

Sure, please see the attachment. The first two pages are "as received" measures, the following two are after the calibration.
All other measures (not included) were within specs, they were adjusted as well

[J-R]

I have a 34401A with a current calibration certificate.  Last calibration date was in 2000.  Calibration count 279.  Here is the voltage calibration data:

[Wintel]

Sure, please see the attachment. The first two pages are "as received" measures, the following two are after the calibration.
All other measures (not included) were within specs, they were adjusted as well

A very good Ref.

Can you also post the current measures or a full report?

[bdunham7]

Sure, please see the attachment. The first two pages are "as received" measures, the following two are after the calibration.
All other measures (not included) were within specs, they were adjusted as well

Thanks for that--it's an excellent data point.  I see that the unit was just barely below spec on the low-frequency AC ranges (all about .12% from nominal) and the HF flatness seems to have curved the 50kHz and 300kHz back in spec.  If you removed and reinstalled the shields during the repair work, that supposedly can cause a shift in the HF response.  So it seem likely that some single element drifted that amount on yours as well, but not as much as mine and after much more use if your display was worn out.  It would be interesting to know what it was used for.