Repair worklog : Old rusty HP 3458A

[TiN]

If KS release 3458B (from Boot, get it?) which look like this (but black, I no doubt), you know where you saw it first  .

I have rack handles for my 3458s too, but removed them to allow easier connection on posts.

[kj7e]

Is there a boot on the rear as well?

[TiN]

Yes, why would I stop half way.  Rear boot fits perfectly there, no need to trim/remove feet like with front one.

[TiN]

Post #3458

Time fly, I was just dreaming about getting one 3458A not very long ago, and now have two, which need attention and minor service after nearly 2 years of 24/7 operation. Nothing serious, just clean the dust off and replace binding posts block. My newer box had posts useable but ugly, and older was fine just old. So I bought brand new terminal block (can't afford it if you asking) so I swapped it on older unit, while original front go to second unit rear and rear on that goes to front. Whole process took few evenings and recorded on short timelapse video. Enjoy

[Muxr]

Nice work TiN! My 3458A is in need of a new front panel and the binding posts as well. I don't know if I missed it, but did you manage to order them from Keysight directly?

[HighVoltage]

Looking great, TiN
Do you still have the original VFD's installed in both instruments?

[TiN]

Yes. One is bit dimmer, but you wouldn't be able to tell unless both meters are next to each other.
I stopped caring about VFD brightness while ago, as mostly they used over GPIB, sitting in other room anyway.

Should try disconnect front panel board completely, maybe I can shove few 0.0x ppms of noise on readings?  .
57XX MFC front panel assembly goes to great lengths to shield the display PCBA from rest of the unit.

[TiN]

Got DIY replacement metal shield for inguard A1 for second meter.



Like a bought one



Installed in unit 2:

[nikonoid]

This forum and this thread specifically is a gift that keeps on giving!

I have 3458a with a 202 Slave Test Convergence error. Board swap with another repaired 3458a proved that A3 ADC board is at fault. Now this bad A3 board has seemingly identical symptoms as ManateeMafia's (quoted from this thread from 2015!).

EL2018CN is now ordered. Keeping my fingers crossed.

Thank you TiN! Thank you MataneeMafia! Thank you everybody else for participation and valuable information!

3-4 years ago, one of the broken 3458A's I bought had a bad A3 board. Given the recent repairs and valuable info from TiN I thought I would give it another shot. All of the voltage checks were performed years ago, but I went ahead and did them again. This was a donor board as I had acquired a replacement from Agilent.

The board had a 202: Slave Test Convergence error. Based on TiN's recommendation, I went through the ZR_LO and ZR_HI inputs from the A9 reference board. Then I went through the +12ref, -12ref, and +5ref to confirm they were working.

I checked the Reset lines and clock to the 8051 (U220). That looked good.

TiN had mentioned the TARM command during his repair that I didn't previously try. Using the command, I watched the outputs of pin 6 (OUT) briefly change on U110,U111,U112, and U140. So far so good.
When I got to U142, I could see the negative transition on the input, but the output never changed from +4.2V. After testing it a couple of times, I decided to replace U142 and fired up the 3458A.

The meter came up and was reading DC. Self-test and ACAL DC successfully.

The EL2018CN was sourced from UTSource .

Thank you TiN and all the others that have added to this thread. . Next step, repair defective A2 board with EL2039 from the same seller.

[TiN]

But don't get hopes too high. My luck with A3's is so far 100%. All must be replaced, that is .
My 3rd 3458C waiting ADC board so far, before doing any more testings. Current one does not give errors, but happily drifts +10ppm/day. It does not slow down in any way, after a month of operation already.

[nikonoid]

SELF TEST PASSED

It was a dead U142 indeed.



Drift test - here we come!

[TiN]

I wish I'd have your luck
GJ!

[ManateeMafia]

That is good news. 

I hope your drift tests are passing.

[nikonoid]

Just a quick report: 4 days of testing this repaired A3. Total drift 0.52ppm over course of 4 days. I.e the drift is about 0.13ppm per day and it is fairly monotonic (I do ACAL and recheck daily).

Keeping fingers crossed to the end of the week!

[TiN]

That's a lot. Unit will be out of spec in a 2 weeks, if drift speed is like that. 

[martinr33]

Well - the out of spec limits is 0.43ppm per day, and the first data here is 0.13ppm per day.

So the unit is in spec - so far -  but it is performing nothing like as well as the better units do.

Interesting that the board has a double fault.

[seebeck]

Hi TiN,
I am a volt-nut, I was emailed to you a few times for the "RAM Test 1 High" for my own 3458A.
The final root cause that resulted in the failure was a battery damaged for a RAM.
Currently, the meter be able to running normally after replacing the broken unit.
Thanks a lot!

[TiN]

Perhaps interesting retrospective test, showing 10V data over ~7 days, measured by five different 3458A meters.

Few ACALs were performed in middle to see jumps and spread. All meters were in variable ambient room temperature.
Source is Fluke 5720A/Hulk1 programmed to output 10V -2ppm value. For verification purposes 5720A and other 10V standards (792X with xDevs.com FX inside, Fluke 732B, Fluke 732A) were logged using Dataproof 160A scanner and Keithley 2182A nanovoltmeter in opposition mode. This data shows maximum deviation of 5720A output less than 0.2 ppm, which is well below daily stability spec of 3458A (0.55 ppm)

Logging settings for each 3458A (bottom side of the plot at around -2ppm for all meteres):
NPLC 100, NDIG 9, DELAY 0, TRIG SGL, Guard float on meters, and local on calibrator. Logging performed by xDevs.com Python Teckit app with thermal control inhibited.



RAW data-file

One can spot vast difference between my old "golden" 3458A's and standard 3458A in terms of 10V temperature coefficient and stability. Ambient temperature in lab was jumping around between 20.6 and 23.4 degrees due to aircon cycling. This little test again shows that even mighty 3458A DMM cannot be used as calibration standard on ppm-level without proper verification and very careful attention to environment conditions and previous historical data/stability validation.

[guenthert]

Perhaps interesting retrospective test, showing 10V data over ~7 days, measured by five different 3458A meters.

      Thanks for sharing.  We live vicariously through your reports. 

[/url]

      What happend July 1st, 12pm?  The great noise-eliminator visited?

[TiN]

Noise eliminator called no data. D3.js library automatically close all sample points with lines, so it looks like "no noise" with just median line.

[ramon]

 It's amazing how closely they track ambient temperature. And much more amazing how 3458N mirrors 3458C.

Experiment: if you monitor a trusted reference at ambient temperature (*) for 5 weeks, which combination provides the better results?

   1) any of your golden 3458 (A or B)? ...
   2) or the averaged result of 3458N and 3458C?

Edit: (*) the trusted reference at constant temperature, only the DMM at ambient temperature. (The meters are the DUT, not the reference)

[TiN]

Well I am not sure what question such experiment would answer, as there are multiple factors involved and its next to impossible to separate them even for one meter just by logging for 5 weeks.

Goal for the test above was:
a) verify A3 ATOD performance of newly repaired Agilent 3458A Option HFL. I also run SN18 type procedure [hence few ACALs during plot].
b) see how good\bad is old HP meter with fancy new RoHS 2019 A3 ATOD.
c) see tempco of meters, and check again excellent stability of my 5720A and golden 3458s.
d) keep meters busy, while I am on vacation :]

If one really care about proper DC voltage results, then one must use two (or more) calibrated DC voltage references and help from differential measurement. No need to have expensive 3458A also, just usual nanovoltmeter (2182, 34420 or older Keithley 182, 181 ) or even decent DMM will be enough to resolve fractions of ppm nicely.

[ramon]

Well I am not sure what question such experiment would answer,

Ok, sorry to interrupt. Ignore my suggestion, and please keep going with your original plan. Thanks for your hard work.

Just a comment. When I worked in testing, some of the 'Surprise!' moments came from simple stupid test setups that was suppossed to prove an expected normal behavior (re-check tests), and many other 'Surprise!' moments came from redundant, weird, stupid, useless, or a waste of time kind of test (on the eyes of the designer engineer). I was so lucky to work with an engineer that he didn't bothered to do even basic checks to its own code ... Completely off-topic here, don't want to derail this post.

[TiN]

Updated plots with 11 days of data.



Here's different view angle on same data points, ambient temperature versus 3458A ppms off 10V.  Perhaps it's easier to see tempco contribution on ppm/C plot.



Color designators are same as time scale plot. 3458A option HFL repeats standard 3458A performance, while meter with new A3 ATOD RoHS board is significantly worse.

All the scatter on the plot comes from noise of 3458A and 5720A output. Total plots have 6840 points per each meter, Sample period is 137.41 seconds per point.

[JR.junior]

Hello.
Did someome fix the ERROR 202 SLAVE TEST CONVERGENCE error. I think it´s a A1 board problem.  A3 A/D board is fresh new, version 4 (SMD parts).
Interesting thing; when I put the A3 board with "114" rundown error, the 202 error does not came along, I can make measurments. With new fresh A3 board I get immediately error 202 can't take any readings, nothing.

Kind Rgards,
JR.