Agilent 3458A - I joined the club

[Grandchuck]

Will you then paint it black?

[aronake]

I swear I'll do it! I'm not kidding, I'll totally do it and I'll be glad, you hear? GLAD!

Is the Agilent panel badly scratched? Generally I think the Agilent panels looks better, and the genuine time correct choice. And i wonder who put 1/2 at the bottom of the keysight panel.......

[TERRA Operative]

I tried to get a replacement Agilent display window, but none available...

The Agilent one is a bit scratched. I'm almost thinking of making a stencil (I have a sticker cutter) so I can polish the Agilent window and repaint the writing and logo with a stencil, maybe.



re. painting it black.... Even I have limits to my madness...

[Dr. Frank]

I tried to get a replacement Agilent display window, but none available...

The Agilent one is a bit scratched. I'm almost thinking of making a stencil (I have a sticker cutter) so I can polish the Agilent window and repaint the writing and logo with a stencil, maybe.
..

I'm so glad, that my unit from 2000 still got this HP logo on the display filter, even though it had some scratches across.
I just used plastic polish to get rid of all but one faint scratch .. I would have felt uncomfortable to have such a legendary DMM with agilent or keysight on it.

Well, the VFD of my unit shows signs of use, but I still have no visual problem with that, as I really USE it when making automated measurements.
For me, it's not a  museum artefact, but a tool.
Frank

[IanJ]

I just looked out my old "HP" front panel and it's certainly got a few dings, bangs and chunks out of it......however, the HP window is pretty nice and I am sure it would polish up.......so maybe I'll replace the Keysight window that came with the new Keysight front panel I had fitted back in 2018.

To me, it's just a personal decision on whether to stick with an old battered front panel or have a brand new one. I prefer pristine test gear hence why I changed mine........but I do think, a nod to the original 2000 heritage of my 3458A is probably worth doing.

Now where did I put that perspex polish.........



Ian.

[TERRA Operative]

Well, I did it. I stuck the Keysight display window in.    More shiny!


I also made a leaf spring (Part number 1460-1311) for the grounding of the front panel.
As previously mentioned, it's a non-purchasable part, so I had no choice but to make my own...
The parts manual says it's a Be-Cu leaf spring and I measured the remaining bit at 0.25mm thick, so I bought some 0.25mm thick Beryllium Copper sheet from Aliexpress and cut a chunk out, drilled the holes and tinned it with the soldering iron.
I contemplated nickel plating it, but that was all a bit too hard. Solder is fine. When this goes in for calibration, I'll try to convince them to replace it with the real part then.

I'll install it Sunday night, once the timer for one week continuous operation has elapsed as specified in the drift calculation in service note 18A, so I don't ruin the test.
Then I can compare ACAL constants to check for the common drift issue, at least for the 1 week test. Longer testing will follow once it's installed back in my rack.

[Dr. Frank]

Well, I did it. I stuck the Keysight display window in.    More shiny!


...

I'll install it Sunday night, once the timer for one week continuous operation has elapsed as specified in the drift calculation in service note 18A, so I don't ruin the test.
Then I can compare ACAL constants to check for the common drift issue, at least for the 1 week test. Longer testing will follow once it's installed back in my rack.

Congrats, and have fun!

As Ian is currently in the thread, let me ask again, whether your 18A drift measurements of U180 (by means of Ians program) take its temperature coefficient into account, i.e. compensate for this effect?

I guess, I'll append my latest measurements for comparison. The U180 drift nearly vanished over the past few years.

Frank

I always monitor CAL? 72 (DCV gain) and CAL? 175 (ACAL DCV temperature).
The first diagram then calculates the T.C. of U180, which fits to the 3 T-points taken during warm-up.
The T.C. was later confirmed by separate T.C. measurements of the overall T.C. minus the T.C. of the LTZ1000A reference.



When I plot the U180 gain constant over time, and substract the T.C. influence, I get a seemingly consistent, temperature-compensated drift curve for U180.
In 2019, I made the latest 10V DC calibration, which automatically gives a small jump of CAL? 72.
The drift was already very small, 0.5ppm/y or 0.002ppm/d and practically vanished for the last 2 years.     

[IanJ]

HP window fitted on my 3458A today…….I feel better!

Ian

[TERRA Operative]

And to provide balance in the cosmos, mine now has a Keysight window. Ying and Yang.


Tonight marks the end of 1 week constant power on my 3458A to check for A3 drift as per Service note 18.
The service note gives a maximum value of >0.43ppm per day, and once I plug my values into the formula, I get a daily drift of -0.0189ppm, so I think I'm good. 


I also installed my little leaf spring too, so that's all good now and the instrument has been returned to the rack to be left running 24/7. (I use the DISP 2 command from the front panel to blank the display when not in use. I also set NPLC to 200 to reduce wear on the SMPL annunciator as well).


So all is happy for now. Next step is to try to figure out how to figure out what is up (if anything) with that -1A range from earlier.
I need to figure out how to measure that with only one 3458A. Or just bite the bullet, get a cal done and hope it is good.... 




Also, I uploaded a video to youtube on replacing the NVRAM's, I think I got the details correct...... 

https://youtu.be/uNbI3SUZ7oI

[Dr. Frank]

There is a special GPIB command which turns the display off completely.
Somewhere in the Metrology folder I have described how to use the proper command syntax. Maybe it's one of the additional BASIC commands, then it might be used with the function keys

Such a command you could find on many instruments, when complete darkness is required for optical experiments in the lab.
Frank

Got it:
https://www.eevblog.com/forum/metrology/repair-old-rusty-hp-3458a/msg800134/#msg800134

Problem with that command  is to properly define the  correct quote marks in your specific programming language

[drhex]

Mine also got a Keysight window, ditched the original HP as it was beyond polishing. Whole front did look like it had been in the dump already. Using @ianj's spreadsheet, I arrived at 0.03155 average drift, given I am in a totally temperature uncontrolled environment I consider this an excellent result. Figuring out TC would be an interesting one actually.

[TERRA Operative]

Well, after a continuous run over the last 60 days since last ACAL, I get a drift value on cal constant 72 of 0.0004ppm per day.

0.4ppb/d isn't too bad I guess.


I'll have to set up temperature logging and start logging data points more frequently to see how it drifts up and down over time.

[Dr. Frank]

Well, after a continuous run over the last 60 days since last ACAL, I get a drift value on cal constant 72 of 0.0004ppm per day.

0.4ppb/d isn't too bad I guess.


I'll have to set up temperature logging and start logging data points more frequently to see how it drifts up and down over time.

Which drift are you referring to?
The U180 will not drift , but the  volt and resistance references inside might drift over time.. which can be monitored by external references, and which do not drift in my instrument (from 2001) at all.

i do not leave my 3458A running all the time, only when in use, and I lowered the oven temperature to about 65°C, from 95°C.
Therefore, the LTZ chip does not drift any more.

Frank

[TERRA Operative]

I'm talking about the Service Note 18A, testing for drift of the A3 assembly.

[aronake]

Don't you think its time for your 3458a to get a brother? Then they can help track each others PPMs.

Here is an interesting grab as you live in Japan. May just be to change the user data and calibration rams to get it working properly.

And congrats on the A3 stability on the one you bought.

https://auctions.yahoo.co.jp/jp/auction/k1173946045

It is allways a bit of A3 stability gambling to get an used 3458a though.

[TERRA Operative]

I wish I could buy a second meter....

But I just bought an Advantest R6161 that set me back a bit, so I have to keep selling off unused equipment to make space and more funds.

[aronake]

I wish I could buy a second meter....

But I just bought an Advantest R6161 that set me back a bit, so I have to keep selling off unused equipment to make space and more funds.

I have one too. With a bit of tweaking of the current of the zener reference, and insulting the zener reference a bit more, i have gotten it pretty stable and pretty low noice for postive voltages. For negative voltages is show for some reason I dont know quite a bit more noise. Ideally I would want to swap the zener for a LTZ1000 circuit, but as the zener is built to sink current in a way a typical LTZ1000 circuit cant handle, I have not came any further on that.

[TERRA Operative]

You don't happen to have any schematics or service manuals etc for the R6161 do you?
I have the English operation manual, but Advantest were generally not very forthcoming with schematics and in-depth service info..

[aronake]

You don't happen to have any schematics or service manuals etc for the R6161 do you?
I have the English operation manual, but Advantest were generally not very forthcoming with schematics and in-depth service info..

Unfortunately not. Only what you have.

[guenthert]

It is allways a bit of A3 stability gambling to get an used 3458a though.

   Just out of curiosity (I don't have the means to join the club and fortunately no need either), how bad is the A3 drift, when it drifts?  I understand that such an instrument won't match (or rather cannot be calibrated to match) its stated specifications and won't be of much use as a voltage/resistance/current meter, but will it measurably affect its use as ratio meter?  I think it was @alm who mentioned that he uses his meters only as ratio meters anyway.  If one is interested only in DCV or DCI and resistance and has good (stable, calibrated) standards available, then a 3458a with a drifty A3 would still be of value, wouldn't it?

[TERRA Operative]

The magic number is 0.43ppm drift per day.

If you exceed that, your instrument is out of spec and needs a new A3 board.

[Kleinstein]

It is allways a bit of A3 stability gambling to get an used 3458a though.

   Just out of curiosity (I don't have the means to join the club and fortunately no need either), how bad is the A3 drift, when it drifts?  I understand that such an instrument won't match (or rather cannot be calibrated to match) its stated specifications and won't be of much use as a voltage/resistance/current meter, but will it measurably affect its use as ratio meter?  I think it was @alm who mentioned that he uses his meters only as ratio meters anyway.  If one is interested only in DCV or DCI and resistance and has good (stable, calibrated) standards available, then a 3458a with a drifty A3 would still be of value, wouldn't it?

HP gave the limit of 0.43 ppm per days, but they should probably better give it as 3 ppm over a week. For a reliable value one would usually want a week or more (if the temperature is not that stable). Usually good meters have way less drift, like < 1 ppm/week and bad units tend to get way worse, like > 10 ppm per week.

A 3458 with a drifty A3 is not really usuably, at least not seriously: the calibration is no longer valid and expect errors.
The drift test checks the ADC gain. No need for an external reference to correct that direct effect. The simple ACAL call would correct for the ADC gain drift - this way the dirft is also measured. However ACAL can not correct for other errors (e.g. linearity - up to the point of causing an error message) that usually come with the defect.  The gain drift is only the error that is relatively easy to detect with no extra hardware.
Ideally HP/KS should add a check of the gain drift as part of ACAL - they could keep a few older ACAL values and see major drift. No sure of the meter contains a real time clock though, but many meters run 24/7 anyway.