[3458A] Yet another 3458 repair thread.

[TiN]

I know some people here might hate me , but I couldn't resist buying another parts 3458A, listed with good price
Got the box today. 



Question is, will I need to spend another 1.5-2K$ on top of the unit price to get it fully function, or luck will turn face to me this time? 

[HighVoltage]

How many 3458A do you have by now?

It looks like someone attempted a repair already? Or why are the push rods missing of the switches?

[TheSteve]

I know some people here might hate me , but I couldn't resist buying another parts 3458A, listed with good price

You're welcome! 
and you should have bought the second one too...

[IanJ]

Looking forward to following the repair........always something for me to learn.

Ian.
[3458A repair newbie]

[Theboel]

just wondering how much the good cost for 3458A with condition like that ?

[martinr33]

Well, at least you don't have to worry about a bad A3 board!

[TiN]

martinr33
Description was not accurate, and unit came with actual A3  So I do have to worry about A3, which likely need replacement. I always have "good" luck with A3's..

But before we go deep into analog domain shenanigans, gotta fix ROM 0 LO error first.

[martinr33]

I guess they wanted to justify the low price, hence the listing claiming no A3 board!

Looks like you are only mssing one bit in that EPROM, should not be too hard to find.

What do you figure the odds are  that you have a drifter - 30%? Sometimes, I wonder if these unist pop up because they drift, but I suspect that most sellers have no clue about it.

And if you want the proper color keypad when you update that bezel, you'll have to call HP as it isn't on the web site.

[picburner]

TiN, you have to upgrade your TL866CS, now we are at V6.82...
I am eager to know the developments of the new 3458A repair...

[TiN]

TiN, you have to upgrade your TL866CS, now we are at V6.82...

Does not matter a single bit.

I am eager to know the developments of the new 3458A repair...

Alright, I'll cut the suspense... Meter is working now, just replaced NVRAMs and reflashed firmware UVROM.
To my surprise it even passed self-test and ACAL.

It's mismatch set of boards:

A5 is datecoded 1994
A6 datecoded 1998
A1+A9 set datecoded 1997, A9 is option 002 with selected LTZ1000A.
A2 datecoded 1997, and it is H01 high-voltage AC option! (usual 3458A specified only for max 750VAC input signals)
A3 datecoded 2004

Few money shots from upcoming article:



Voltage reference, marked as 03458-66519 (Option 002 4ppm/year).



Started initial calibration after few hour warm-up, just to see where we stand.



DCV voltage calibration vs 792X.



So obviously calibration ROM contents was not correct for this meter.
I ran quick calibration to my 792X 10V and SL935 10kOhm and now meter is running performance verification procedure against Hulk. Full report will be cooked in 20 odd hours or so.



It remains to be seen, if A3 will need to be replaced.

[dl1640]

Subscribed!

[HighVoltage]

wow, congratulations !
It seems you got lucky this time, some pay more for the 002 option LTZ board alone.
And you got the 1000VAC option as well.

Your collection of instruments are really impressive !

[dl1640]

i once applied 1000V AC into standard 3458

[PartialDischarge]

Nice, I know you're going to be lucky this time...it is time you deserved a 3458 with a working A3 

[TiN]

So far 10V comparison look good.

[Zucca]

[TiN]

And I spoke too fast...

A3 ADC does not look good a single bit.
2.9 ppm / day.

Oh well, you didn't really expect to have cheap 3458A with good ADC, right?

My luck with 3458s proven once again, every single ADC in my boxes must be replaced. 

[ManateeMafia]

Keysight thanks you for your continued support.   

[MadTux]

Is there a reason for the A3 U180 cancer, like stored in damp shed for long time, degradation over time, diffusion of water into chip...?

Not much fun to buy a $4000 DVM if it drifts like a $200 DVM after a while for unknown reasons and you can't do anything but to buy new board from HPAK for another $2000.

BTW is there a 8 1/2 digit DVM that isn't build from custom parts? K2002 seems relatively free from unobtanium while Datron/Wavetek is full of it. HP 3458As have drifty unobtanium ADCs. How about Fluke 8508, same madness like Datron or have Fluke engineers improved upon that? Also never seen a Solatron 7081 from the inside....

[dl1640]

to leave the 3458C powered on for a few weeks then check again?

[ManateeMafia]

Leaving them on has not shown to improve the drift. I have had a few do the same. It just delays the repair and is a daily reminder of how expensive it can be to play in the extreme volt nuts game.

Buying a used 3458A should always include the expectation of replacing the ADC as a minimum though you can get lucky, for a while.

[ap]

I have seen many of them with this fault. Interestingly enough, my experience is that the eray units (say in the 4 digits SNs) do show this issue pretty seldom, its mostly higher ones. Just my onbservation, but having serviced dozends of them. The drift rates I have seen was between roughly 1 to 20 ppm/day, so can get pretty dramatic, but can also be so low that you can essentially continue to use it (spec 0.43ppm/d). Important is to test over a couple of days (service note says 7). So don't test just a day, that will be misleading.

Re. the questions about custom parts (or custom selected), the 3458A has many of them, true, some being available, some not, just board level repair then, unless you have a donator board. Selftest not very helpfull, does not mean a lot if passed (as many sellers maybe think/pretend).
The Datron meters have a few custom parts too (mostly gate arrays, reference RN, transformers...), but here the issue is you cant get any of course. Display and also mechanical parts (posts) are sometimes an issue too.
The 7071/7081 meters usually are better to repair, most parts available, and straightforward technically. Even displays available. Underrated meter imo. But has its challanges (leakage, other stability issues; selftest is a good indicator though)

[martinr33]

That the A3 board works, but drifts - says someone knew what they were doing. I've always wondered how good cal labs are at spotting this.

I'd guess that  the unit failed in service in the last year or two, judging by the calibration sticker.

So someone likely switched the ADC in with one from a drifter unit.

Looks to me  like you have about a 50% chance of drift in older units, more if you have one of those mentioned in the  service note, and less if it is older. TiN's a bit on the unlucky side with these, but he's getting his luck in other places.

[TiN]

Drift didn't cure itself, as expected.

A3 drifted -29 ppm, while A9 LTZ ref changed it's value only -0.34 ppm



Options from now are :

Plan A:
* Continue with refub of the unit, replace all busted mechanical parts and order new A3. That will set back the project cost to about ~2K$, similar to my other 3458s. 
Hey, I knew what I stepped into, I expected no A3 at all from the listing description  .

Plan B:
* Give up on A3, and use this box as Frankenstein box to test wild ideas, which we all wanted to try, but never dared to ruin good working 3458A for. 
For example make DIY references, swap opamps, try different 40K resistors as reference, remove high voltage protection for sake of better LV VDC perf?
It will be bit tricky to test results of the mods with drifty ADC, but we should be able to see if short-term noise/tempco performance improved even with slow drift.
That's given the ADC does not completely die in coming weeks , like 3 of my other A3 boards...

Plan C:
* Give it away for somebody in community who want to play 3458A game to trade for some other toys? 

What say you?

[Dr. Frank]

Plan B  makes no sense, as these experiments require a stable A3.

If you have no personal need on another working 3458A, as I have the impression, just replace A3  and sell the unit to another hungry volt-nuts.

Frank