3458A CAL & repair

[IanJ]

Hi all,

So I put my 3458A to Keysight here in the UK for CAL and they've come back (email) today saying it has a "faulty front/rear terminal switch which is causing issues taking measurements and will require repair".

Am awaiting the repair engineer to contact me which I presume is to ask if I want to go ahead with the repair etc.....but I've no idea if they mean the actual switch mechanism or if it's any related relays (if there are) etc.

Any thoughts?

Ian.

[Dr. Frank]

You have bought your 3458A 'used', and it's more than 10 years old, isn't it?
And it fails 2W Ohm zero @ 10 Ohm and 100Ohm range?

It's very probably the same error as on the well known one of the 34401A:
https://www.eevblog.com/forum/testgear/hpagilentkeysight-34401a-2w-problem/

The switch is very similar, at least same supplier.

My 3458A (vintage 2000) also showed signs of rising contact resistance, which I could mitigate by frequent actuation of front / rear switch.

Maybe they can calibrate front or rear only.

You may then clean it on your own, and check zero on your own.
As in the 3458A, the switch assembly is relatively small, it may not cost so much to replace it @ KS.

regards Frank

[IanJ]

Out of interest, is there any way to tell age of the unit from the serial no.....?

Mine is an HP branded unit (with a new KS front panel).

Ian.

UPDATE: Mine is a 2823A prefix so from what I read means 1988 through to mid 90's.

Ian.

[kj7e]

Not exact age, but it was built sometime not log after 1988.  The prefix only shows the model year series until the next revision was made.

I had a 3478 with a flaky front/rear switch.  Everything was fine until one day I pressed the switch then the Ohms readings where no longer stable.  I suspect it had been many, many years since it had been pressed.  A little contact cleaner and lots of cycling helped resolve the issue.  I looked in to replacing it, but it was no longer available.  If you can get a new switch for not much $, I would consider it.

[Dr. Frank]

Out of interest, is there any way to tell age of the unit from the serial no.....?

Mine is an HP branded unit (with a new KS front panel).

Ian.

UPDATE: Mine is a 2823A prefix so from what I read means 1988 through to mid 90's.

Ian.

Nope, that doesn't work for the 3458A. From the Prefix, you can't tell the Vintage directly. Mine from 2000 has the very same one.
The serial number itself will tell, Joe Geller once had a list on his site, but that link vanished - pity:
http://www.gellerlabs.com/HP%203458A%20DMM.htm

As you have a HP branded instrument, it's definitely before 2001, as that's about the year when agilent took over.

Ser No's 2823A25xxx are about from 2000.

If you registered your instrument on 'myKeysight', you may also find the origin of your instrument and the vintage on the KS Infoline, like so:
https://service.keysight.com/infoline/public/details.aspx?pn=3458A&sn=2823A25500&i=DOC&from=details&BTN_QUERY_MODEL_SERIAL=Next


Please check with KS the exact error.
If it's what I suspect, you may save a lot of money. If it's something different, as a failure in another mode, it may be a relay, as you supposed, which means that KS will replace the analogue board for a fortune.

Frank

[MisterDiodes]

Another clue on 3458a age (it's not serial number I know) - just open it up and look for youngest component date codes.  It won't be exact, but at least you'll have a general idea of what era you've got, or at least when an assembly was replaced.

If the unit was always cal'd at HP / Agilent / KS they will have records - that doesn't always work but worth a shot asking.  When we send our units in the paperwork comes back with a complete repair / cal history.  They should at least be able to look up sales date based on serial number.  We've had luck with that before.

Not sure how it works in the UK - but in the US you never send in your 3458a without first buying the extended warranty...if the unit is in current cal they will sell you the warranty, usually less than $200/yr.  Then the repairs are -much- more reasonable, and you get a discounted price on the cal too.  Worth every penny if you depend on your equipment that might need a repair now and then - and whenever time = money.

[Andreas]

The serial number itself will tell, Joe Geller once had a list on his site, but that link vanished - pity:

I downloaded this article from Joe. (you never know when you need it).

with best regards

Andreas

[martinr33]

Andreas - do you have the Geller serial number file? It isn't in the archive you posted.

And the switch assembly is available from Keysight but...

A10 Front/Rear Switch for the 3458A   Orderable      US$ 684.00

I've looked for the switch alone from distributors, but can never find the gold version. TiN reports getting one from HP P/N  3101-2969 for $75, but that is not in the current parts list - I suspect he got it through some other route. Either way, you will probably want to get the meter back so you can fix it, assuming you don't want to pay  the $684 + labor. If you have a drifty ADC (good bet that you do) and a weak display, you are close to the HP Big Fix price.

One failure cause (not in your case, but to be watched for) for the switch is, if the front retainer pops off the switch actuator, you have to find the flying spring and watch out for a little piece of bent wire that is the front/rear latch pin.

I had a Keithley meter with a flaky F/R switch. I used Deoxit - which fully restored operation.
I flushed the Deoxit, and it went flaky again.
So I left a little Deoxit in there, and all is good.

[Andreas]

Huh,

I have only the HTML-text. no extra file.

with best regards

Andreas

[IanJ]

Update:-

Keysight got back to me today.

They won't replace the switch, it has to be the whole A10 board.

EUR 52.93 (SHADOW SWITCH P/N=3101-2969)
versus
EUR 618.00 (A10 FRONT/REAR SWITCH ASSEMBLY P/N=03458-66510)

I suggested they could remove A10, send to me, I'd replace switch and send back but they are not biting.......saying that the 3458A has to come back to me, then when it's resent I pay for full cal again.

So, I've told them I am only interested in DCV right now (I manufacture voltage references)........so conditional calibration for the rest will do.......then I'll get a new switch in time for next year recal. Hopefully they will be ok with that.

QUOTE FROM KEYSIGHT:-
It seems to be behaving itself at the moment, ie. It passes the front and rear 0 Ohms Self-cals, but during calibration they couldn’t get the correct readings.
(approx. 2.4 Ohms on the front and 1.6 Ohms on the rear).
We can return the unit to you as a Conditional Calibration (there are also some “Undetermined” results on some of the AC measurements) and you
will be charged the cost of an Accredited Cal.


Ian.

[IanJ]

Update: I asked Keysight when my 3458A was made.......they've come back with Feb. 2000. Good to know!

Ian

[Dr. Frank]

Update: I asked Keysight when my 3458A was made.......they've come back with Feb. 2000. Good to know!

Ian

So all my guesses came true, including vintage / ser. no. of your instrument.
And you may repair the switch by yourself, afterwards repeating the zero calibration, which does not spoil the 10kOhm / 10V basic calibration.

Frank

[TiN]

If calibration was done with faulty zero, resistance calibration will be spoiled after repair as well. 

[Dr. Frank]

If calibration was done with faulty zero, resistance calibration will be spoiled after repair as well.

I doubt that, because 40kOhm value is determined in 4W mode, where zero calibration is not relevant.

This particular error affects 2W Ohm only.

Frank

[IanJ]

And you may repair the switch by yourself, afterwards repeating the zero calibration, which does not spoil the 10kOhm / 10V basic calibration.

I hear you, but I'll not replace the switch until just before next recal. Don't want to upset anything.

Ian.

[ap]

I hear you, but I'll not replace the switch until just before next recal. Don't want to upset anything.

Thats the right approach. There are users who state they never toggle the switch because they have no rear connections, and still the item fails when sent to cal. So if you replace it now, that may not necessarily help you. I would even go further and just start toggling the switch prior to sending it to next cal, and watch how it behaves over a period of time. You can easily verify the 0 ohms. If it is unstable only, which I doubt, then I would replace it.
The other thing is, if you just need DCV calibration, any good cal lab can do that, and it will be at a much better price. Keysights uncertainty is not very exciting, 2ppm as per manual, which adds to the uncertainties in the data sheet. They may be lower when they do their gold cal or whatever, probably also depends on where its done.

[Tony_G]

Thats the right approach. There are users who state they never toggle the switch because they have no rear connections, and still the item fails when sent to cal. So if you replace it now, that may not necessarily help you. I would even go further and just start toggling the switch prior to sending it to next cal, and watch how it behaves over a period of time. You can easily verify the 0 ohms. If it is unstable only, which I doubt, then I would replace it.

I think that is because the cal process does both front and rear connectors - If you could just get a "Front cal only" then it'd probably work for those customers.

Mine failed the same part of the test but it turned out to just be the crimping so recrimping fixed that.

TonyG

[IanJ]

Back home........

Ian.

[Dr. Frank]

Hi Ian,
the calibration documentation of your instrument are not online...
I'm curious, how this conditional calibration looks like, would you mind copying the report here?
Maybe you have access to the file ones  KS site, so that you may link the complete doc.
Thank you!
Frank

[IanJ]

Hi Ian,
the calibration documentation of your instrument are not online...
I'm curious, how this conditional calibration looks like, would you mind copying the report here?
Maybe you have access to the file ones  KS site, so that you may link the complete doc.
Thank you!
Frank

As attached........comments welcome.

PS. The docs are online but the case is still recorded as open so maybe thats why you couldn't see.

Ian.

[technogeeky]

Back home........

Ian.

That's a pretty awesome bit of kit to have, and an awesome sticker to have on it. Especially as a "hobbyist." Congrats!

[IanJ]

Back home........

Ian.

That's a pretty awesome bit of kit to have, and an awesome sticker to have on it. Especially as a "hobbyist." Congrats!

The 3458a earns its keep........I use it to cal the PDVS2's that I produce and sell.

Ian.

[Dr. Frank]

As attached........comments welcome.

PS. The docs are online but the case is still recorded as open so maybe thats why you couldn't see.

Ian.

Ian, thanks again, it's really interesting / puzzling...

Obviously, your instruments passed Ohm completely after adjustment (As Completed), also this 2W zero error of 1.6 .. 2.4Ohm has been successfully 'removed' by adjustment.
As they had to actuate the front/rear switch for that check, and as the zero reading was ok afterwards, maybe the switch itself might be ok, but the crimping not.. depends a bit on when they checked the zero reading, either directly after adjustment, or afterwards, in a 2nd attempt.
As I already assumed, this error does not affect at all the gain calibration of the Ohm mode, 10k and 100k are spot on to < 1ppm after adjustment.
Same goes for the current ranges, which were "Undetermined" before.

It's a bit strange, why they set status Undetermined in several modes/ranges during As Received, and even after adjustment.. there's no obvious failure.

Frank

[IanJ]

maybe the switch itself might be ok, but the crimping not..

I'll have a look when I remove the board to replace the switch in a years time......Any ideas what/where to look?

Ian.

[Dr. Frank]

Yes, sure.. that's assembly 66510 (see CLIP).

I'd check the resistance between the jacks (front/rear) and the corresponding soldering junctions directly at the switch. 2nd, the resistance values across the switch contacts. 3rd the resistance from the switch to the analogue PCB.

Frank