HP 3458A ; A/D board spare part availibilty and cost

[voltampere]

Dear voltnuts,

i aquired a HP3458A fromm ebay in decent optical condition.
According to the date codes on the ic's it should be from 1989.
The A9 reference seems to be stable but i suspect the A3 PCB ( A/D board ) to be drifting seriously.
The drift is very random and unpredictable.
Sometimes it will not drift for about 12 hours. Sometimes it is running away 3 to 5 ppm in 12 hours.
It will always come back within 0,2 ppm after an ACAL DC.

I got in contact with Keysight Germany to get an idea how much a replacement A3 board would be.
First problem was finding the right part number. The part that is inside my unit seems to be the first edition.
The part number is 03458-66503. In the keysight part database you will get 03458-66513 as the replacement part number for it because the 66503 is obsolete now. But findapart@keysight.com emailed me that this part is"NFTS" meaning not for trade sale and that i should contact tech support at keysight germany for a repair of the HP3458.
I asked KS germany to quote for the 03458-66513 and the first reply was that they only have a few of these boards and they will not sell the spare parts but i can send the unit and they will replace the board. Material + labour = 2.000 € ( + VAT ).
I talked to some KS support guys on the phone and asked which part they would use to repair a new HP3458A ( RoHS version ).
They were unsure if that part would be compatible and said to contact me when they found out.
I read a post here where somebody confirmed that the new version of the A3 board ( part 03458-66523 ) would definitely work in older units ( fully backward compatible). So asked them if they would sell the 66523 spare to me.
Got an email two days later where they told me that they usually dont sell the new version either but would make an exemption and quoted 2.000€ ( + VAT ) for the board only!!!

That is just a bit too much i think.
Now my questions.

Does anybody have any experience about getting an A3 spare board ( tested, no second hand ).
Should i try to get it from another support center , UK for example?
Are the old version ( 66513 ) boards really that rare already?

I dont think i want to spend that much many for the spare board.
How are the chances to get my drifty boar cured?
I had the ( crazy ? ) idea to put the board in the oven at 60 degrees celsius for some hours to get the humidity out that might be the cause of the drift.
Would it cause more harm than help?

Hoping to get some ideas from the HP3458 repair experts...
( TIN ?!!! hello...  )

Thanks to all for any input on this

Best regards

joerg

[TheSteve]

Your only choice is to buy a refurb board from Keysight as far as I know. They tend to have unknown drift/noise, some are certainly better then others.
If you can't find anything else and do really need one I have a refurb that has been tested for noise and drift I can sell.

[Edwin G. Pettis]

Voltampere, your drift problem sounds suspiciously like a bad solder joint or cracked PCB trace, you might take a very close up look at your board and see if you can spot anything before popping for a replacement board.

[TiN]

Graphs. This thread is pointless without graphs. I'd also wouldn't have any conclusion about ADC stability untill you ran the unit for a month, non-stop, to rule out storage/settling issues.

There are also more than two A3 versions, but even cheaper refub one would set you back $900+ USD if you are in US. Since you are in Germany, add extra and transform currency to EUR. So 1500 EUR+ is what I would expect as reasonable.

[voltampere]

I did a quick inspection of the board. But will have a closer look again.

TiN ar you saying that i could still get a -66503 verision refurbished/tested from keysight?
Where ? At what price?

I did let the unit run for about 10 days and if i remember it right the situation got worse.
I want to do a longer run with logging. But before i do that i want to be sure that power supply incl. all relevant components are in good health. Dont want that ugly smoke coming out when i am not around...

I think that my board is the oldest version. Pictures attached.

Regards

Joerg

[MiDi]

Not long time ago the refurbished A3 were orderable from service Homepage at keysight.
I do not have the link on hand, but I think it was posted in the 3458 black thread.

[Dr. Frank]

The refurbished A3 boards probably were also not offered any more, as far as we have found out.
The new designed RoHS compliant A3 board is in series production already since Spring last year, and therefore this is the only version left for spare parts.
And yes, all new boards are fully backward compatible.

Frank

[TheSteve]

Keysight stopped selling the refurb A3 boards a while back. They then sold them again after a few phone calls but it seems that it was short lived and there is no longer anyway to buy a bare A3 board. They offer only a service option.

[ScoobyDoo]

Hello - I have been following the price evolution of the A3 board over last couple of years:

03458-69503                     755€                     (One board – through hole components)
03458-66513                     1258€                   (multiple stacked boards)
03458-66523                     2046€                   (SMD version - WEKOMM")
03458-66533                     1004€ ... 1068€    (SMD version - KS as used in black edition)

Yes - the public sale of these A3 boards is set on hold for an undisclosed reason.
Also the 03458-66529 VREF board with 2ppm HFL spec was withdrawn despite market demand.

Also new A9 DC Reference boards are listed:

03458-66539   PC Assembly-DC Ref Tested 8ppm              469€ (orderable)
03458-66549   PC Assembly-DC Ref Tested 4ppm              469€ (orderable)

We spotted same price info for both 4ppm and 8ppm board on KS site.
These superseed the 03458-66509 and 03458-66519 DC Ref boards

Herzliche Grüße/Meilleures salutations/Best regards
ScoobyDoo

[notfaded1]

I just looked it up on USA Keysight found same basically.  Not in stock and the price is the same?   which would you choose lol

[rastro]

Sounds like KS is may be using the supply of new boards for their 3458A Black rollout and keeping the stock of older boards/refurbish to support units that are sent back for repair.

[Kleinstein]

I just looked it up on USA Keysight found same basically.  Not in stock and the price is the same?   which would you choose lol

There may be a small down side to the 4 ppm reference. I have not checked the specs, but chances are the high stability reference runs at a slightly (e.g. some 10-20 K) lower temperature and this would also mean that the maximum environmental temperature is lower. Not a problem for use in an air conditioned lab, but it could be in a hot environment in a populated rack with additional heat sources. So chances are the 4 ppm/a version would be better for most, but not for all.
If the main difference is a slightly different resistor value, there is no need for a different price.

I can somewhat understand they don't want to sell the new A3 board. As a service only option they can do a few more checks to make sure the new board really works OK in older units. Even if designed as a 100% replacement a few more checks may be needed until they have enough experience.  Even with ACAL one may still want a new calibration after such a repair - so the meter would have to be send out anyway.  For large customers who can do the calibration in house Keysight may make an exception  , though they may still out a meter for repair.

[e61_phil]

I have a 4ppm reference in my 3458A and it has the same 15k resistor.

[TiN]

All A9 variants use same 1k/15k oven setpoint resistors, there is no BOM difference, only selection process.

[notfaded1]

I would for sure order the 4 ppm model if the prices are correct.  My lab is almost constant 74°F in the winter and 76°F in summer.  I live in desert of the southwest USA in Arizona with always low humidity.  On another note to partner with my 3458 I bought this today.  I'm still building some my own references because it fun to build things and a hobby for me.  I found out the seller ran a lab and was selling all his gear because of a death in the family.  I asked the seller to include any history he has on the 732A as well and he's shipping it in a foam blown thick walled box.  Unfortunately it hasn't been powered the whole time... I couldn't get that lucky... not for what I paid.

BTW... you forsaw this TiN.  I've got some nice resistance decades but... how can I not want a certain certain 10K.

Bill

[voltampere]

I owe you some graphs.

First one is from my healthy 3458A.
Second is the "sick" one.

I did one measurement every 20 minutes. Temparature ( command "TEMP?" ) is the orange line. You can see the daily temp rise in the afternoon when the sun is warming the room that has a window to the southwest. The big temp excursion happened when i got bored and started the Datron 4700 which immediately heated the area around the Equipment by another 2 degrees. That is followed by a sharp drop wheni opened the window to let some fresh air in.
The cal constant ( command "CAL? 72") for DCV 10 gain is the blue line.
The good meter has a bit of drift but that should be in spec. Less than 0.5 ppm in 4 days. It has a negative temp coefficient.
The sick meter has a positive temp coefficient and a strong drift.  2.5 ppm in 4 days. That is around double what is acceptable according to keysight Appnote.

I bought a fresh spare board from KS germany and have the sick board on the table now.

That brings me back to the questions of my starting post.

Does it make any sence at all to try and fix/cure the sick board?
I am thinking about to give it some temp treatment. One hour in the oven at 60 deg celsius.

Any strong advice against this experiment?
Somebody mentioned a possible bad/cracked solder joint.
Should i try to reflow the U180 pins?

Any better ideas?

Best regards

Joerg

[Dr. Frank]

The sick board seems to stabilize somewhere, after some days, but anyhow, that's what TiN also encountered on one of his bad A3 boards, years ago already.
That U180 really seems to be defect.
TiN also tried everything to treat it somehow, long stabilizing on-time, temperature shock,  he even opened the chip and tried to replace some of the resistors inside.. no chance (in German: vergebliche Liebesmüh).

Maybe you search on his site for his experiments, or somewhere here, in the accompanying blog.

Sorry. Frank

[voltampere]

Right, from the plot one could imagine that it  would settle. Maybe...
But then you still have this terrible tempco.

I think i read every info on TIN's site regarding 3458A in general and also the A3 board and U180 problem.
Did not really expect someone to come up with a successful recipe for healing this board.

But i am not ready to throw it in the bin yet 

regards

Joerg

[branadic]

What I was told at wekomm is, that there is no magic treatment to heal U180, but they are replaced and boards are then tested, until the board achieves the specs.

-branadic-

[martinr33]

You might try logging both meters against a known voltage, see what that gives you. A good ADC board doesn't drift at all in the medium term, and running autocal can shift the voltage by some random (small) amount.

So:
        - hook both meters to a known stable voltage
        - Log the voltage on both meters

If your voltage is really stable, you will be able to see the drift on both units and also derive your thermal characteristics. If nothing else, the delta between the systems will give you an idea as to how bad your drifty board is.

The reason is, autocal injects some noise by shifting reading a little. 

Then finally - repeat with new board.

The reason for the Agilent method is you might not have a stable reference.

[MiDi]

My unit showed initial drift of ~0.6ppm/day for CAL 72, it took several months to settle to a reasonable drift rate - now after more than a year it has < 0.04ppm/day.
CAL 72 shows a TC of -0.33ppm/K, which is in compliance with some others units I had ability to compare - up to 0.5ppm/K is acceptable for ADC according to HP Journal.

[notfaded1]

@MiDi what tools did you use to create that CAL 72 graph?

Bill

[voltampere]

You might try logging both meters against a known voltage, see what that gives you.
-------

That is what i did. It was running away a couple of ppms per week. That is why i go suspicious and started the CAL 72 log.

A drift on a log of a stable voltage source should also show up in the change of CAL 72 in my understanding of the ACAL function.

MIDI:
My eyes ar not good enough to read the horizontal scale of your graph. Is it one week per division?
Your meter was on 24/7?

Regards

Joerg

[TiN]

voltampere
I don't see much drift in your hp3458z plot. Just normal settling for warm-up time.  So don't understand where you get conclusion that your A3 is bad.

[MiDi]

@MiDi what tools did you use to create that CAL 72 graph?

The chart is made with Excel and updated automatically from csv with the captured data.
The extra lines and text are inserted manually, the TC compensated dppm are calculated with determined TC of CAL 72.
For logging I use RasPi with Linux Gpib and Python program that logs the CAL values every hour into csv.

MIDI:
My eyes ar not good enough to read the horizontal scale of your graph. Is it one week per division?
Your meter was on 24/7?

Main horizontal scale is one week, starting at 2018-11-26.
You can click onto it to get the full resolution and zoom in to read it better, but I have to admit that the horizontal scale labels are getting a bit to small for presentation.
Unit is on 24/7 with some exceptions.