HP 3458A - Fails Slave Test - Convergence

[Ade911]

Hi team,

I have an HP5458A that has started to fail the bootup self test - Fails Slave Test - Convergence.
The manual suggests this relates to A1 PCB DC Input. Ive had a probe around and I've noticed that Q106 is on which is fine but the source voltage slowey drifts from zero to above 10volts over time. I did remove Q106 which stopped the voltage rise so it was replaced just in case it was leaking. However the fault is still there so before changing any other parts I did wonder if any one else could recommend where to look.

Any support and advice would be most welcome.
Many thanks,
Ade.

[Swake]

Is that thing trowing an error 202?
The 'internet' suggests several things:
Jumper PCBA on board A1 is missing or configured wrongly. It must be placed in position 'normal' over pin 2 and 5.
If that is check then proceed with an autocal.

[Ade911]

The jumper is in the correct position, autocal fails. probing TP11 on A1 you can see the voltage jumping most times to 10V, then 1v, cant read the 0.1 on the scope in the time given but still then after it fails the test. After doing this a few times I now have a further error ACAL DCV, OHMS R. I dont know if its a temperature issue or something is failing the more I test it. I still get when checking the error code
202,"HARDWARE FAILURE -- SLAVE TEST : CONVERGENCE

[Ade911]

Hi , I really need some help please.

I started with Fails Slave Test - Convergence

I now have this plus ACAL DVC, OHMS R error.

Dr David Partridge on xxx@groups.io suggested i try a Autocal ACAL DCV which i tried but still i have the above errors.

Monitoring TP11 on A1 I can see a signal jumping from zero to around 7 volts, the screen also shows its sampling, then it beeps and stops , or some times it will sample , a relay will click and TP11 will jump to 10V before failing again.

Ive checked the Convergence circuit with my very limited understanding of the circuit. FETs are on and off , ive cycled functions to see if the FET will change state just to see if its working.

As the test sometimes goes further than other times I took the decision to replace U7,8,9,20,17, 14,11 [I know some chips don't need changing but desperation cut in]

After powering the system back up I still have the same issues.

Has anyone seen these issues, can some one please direct me to where I maybe should be looking as obviously Im in the wrong area.

The unit was working fine several months ago, the was a bad connection between -20v (-21v) and the power supply but that was resolved by reseating P3.

All voltages recorded on the schematics have been checked and are close to the spec [maybe my old fluke meter needs calibrating as well]

Any advice and support will be most welcome or direction to any manuals explaining in detail how the circuit works would be very much appreciated.

[Ade911]

This afternoons update.

Still shows Fails Slave Test - Convergence on power on

After finding -20v missing through P3 to PCB A1 i reset all the DIN cables , magically the system started to sample more when testing DCV.

Autocal DCV

Sample 10V OK
Sample 1V OK
Sample 0.1V random samples
Sample Zero DCV 100
Sample Zero DCV 1KV

Click of Relays
Screen shows VDC [ERR]

Reset
ACAL OHMS REQUIRED
Sample ACAL OHMS 10K
Sample DCI 1/10u
Beep

Screen shows VDC [ERR]
Still has ACAL OHMS REQUIRED

Any thoughts what to check next please

[Ade911]

Todays update

Managed to borrow an A1 bd, replaced and still the same problems.

Should I replace the memory devices on the CPU board as they are rather old and maybe the batteries are dead.

Support most welcome please

[Dr. Frank]

Hello,
it's never a good idea to blindly change components on suspicion, especially not on the 3458A.
The nvRAMs very probably have nothing to do with this error.
Check their Date Codes, please. As well, provide Date of Production (DC on other components) for your instrument, so that we have the possibility to judge possible errors.

This kind of error often relates to the fast comparators EL2018 on different boards, in this case I assume U142, or U181, less probable U405 on A3 ADC board.
There's another of these comparators on the AC board.
You get much better ideas what might be wrong on xdevs.com, please check his different repair articles (3 or 4?).

At first, please make regular measurements in all modes, which fail and which work.
Short NPLC and longer NPLC in DCV might have different behavior and might indicate a failure of said EL2028.
You did not provide any details, how and when you got the instrument, if it was DOA, or failed after transport, or some usage time.
If latter cases apply (vibrations?) then you should check all of these serial fiber optics interfaces, if they are properly stuck into their counterpart.
Sometimes, some of these optical interfaces fail.

Please make systematic failure analysis on this instrument, so not to increase your trouble.
Frank

[Grandchuck]

https://xdevs.com/fix/hp3458a/

You are probably already aware of TiN's work ... but, just in case ...

[Ade911]

Thanks Grandchuck for the link, Ive started reading , truly amazing what was achieved.

Dr Frank, thank you to for the support , you have so much understanding of these meters its truly outstanding.

My initial problem relates to the understanding of the circuit, I thought HP would have a guide showing the flow of signals through the system but I couldn't find anything , only the amazing write up re the A/D and signal conversion.

System history,

Not sure how you confirm the age of the unit but I guess around 2001 due to the old labels on it.

Managed to purchase maybe 5 years ago , upon trying to power on at that time I found the inguard board was burnt around CR11,CR12 maybe caused by the incorrect power setting from the previous owners. These were replaced and the system powered on, but due to travelling allot at the time it was switched off and left for 2-3 years.

After getting more time power was reapplied to the unit, only to then me confronted by the smell of death from the input filter frying. This I replaced along with some of the large caps on the two power supply boards [ recommended on line to change ]

 Again due to time away the system was left off for another 2-3 years but now I have time to use more. The last time it was powered up the unit sampled correctly, passed all internal self tests so it was a surprise to be confronted with the initial error. when i switched it on again.

I focused on A1 due to the manual fault finding recommendations by HP but as now proved there guidance note was incorrect I believe although its doing more now that before I changed the devices and re-seated all the interconnects so not 100% sure what helped.

Ill trace the signal as you suggested around these devices , as nothing works , well I cant get the meter to read any values , is there a way to bypass the initial startup tests to be able to read values.

Many tx Ade.


 

[Grandchuck]

http://hparchive.com/Journals/HPJ-1989-04.pdf

[Dr. Frank]

you may read date codes from the nvRAMs and from any IC on A1, A5, etc.
First will give you a hint about the sanity of the calibration constants, 2nd will give the age of the whole instrument. there also some stickers on stamps inside. The serial number as well gives a hint.. I have a table somewhere, to match production date and ser. no.

If its really from 2001, like my unit, then 20 year old nvRAMs should be replaced.. save content of calram at first

There's been a lot of recent damage inside the instrument... so that's unpredictable, what has happened.
if the unit worked until recently, then I tight not have affected the comparators, which are unobtainium.
these usually  fail for very old units with high use in high temperature environment, due to heat death.
check all the fibre optics. maybe the shift register chain is broken, I remember one instance.
try to change NPLC to small numbers <1 , hopefully the instrument reacts to commands.
BTW.: TiN / xdevs has most experience on the 3458A and really disassembled each and every part of this DMM.
Anyhow, thank you for the compliments

[Ade911]

I checked U181 and this looks fine,

All signals are fine except pin 3 where it toggles only to 7v during a DCV cal then fails / stops sampling, or 10v when it does a little more sampling and the output jumps to 5v [plus relay clicks on].

U142 , again I believe is ok, the output is toggling between 0 and 5v, however if I scope the input pin 3 the sampling fails [TP140] , its as if the system doesn't like the extra load, which is strange as its only a normal scope probe.

You mentioned the shift registers , where or is there a guide explaining how they work [plus other sections, signal routs etc], i confess i didn't read all the HP Journal shared by Grandchick because it seems to focus only on the logic behind the A-D conversion process.

OK , of to test the other items you mentioned,
thanks again
Ade.

[Ade911]

Dr Frank,

Details on the Dallas chips

DS1230Y-150
9538U 067285

DS1220Y-150
9547D 070606

Not sure if that is the information needed to confirm the age.
BTW , is there a procedure I can follow to download the stored data / replacement please.

Many tx
Ade

[Ade911]

update

So I guess the dallas chips are from 1995 week 38 / 47 after watching videos about them, ill order replacements and hope you can clarify how i can download and replace.

I also found an old communication from TiN re the TARM command, OMW , after doing this the meter came to life and did a few samples before stopping again [but it is repeatable]. With nothing connected im getting various voltages between +5.xxx and +10.xxx , If you do an autozero i then get +10.xxxxv.

[Ade911]

Todays update

only samples if i manually trigger TARM command, no idea why.

when i initially started to look inside i found the voltage on U100 pin 6 is slowly drifting up , its still doing it now. Unfortunately I damaged Q106 but was able to replace so didnt think any more about it, but today I find myself again back here as the signal DC_AD is slowly rising and is reflected on the screen [least I know the DAC is ok]

screen readings

DCV volts ? +0.689 max then freezes - error message time out read A/D 108 [but it cant be the DAC]
ACV mv samples for ever 1.945mv AC
ACV 000.081 and rising VAC
OHMS +0.VLD GOHM
OHMS +OVL.D MOHM stops sampling
OHMS +OV.LD KOHM
OHMS +OV.LD OHM
DCI nA 000.173
DCI uA 0.000159 stops sampling
DCI mA 0.0000248
DCI A 0.0000135
ACI uA 000.0434
ACI mA 0.000286
ACI A 0.000268
FREQ 50.xxx HZ

Data error matches system asking for DCV and OHMS calibration

Any support most welcome please

Tx Ade

[Kleinstein]

The convergance failure / salve test fail and hang up state point a little towards the ADC board (A3).  Here known failing parts are the already mentioned comparators, the fiber link (e.g. loose connector) and also the worst case U180 failure.

A drifting signal in the voltage reading mode cold be something like a porblem on A1 with the input path (e.g. relays or as TiN just had it with the protection resistors).  A open path there can make the amplifier slowly dirft. As the readings in current and AC mode seem stable, this points in this direction. Of cause just an open input (e.g. switched to rear) could also cause drift and an overlaod reading with ohm tests.

The odd trigger behavior could be as simple as a odd configuration still remembered in the buffered RAM.

[Ade911]

Kleinstein thank you for your feedback and advice. I want to believe that the A3 Bd is OK only because  it does measure a value  plus its expensive to replace.

Plan of action,

Purchase new memory devices and try to copy over the existing data.
Break the loop on the A1 Bd and see where the rise in voltage is coming from around Q106, Q107, Q108.

I read that the original FETS are obsolete, has anyone found an alternative device just in case one is leaking ?

Many tx Ade.

[Ade911]

Sorry meant to ask, as the Dallas models are now obsolete which version should i purchase.

Also what type of reader , programmer is recommended wih out spending mega bucks.

Many tx Ade.

[Ade911]

Update

Purchased a GQ-4x4 programmer , removed all Dallas devices and copied data to new ones , replacing in the meter , now suffer from needing all calibrations , not sure if the copy was OK although the programmer said they were - really confused over this. BTW the programmed sometime [if you programmed twice or  ore] would randomly say there was a problem with the Dallas chips , not sure if others have had similar problems with these programmers.

In parallel I found the output of U105 which drives analog switches on A1 was floating all over the place , what ever i tried on the input had no effect on the output so removed , need to wait for a new device but for now running LM709 which is not ideal but I dont mind as the out put which drives U104 is grounded anyhow so I could actually run without the device I believe at present.

As suggested I also checked all the fibre connections which seem to work OK well I believe they do , scoping one end and the other the signals jump up and down at the same time.

So now things get worse, the system was trying to calibrate DCV before I checked the fibre links, it was sampling but now nothing.

It only samples when I activate TARM and in KOHMS although it shows overload.

TP100 is also slowly drifting up to the supply rail voltage, Ive replaced U103, U108 and still no joy.

The input to Q103 is stable , zero volts if i recall , if I remove U103 everything is fine and I don't see any drift so could it be the FET Q107 Q108 well not sure. As I don't have spare and Im not sure if we can get replacements - SOMEONE PLEASE CONFIRM IF IM WRONG as Im scared to remove them.

Can someone please also confirm where the calibration 10V comes from as well , again I'm confused to know where its generated and fed from , you normally see in the screen when sampling but I don't.

Starting to wonder if Ive started on something i cant repair.

[Kleinstein]

There are too many, more random changes of parts and to little measurements / checks before. This way lots of parts get touched and possible new problems added.

The part around U105 and U104 should not effect the normal voltage readings in the 10 V or 1 V ranges, even of the part is bad or possibly missing. So no need to worry about them for the start.
These 2 may effect the 100 mV range and especially the 4 wire low value resistance ranges.

Removing U103 (LM399) would leave more than 1 FET switch active at the same time for the gain setting. Worst case this may damage things, though chances are the parts would survive.

For debugging the input amplifier it would help to start with the non AZ mode, though with a stable 0 V at the input it would not make a difference.
For the 10 V range pins 1 and 2 of U103 should be at the negative rail, some -18 V.
Pin 14 should follow TP100 = the amplifier output.
TP101 as the buffered feedback signal is also of interest.

The data for the 10 V calibration constant should come from the CAL ram chip.
A 10 V signal is used in the ACAL procedure in the calibration of the HV divider. Where that signal is comming from is a good question. I don't know for sure, but it could come via the shunt chain and Ohms source.

[IanJ]

Update

Purchased a GQ-4x4 programmer , removed all Dallas devices and copied data to new ones , replacing in the meter , now suffer from needing all calibrations......

Did you try extracting the Cal data via GPIB first?
There are a few methods to do so.

Theoretically, extract the ram data, (peace of mind achieved that you have it all backed up), then send to a programmer and burn the new Dallas ram IC's without disturbing the existing ones until you really have to.

There are different variants of the Dallas ram IC's.

PS. I use a Dataman S4 programmer.

Ian.

[coromonadalix]

some search here may helped

https://www.eevblog.com/forum/repair/minor-repair-hp3458a/
https://www.eevblog.com/forum/repair/hp3458a-sram-replacement-copy-or-calibrate/
https://www.eevblog.com/forum/metrology/yes-another-3458a-repair-and-nvram-replacement-to-ramtron-fram/
https://www.eevblog.com/forum/metrology/3458a-how-do-i-read-the-dallas-non-volatile-sram-date-codes/

https://www.febo.com/pipermail/volt-nuts/2009-August/000052.html
https://www.febo.com/pipermail/volt-nuts/2014-August/003482.html

https://xdevs.com/fix/hp3458a_u4/
https://xdevs.com/fix/hp3458a/#diag1                 repair of the slave test convergence ?? and some software down the page
https://xdevs.com/article/hp3458a_gpib/#calgpib

[coromonadalix]

Annnnnnnnd this software piece to control the meter and do the nvram backups

https://www.eevblog.com/forum/metrology/3458a-logging-via-windows-app-revisited/

1st reply  down

https://www.eevblog.com/forum/metrology/3458a-logging-via-windows-app-revisited/?action=dlattach;attach=1782284

[Dr. Frank]

@ ADE911:
It's a pity that you did not use the search function of this forum upfront, and you listened to people who don't even own a 3458A.
The TL866 plus would have been the better and maybe cheaper solution. It would have been necessary to download calram content via GPIB first.

Now, after all these unthoughtful changes you made, it seems that calibration constants are probably lost.

Frank

[Ade911]

Thank you all for your advice and feedback, yes not proud of what i did but you sometimes get to a point where you do big changes usually around 2AM.

I don't have a GPIB interface so hence the decision to very carefully remove the Dallas chips and copy. I've attached copies of the data uploaded from the chips, I believe its still ok unless I dropped a bit somewhere or you think otherwise.

Re your advice and comments below, Ill carry these out today and report back.

For debugging the input amplifier it would help to start with the non AZ mode, though with a stable 0 V at the input it would not make a difference.
For the 10 V range pins 1 and 2 of U103 should be at the negative rail, some -18 V.
Pin 14 should follow TP100 = the amplifier output.
TP101 as the buffered feedback signal is also of interest.