[FIXED] HP 3456A DMM fault: displays random numbers

[enut11]

This morning my faithful old HP 3456A 6.5d DMM failed when I switched it on

Yesterday I logged about 12hrs of data and then switched it off. This morning I find that it displays random numbers?

This is what I have found so far:
DCv no logical response - random numbers around 0.5xxx
ACv no logical response - random numbers around 0.05xxx
2W Ohms no logical response - random negative numbers even when open circuit
However,
4W Ohms accurate response on all ranges - interesting

Pressing the 'Test' button flashes -4.00000

Just flagging this for comment while I delve further. At the moment the DMM is under a lot of other test gear.
enut11

[MadTux]

Already recapped?
Usually the first thing that goes bad on these, especially if not abused, as in your case.

[Kleinstein]

If 4 wire ohms works, but 2 wire ohms not, I would suspect trouble at the very input of the main input terminal somewhere to the point were the Ohms sense Hi joins the path. This would be a rather limited area, like input protection, maybe a cable or relay.

Higher higher DC volts (e.g. 100 V range) could also be interesting - chances are one would read near zero.

[enut11]

If 4 wire ohms works, but 2 wire ohms not, I would suspect trouble at the very input of the main input terminal somewhere to the point were the Ohms sense Hi joins the path. This would be a rather limited area, like input protection, maybe a cable or relay.

Higher higher DC volts (e.g. 100 V range) could also be interesting - chances are one would read near zero.

@Kleinstein
Applying a 30v DC voltage produces only random numbers or 'OL' if manually switched to a lower range.
I am still trying closed box testing at the moment.

[enut11]

The service manual points to an error in the input switching or input amplifier.

Still, why is it behaving perfectly only on 4W ohms??

[The Soulman]

front/rear terminal switch?

[enut11]

I cycled the front/rear input switch and it made no difference.

When I place a short across the input terminals there is no change to the display
eg, DCv is 0.57xxxV open circuit and similar number with short circuit input

[Kleinstein]

One could see if the connection from the input to the 10 M divider for the high votlages is still connected: so use a different meter to measure the input impedance if in 100 V mode.  If present the input switch would likely be OK.

After that it is probably time to open the unit. Not much more to do closed case.

[coromonadalix]

before going too far  you may have some cold sloder joints  etc ... if your meter was doing fine before a shut down : technically it was working fine .......  check every voltage test points, if you have socketed chips, carefully remove the and re seat them  etc ........

I see lots of people going / thinking too far away ....  maybe some capacitors died or were near of death .... some bad connectors contacts 

[enut11]

100v DC range input resistance measures 10Mohm, so OK.

I have run out of ideas so I will need to open it up and do a close inspection, re-setting all plug-in interfaces first. Thanks @coromonadalix.

Thanks for the feedback so far.
enut11

[Kleinstein]

If the 4 wire ohms mode still works, this narrows down the failure quite a bit, to a small part of the input switching.

For the 100 V range this essentially means Q109 not turning on for some reason. So maybe a fault with U101 or the supply to this.
For the 10 V range this would be the path with Q116 and K103. So maybe U102 or the supply to this. The 2 V ref. Level for the comparator could also be such a point that can effect several channels, but maybe not all.

For some reason Q105 and Q115 seem to work for 4 wire ohms.

Checking the supplies is still a good idea, as several switch paths seem to have failed.

[enut11]

Trouble shooting the HP3456A DMM is proving tricky.

All supply voltages are normal
4W ohms is accurate on all ranges

2W ohms is faulty
DCv is faulty
ACv is faulty

Spent some time around the input switching circuit and suspect one of the relays may be faulty.
I have not worked out how to check for noisy relay contacts or noisy switching FETs.
enut11

[vindoline]

enut11, do you have the service manual? It has extensive troubleshooting procedures to help you narrow down the problem. Highly recommended!

[Kleinstein]

The service manual gives a lot of hints and also schematics.

I don't think it is a relay or the front/rear switch, as the 10 M resistance is still there and the 4 wire ohms mode still works.
This really points to some FET (e.g. Q109) are not turning on.  Not turning on is usually a failure on the control side and not the JFET itself. So it's the part around U101 and U102.
With the AC ranges also not working this makes me more suspicious about the 2 V level for the comparators used to switch the FETs, as to high a level there (e.g. near 5 V) could cause several FET switches to stay open.

[enut11]

enut11, do you have the service manual? It has extensive troubleshooting procedures to help you narrow down the problem. Highly recommended!

Yes @vindoline, I have the hires version of the 3456A manual

[enut11]

The service manual gives a lot of hints and also schematics.

I don't think it is a relay or the front/rear switch, as the 10 M resistance is still there and the 4 wire ohms mode still works.
This really points to some FET (e.g. Q109) are not turning on.  Not turning on is usually a failure on the control side and not the JFET itself. So it's the part around U101 and U102.
With the AC ranges also not working this makes me more suspicious about the 2 V level for the comparators used to switch the FETs, as to high a level there (e.g. near 5 V) could cause several FET switches to stay open.

The 2v ref for the comparators checks ok. I will do some more checks around the FETs and ICs.

[enut11]

Also, when I injected a 1v or 10v signal into Hi input terminal, I was able to measure a signal at the node point so at least Q116 is being switched on.

The manual says to check for noise on the signal path but does not mention what levels are significant.

[Kleinstein]

The problem described does not look like a noise problem, more like a broken signal path.
At there is only switching, there should be essentially no extra noise visible with the scope.

Getting the signal past Q116 is odd, as from that point on the signal for 4 wire ohms would take the same path (coming from Q115).

[enut11]

I was able to verify that both Q116 and Q109 switch on correctly in relation to the applied input terminal voltage by measuring the path all the way to the node at R125.
In both cases Q115 was off.

I then went on to tracing the LO circuit from the input terminals.  I injected a DC voltage on the 1v range and found that Q104 was not being switched on by comparator U104.
This traces back to U103, an LSLS02N logic gate, and here I am stuck 

[Kleinstein]

Q104 not turning on makes absolute sense to explain the errors 

U103 should be a normal 74LS02 chip. If not available one should be able to use an 74HCT02 as well.

The relevant pins are 2 and 3 as the input to the NOR gate. If these signals are OK, it would be about replacing the chip.
If in a socket one could try to reseat it - bad contacts can be a problem.

If the signal at pin 2 is stuck high  (the signal at pin 3 should be OK as it is also used in 4 wire Ohms mode) it could still be the LS02 pulling the signal up. A socket with U702 could also have a contact problem.

[enut11]

 Pin 9 U104 needs to be high to turn on FET Q104, and this would come from pin 1 U103 (NOR gate)

I read that for a NOR gate to output a high, both inputs need to be low.

Pin 2 U103 is low so the problem is with pin 3 (currently high) and this goes all the way back to microprocessor pin 32.

I am missing something here but not sure what...

[Kleinstein]

A fault at pin 3 is odd: this line should be working when in 4 wire Ohms mode. This than mean more like a software problem / broken program ROM.

[enut11]

Over the past 2 days I have done extensive probing and signal tracking through the analog circuits all the way to the A/D converter with no luck.
There is still that mysterious 4W circuit behaving.
It is starting to look a bit more serious and beyond my knowledge to troubleshoot.
Haven't given up yet...

[Kleinstein]

Th signal at pin 3 of the NOR gate is also used for the other gate to control the auto zero for 4 wire Ohms. So the Hardware side from the µC must be OK. This would be more like a digital problem, like a slightly corrupted ROM. So maybe check the ROM and digital supply.

I don't think the actual ADC is involved, as the ohms mode works.

[enut11]

@kleinstein, I have no idea how to check a ROM