HP 3478A - unreliable input switching

[dom0]

My 3478A works fine, often, but sometimes displays just an all zero reading.

The service manual suggests that a probable cause is some error in the input switching circuitry. Checking the relais unveiled the error ; in this "all zeroes" condition all input relays are on (approx 5 Volts across the coils). I checked the schematics and saw that the  relay driver H-bridges are inside the input hybrid - Oops!

Out of curiosity I looked at the coil voltages with a oscilloscope and was startled by what I saw (and this is reproducable!): the coil voltages are not DC. Instead, they are switched approx once per second between not-so-high (4 volts) and yep-thats-high (4.8 volts). I have never seen a fault like this before. How can this happen? The 5 V, 15 V and -15 V supply rails are perfectly clean while that is happening. Maybe someone saw something like this before?

To reliably enter this "all zeroes" condition I just apply no input signal and enter auto-range mode in the DCV mode, after a few times switching the relays around it'll settle on the smallest range displaying all zeroes.

Now I have a idea or two on how to fix this. While HP doesn't reveal any details on the hybrids internals, I'm pretty sure that the hybrid is just a shift register to the A/D processor. I came to that conclusion after seeing the inteface between the hybrid and the MCU. If only the relay drivers are somehow broken it should be reasonably easy to put a small MCU on the digital interface and let this small MCU drive the relays instead. However, I haven't yet checked the digital interface with my logic analyzer. Will do that before doing any soldering inside that meter.

[toomas]

I'm a bit surprised and glad to find this topic. I am troubleshooting my HP3478A input hybrid also. The fault that I am getting is that when I turn it on, then everything works fine. All ranges work and accuracy is within specs (triple checked against other known working 6 digit multimeters). Then after a while the value starts to shift, say 3.00006V input will become 2.88...V and then bang, all the ranges (except AC voltage and AC current) will show overload. Then after turning the power off for a while, then sometimes it will become sane again, and sometimes not.

I am suspecting that it is a stability issue, but haven't figured out the cause yet. Maybe even on the digital part on the analogue side. ADC seems to work just fine btw. The reason why I'm suspecting it is that Y463 oscillator amplitude is not perfectly stable although frequency 10.9MHz seems right. Also at some point I was seeing exactly 25ms period square wave coming out  from input hybrid pin 10. Before this issue appeared then something whined briefly inside the unit when ranges switched. I didn't bother much, because everything was working, but it may have been that relays were not getting pure DC maybe similar to your case. Power rails seem to be stable, but -15V is a bit low, it's 14.5V while +15V rail is 15.06V or so.

I'm investigating it further... have you found a solution?

[dom0]

Not yet, haven't had time to check back with the DMM. But it may be wise to do more thorough checking with the signature analyzer, I didn't check any signatures at all, since all my suspicion was directed towards the input circuitry...!
In any case, it's time for some coolant spray I guess...

Funny thing is, I remember once seeing these 2.8... V regardless of input, too. I guess it's a common "residue voltage" (due to parasitics in the design) if the amplifiers' inputs are floating.

Did your unit suffer a high voltage surge on the inputs like mine did?

[toomas]

I haven't abused the input as far as I know, but I don't know what previous owners did to it. I bought it in non working condition, but fortunately there were only two things to deal with. The main big electrolytic cap had dried up and the unit show "uncalibrated" message. Recalibrating 2 voltage ranges and 1 ohms range fixed it. After that I could use it occasionally for half a year.