Fluke 8800A increasing offset

[hajo0812]

Hello all, thank you for reading this. Possibly you can help me with your experience.

I have an old Fluke 8800A multimeter on my bench, with a somewhat complex fault in the DC area.
When in 200mV, 2V and 20V DC ranges, without anything connected, the offset slowly but steadily decreases, in the 20V range down to -4,5 V.
As soon as I short the input, it displays 0V as expected.
Also, switching to the 200V and 1200V DC ranges, without anything connected, it returns to 0V immediately.
Measuring inside, I saw that the voltages supplying U1 and U2 are largely offset, not symmetrical any longer.
When the 200V and 1200V ranges are selected, the relay K5 on the left connects the input to ground via 100kOhm, this is sufficient to bring everything back into balance. Even measuring voltage on the base of Q10/cathode of CR6 with another DVM (my good old HP466A) with its 10M input resistance is sufficient to have it back to working temporarily.
The 8800A has >1G Ohm input resistance in the lower ranges, this is probably what makes it so sensitive.
The troubleshooting section of the service manual mentions to shorten TP4 so that the bootstrap amplifier (U2/Q16) do not drive the DC buffer out of range. Doing that, the offset remains stable to some mV only.
I checked/replaced CR32/CR33 and Q16 so far, but no change.
Zeners CR11 and CR12 show a little deviation (one 5,9 V, the second 5,7 V). I don't know if this is important here. Have ordered replacements for them. My current thought is probably to replace U2, but that is an old OpAmp which is rare to get.
Possibly you have some experience with this kind of fault?
Thanks a lot from snowy Austria,
Hajo

[Alex Nikitin]

Hi Hajo, and welcome to the forum!

What you see is not a fault but a typical behaviour of a meter with a very high input impedance. There is always a small leakage current which would charge the meter's input capacitance and that appears as a changing offset voltage. If that bias current is below the the value shown in the meter specification there is nothing to worry about, it is perfectly normal. You can measure the input bias current by connecting a large resistor (i.e. 1 Gohm) across the inputs (preferably shielded) and measuring the offset voltage on that resistor - 1mV will be the equivalent of 1pA input current. The Fluke 8800 manual specifies the bias current below 15pA so you should have less than 15mV offset with 1 Gohm resistor connected across the input. A larger resistor will give you a better accuracy: 10 GOhm will produce 150mV for 15pA bias.

Cheers

Alex

P.S. - there is a bias current adjustment in the input circuit of 8800, you may try to null the current by the slowest drift on an open input.

[hajo0812]

Hi Alex,

thank you for your fast reply and the explanation. Makes sense of course!

It simply made me nervous seeing a -4,5 V indication on a totally unconnected DVM...  .

I will try the bias adjustment. Until now, I did not touch any trimpots.

Best Regards,
Hajo

[Alex Nikitin]

Hi Alex,

thank you for your fast reply and the explanation. Makes sense of course!

It simply made me nervous seeing a -4,5 V indication on a totally unconnected DVM...  .

I will try the bias adjustment. Until now, I did not touch any trimpots.

Best Regards,
Hajo

Hi Hajo,

First things first - measure the bias current - or at least check the speed with which the voltage is changing on an open input, the filter capacitor on the input is 22nF, dU/dt = I/C , so for less than 15pA bias the input voltage change rate should be less than 0.7mV/sec = 7mV/10sec  = 70mV/100s .

Only if the input current is not in spec, try the adjustment.

Cheers

Alex

[hajo0812]

Hi, I just wanted to mention that after adjusting the bias the behavior has improved significantly.
The manual describes the adjustment procedure using a 1 Meg resistor, so fairly easy to do.
Thank you again and best regards,
Hajo