Help with repair of Solartron 7150

[giosif]

Hello,

I am troubleshooting a Solartron 7150 bench meter which constantly displays OL for measurements, at least in DCV and ACV modes.
I have traced the problem to be that the non-inverting input of op-amp IC1 is being fed +3.5V or thereabout, with nothing connected at the inputs of the meter.
This +3.5V is outside the -2.6 - +2.6V range IC1 is supposed to see (if I'm reading the service manual correctly).
Now, the source of that +3.5V seems to be N-channel JFET TR5 (I think it is an N-channel, although I could not find a datasheet for P/N WN1001 to confirm).
With the meter in DCV mode, when TR5 is supposed to be "on" (i.e. conducting between source and drain), its gate is driven with approximately +15V.
To me, that +15V doesn't look right but, then, tracing the source, it comes from IC14 (an LM339 voltage comparator), and that part of the circuit seems to work as intended.

Am I missing something here?
Is TR5 supposed to be ok with +15V on its gate and not leak any voltage into its source or drain?

[Kleinstein]

The way it is drawn and the circuit around it indeed looks like 2 channel JFETs as shown by the symbol.

In this case the gate drive should not be +15 V but only as much as the signal to switch. If off the gate should be at some -15 V or similar voltage more negative the signals to switch.
So check IC14 and the drive signal (to the right of the part shown).

[giosif]

Thanks for confirming my thoughts!

About the +15V and IC14, this is what confuses me: I think it is working correctly.
If you look at the schematics, the output pin 14 on IC14 (i.e. the one driving the gate for TR5) is dependent on the values at inverting input pin 8 and non-inverting input pin 9.
And the non-inverting input pin 9 is getting a constant value of about 1.3V from the diode series to the right of the schematics I provided (what is not shown in the attached schematics is what the other end of R13 is connected to - i.e. the +15V rail).
So, this value is correct and constant.
The only thing which would change here is the value on inverting input pin 8, which is driven by IC103, the "floating logic MPU".
Accordingly, that pin can be either 0V or 5V, sitting at 0V when in DCV mode (with a short exception every 10 seconds, when it goes to 5V, for "drift correction", as the manual calls it).
Given all this, the output pin 14 can only sit at +15V, as per the normal operation of the voltage comparator.
And, just to be 100% sure, I've replaced the original IC14 with a new LM339N and the situation is the same.

I am starting to wonder whether IC14 should instead be supplied with 0V on pin 3, rather than +15V.
But, then, this meter would never have worked since coming out of the factory, which I doubt was the case.

[giosif]

This is interesting: looking at the schematics for the 7150 plus model (an improved version of the 7150), they've added a 1MOhm resistor between the gate of TR5 and the output pin 14 on IC14.

[Alex Nikitin]

LM339 has open collector outputs, unless it is faulty (an internal short to + supply etc) it can not produce a positive output voltage on its own. So this LM339 is your primary suspect. I've noticed you have replaced IC14 already. In that case you need to find the source for +15V on the JFET, as the comparator can not provide it!

Cheers

Alex

[Swake]

Or the new opamp died already with a short to +15V too.
Power down the DVM and measure the resistance between the +15V pin an pin 14. If it is near zero ohm, then I sugest to remove the Ic14 and check again for a short the IC and on the circuit to determine if the short is in IC14 or elseware.

[giosif]

I was not aware of the output characteristics of LM339, but you are right and the source of the +15V is not pin 14 on IC14 (and IC14 is not shorted).
Thank you for that important detail!
With this new information, I was able to trace the source of the +15V to IC8 (specifically, IC8b).
Strangely enough, I had some suspicions about this IC at some point in the troubleshooting, so I desoldered it from the circuit.
I then tested it, but it seemed to work fine. 
In any case, I now need to source an OP14 dual op-amp or equivalent.

[Alex Nikitin]

In any case, I now need to source an OP14 dual op-amp or equivalent.

I think LT1013 should be a close enough equivalent.

Cheers

Alex

[shakalnokturn]

Any chance IC2 was providing the wrong input to IC8b ?

[Andy Watson]

If TR5 is a JFET, as drawn, the +15V on its gate should be transfered to its source via the junction in the FET - i.e. it must be dead.

[ch_scr]

In any case, I now need to source an OP14 dual op-amp or equivalent.

I thing LT1013 should be a close enough equivalent.

Cheers

Alex

OP200 is close to a like-for-like replacement/slight upgrade to OP14 (the Analog Devices website suggest it as an non-EOL upgrade as well!).

I have a meter with both OP14 replaced to OP200 and no discernible difference. Also all DG211 were replaced with DB211B with no discernible difference.
This was tested by comparing to a second unmodified meter, watched warm-up drift (for zero and with an ext. voltage source) and reading noise. Had hoped for even a (slight) improvement but nothing apparent.
For the latest test I replaced IC11 (74LS175) and IC17 (74LS00) with the corresponding 74ALS types, and IC102 (74LS04) with a 74F04 (74ALS04 should work as well).
It seems to me the reading is a bit less noisy now, but I have to take a datalog to be sure. What did not work was to replace IC102 with either 74HCT04 or 74ACT04 - results in the meter stuck in a reset loop.

With a positive voltage on the gate of TR5 (N-Ch J-Fet, I'd agree) there might be "high" leakage into the high impedance input of IC1, explaining the OL Voltage. TR5 might as well turn out damaged or degraded (?) as a result.
I'd try a "InterFET PN4118A" as replacement (does the pinout fit?), with emphasis on try though.

[Kleinstein]

The IC8b part is non critical, but the IC8a part is the main OP-amp in the integrator. So there is could help to have a good OP-amp.

The OP200 (or AD8622) would be an upgrade from the noise side, but slower, which may be an issue with the ADC. Another alternative could be OPA2202 or ADA4077 that are a little faster and also lower noise, but would need a SMD to DIP adapter.

[ch_scr]

I took some scopeshots to make sense of the explanation in the manual regarding the "forcing waveform". The trace colors correspond to the same color markings in the schematic.
Maybe it helps to decide how fast the integrator ought to be. This is prior to modifications IIRC.

[Kleinstein]

The short pulses of the green waveform can be down to some 50 µs, maybe a little shorter with a different input signal, but likely not much.
The integrator should settle well in this time.

Another point with the speed is that the integrator is using a single OP and the input voltage is thus depending on the OP output slew rate. It should be  about I / (C * 2*pi * GBW). With the forcing and FB signal there may be nonlinear effects, depending on how much the 2 are correlated. So a bit more BW would definitely not hurt and less may be an issue.

P.S.:
The length of the shart phases is probably OK for settling.

The tricky part may be the nonlinear part. The forcing signal can be some +-100 µA and with 1 MHz GBW for the OP-amp this would result in some 1 mV at the integrator. This extra voltage can make a difference in the short times with both references active, which make up about 7% of the time. Depending on how much of the forcing signal is active for these phases the error could be up to 1 mV/3.1 V * 7 % or some 20 ppm. This would be an upper limit for the INL effect and there is likely not that much variation of the forcing signal during the transition phases. The expected INL effect may be more on the order of 10% of this upper limit.  A slower OP-amp may still not be such a good idea.

[giosif]

Thank you all for your input!

I took OP14 out of circuit and tested it in a unity gain setup, feeding a 10Hz, -5V to +5V square wave at the non-inverting input pins.
The power supplies where +15V and -15V.
And the results are interesting:
    - Half of the IC (i.e. the one corresponding to IC8a in the schematics) is mostly working, although with quite a bit of noise introduced at the output.
    - The other half (i.e. IC8b in the schematics, where we have been investigating our +15V source issue), it has an interesting behaviour - see attached picture (please ignore the slight offset for 0V for each trace; that's just my scope where I cannot run a full SPC).
      That is an interesting failure: when the non-inverting input goes negative, the output & inverting input go positive, to almost +15V.
      I've just realized something: could this be a short to the positive rail when the non-inverting input goes negative?
Anyway, this confirms OP14 is toast, so I went ahead and ordered a replacement OP200 (thanks, ch_scr), as it was also more easily available to me and at a decent price.
I will keep LT1013, OPA2202 and ADA4077 in mind (thanks, Alex Nikitin and Kleinstein), in case OP200 turns out not to be a good replacement.

And since I was investigating that area of the circuit, I decided to remove and test IC2 - a DG211 quad analog switch IC.
The IC turned out to be ok for 3 out of the 4 switches - the 3rd switch was constantly off, though, no matter the input.
However, that would have had no bearing on our initial problem.
Still, I've ordered a replacement DG211.

shakalnokturn
I considered the possibility of IC2 leaking via its inverting input pin, but it looks like OP14 was the main culprit, as per above.

Andy Watson
I suspected TR5 might be bad but, based on some initial tests, it seems to still be ok.
However, based on the experience with OP14, the reliability of my initial tests doesn't look too good at the moment. 
I'll check it again, just to be sure.
If it is indeed dead, it's going to be a pain to find a suitable replacement.
ch_scr is suggesting I try a PN4118A as substitute but, if I read the datasheet for it correctly, the ON resistance is in the kOhms, while the ON resistance for the WN1001 in the same region is in the low hundred Ohms.
Also, the 7150 is actually using PN4118A for TR7, while using WN1001 everywhere else, so I guess there was a reason for them to use different parts.

I will provide an update once I've received and installed the replacement parts.

[giosif]

The replacement parts have arrived and, after installing OP200 as IC8, the 15V on the gate of TR5 is gone.   
I've also installed a new DG211 IC in place of the partially defective one and, doing some initial tests, it looks like the meter is fully working now. 
This also means that TR5 has somehow survived.

Thank you all for the help getting this baby back to health!