Oscilating amplifier in multimeter (Philips PM2527)

[TheUnnamedNewbie]

When calibrating a meter I noticed that I was having a lot of difficulty nulling the input amplifier. In order to investigate I attached an oscilloscope, and noticed that the output of the amplifier is oscillating. The thing is that it is just barely unstable - reset the meter and everything seems good for a while until after a significant amount of time (5 minutes to two hours), without any repeatable warning, the output starts oscillating. It always ends up at 18 kHz, 500mV peak-to-peak.

Looking at the schematic, I can see that they already added capacitor C1201 after the first version - in the meter currently on the bench it is bodged on the bottom across the pins, on a second (newer) meter, there is an additional footprint and pcb traces provided for this capacitor.

My initial thought was that maybe the capacitor had failed or at least was degraded in performance, but after replacing it with a new capacitor of similar specifications (I think actually exactly the same capacitor from philips, just newer), the problem was still there.

Ofcourse, I can just slap on some more capacitance in the hope that it further stabilizes the amplifier but it is hard to know how much I should add. In addition, I don't want to alter the circuit too much as I don't want to negatively impact the performance.

In addition, how do I actually know I fixed the problem? If it is unstable it could be it take 5 minutes, 5 hours, 5 years before it actually starts oscillating...

[Marco]

The output impedance of that source follower is less than 1K, so at least it's not surprising the capacitor is useless at 18 kHz.

[TheUnnamedNewbie]

Ok, but do you have any suggestions how to stabilize the amplifier? Or should I just leave it oscillating?

[Kleinstein]

If the amplifier is close to oscillation, one would start to see ringing with a rectangular input signal.

One could try a little bit of capacitance from the OPs output towards the gate of the JFET in feedback - this could be as little as an insulated wire close to the FET.

[TheUnnamedNewbie]

Another thing I found is that I can't get rid of the offset fully when I try to do so while the amplifier hasn't started oscillating yet. Is this LM741 simply out of spec?

[Kleinstein]

The adjustment at the OP is usually only to adjust the OP's own offset. Here most of the offset likely comes from the JFETs.
Usually it is no such a good idea to compensate additional offsets with the offset adjustment of the OP as this adds additional offset drift. It might be better to adjust the JFETs offset by having an additional resistor in parallel to one of the source resistors.

[duak]

What frequency is the oscillation?

I agree with Kleinstein - try a small capacitor of 10 - 50 pF across R1212.

I've attached a balance circuit sketch.  It isn't quite the same as the Philips circuit but I hope you get the idea that balance is adjusted by changing the ratio of the load resistors and drift is adjusted by changing the drain to source voltage.

Cheers,
 

[TheUnnamedNewbie]

What frequency is the oscillation?

18.56 KHz. I'll try adding some capacitance later then, and see if it helps. I can look at trimming the offset of the JFETs by adding some trimming.
Perhaps I'm missing something obvious, but from what I can tell the JFETs are pretty much just forming a low-noise buffer input pair for the opamp - how can I tell the difference between JFET offset and opamp offset in measurement, without taking the entire thing out-of-circuit?

[Marco]

They do nothing for noise, because there is no amplification in the source followers. They are simply there to provide a high impedance input.

[Kleinstein]

In this circuit the JFETs work as buffers (source followers). Even without amplification a buffer can add noise. Here the main noise sources are likely the 560 K resistors.

To adjust the offset of the OP and JFETs separately, one could shorten out the OPs inputs and than adjust the OPs offset (toward the point where the output will change sign. This should to a large extend adjust the OPs offset by its own.
After that the JFETs can be adjusted with something like a resistor in parallel to R1210 or R1211 (depending on the sign). This could be a rather high resistance (e.g. > 1 M) - a series resistor would be possible too.

Another option would be to set the offset pot to about center and than do a first rough adjustment at the FETs (e.g to less than 10 mV) and than do the fine adjust at the OP.

[David Hess]

I would start with a bypass capacitor across R1212.  It's high value combined with the capacitance of the JFET may be enough to cause a problem.

C1201 is largely useless because of the low JFET source resistance.  A pair of resistors could be placed between the JFET sources and operational amplifier inputs to make C1201 more useful however bypassing R1212 is easier to do.  Increasing the value of C1201 would be my last choice.  As shown I might remove it.

The 741 could be replaced with a slower operational amplifier.

Or replace the whole circuit with precision JFET/CMOS operational amplifier or maybe an LM308/LM11 or LT1008/LT1012/LT1097.

[TheUnnamedNewbie]

Or replace the whole circuit with precision JFET/CMOS operational amplifier or maybe an LM308/LM11 or LT1008/LT1012/LT1097.

I think that might be the easiest option. I will see what amplifiers I have laying around.

[David Hess]

Or replace the whole circuit with precision JFET/CMOS operational amplifier or maybe an LM308/LM11 or LT1008/LT1012/LT1097.

I think that might be the easiest option. I will see what amplifiers I have laying around.

I only suggested this for if the simple solutions do not work.  Installing a low input bias current precision amplifier would likely be easier than finding a replacement for the dual JFET.