Filter-Amplifier Combination Instability

[philwinder]

Hi,
I have a filter-amplifier setup that has to deal with very low
frequencies. 0.1-5 Hz. See attached simulation and schematic.

The problem I have is that the amplifier adds another pole and causes
a 180 degree phase shift right in the middle of my operating range. It
becomes unstable at that frequency because of the gain. I have tried to read a lot about it, but I can't find a definitive answer on how to mitigate it.

How can I compensate for/fix the instability?

Thanks,
Phil

 

[LvW]

Question 1: What is the purpose of the second stage ?
Question 2: What is the indication of instability ? Why do you think the circuit is unstable ? I see no stability problem.

Regards

 

[philwinder]

Question 1: What is the purpose of the second stage ?
Question 2: What is the indication of instability ? Why do you think the circuit is unstable ? I see no stability problem.

1) Gain. Lots of.

2) Experimentation. It is oscillating at the point where the 180 degree phase shift is.

Phil

 

[LvW]

1) Gain. Lots of.
2) Experimentation. It is oscillating at the point where the 180 degree phase shift is.
Phil

1.) If you need only gain, why not use a simple gain stage without any capacitors ?

2.) Although I don´t see the reason for the shape of stage No. 2 (1+bandpass), it is stable and I still see no stability problem. Did you realize already hardware ?
In this case it should be a problem of layout parasitics.

 

[philwinder]

1) Because its single supply, so the signal is biased at 2.5V. Do I need the big cap to AC couple it.

2) Yeah possibly, the input is ultra high input impedance, so it could be coupling back into that. Not sure how though... I presumed it was because of that phase shift.

So just to clarify, it is only unstable due to a 180 degree phase shift when the gain is 1? And it is ok when the gain is above or below 1?

Thanks,
Phil

 

[LvW]

1.) OK, I see. That´s the reason for C1. But for which purpose C4 ?
2.) Decrease the feedback resistor from 1Mohm to 100k and the other resistor to 1kohm.
3.) Don´t worry about any phase shifts. Perhaps you mix something. Such a phase shift is only critical for the loop gain of a closed loop system. Obviously, this is not the case here.

 

[philwinder]

1) C4 is to limit the gain at higher frequencies. It is set so that anything above 10Hz will start to roll off. Essentially another filter.

2) If I did that then I would have to make C1 100uF to have a high pass cut off of 1Hz. Capacitor sizes at these values become pretty large.
So I cannot do this permanently, but I will give it a go.

3) Ah ok. Great.

Cheers,
Phil

 

[LvW]

OK, I realize your problem.
But I can assure you that´s always a problem: Low- or Highpass filters with very low cutoff frequencies normally require rather large R-C combinations.
And this reqires a compromize between R and C . My concern was the 1Mohm feedback which could be one reason for instabilities.
In general, I think that in RC technology there is no real wayout.
Subaudio filters very often, therefore, are realized in gm-C-technique or with switched capacitors.

 

[philwinder]

I have tried altering the value of the resistors to 100k and 1k, and there isn't any difference. The attached image is a screenshot of the oscilloscope just after the filter. If I remove the gain, then it works fine. If I add the gain, I get a signal like the one in the image.

Also, I think that if I reduce the gain to 10, so 100k and 10k, the oscillation goes away. I think so anyway. Tried so many combinations.

Cheers,
Phil

p.s. Oh and I redesigned the test pcb. No ground or other signals around the Hi-Z input, so hopefully, no parasitics.

 

[zorro]

Hi,
are you sure that power supply is well decoupled?
Regards
Z

 

[philwinder]

Its decoupled, how well is well? I have a couple of 10uF's and a couple of 100nF's. Thats usually ok. But I havent got any power supply filtering as such. Straight from the bench power supply.

Phil

 

[LvW]

Hi, philwinder !

Question: Did you check the last gain stage alone - that means without the filter circuit ?

 

[philwinder]

Good point, no I did not. Just the filter and front end. Unfortunately I am going to have to use a signal generator because the incoming signal is contaminated with mains. And I would bet that it will work fine.

Even so, I'll give it a go, thanks.

Phil

 

[LvW]

Hi, PHILWINDER.

As far as I remember - we didn´t speak up to now about the opamp TYPE !
Perhaps you are using a type which is not universal-compensated, which means not stable for low gains ? It just came into my mind..........
Good luck

 

[philwinder]

Hmm, I'm using a TLC27L4 at the moment, but only because I had it in my parts bin. It doesnt mention anything about being compensated, so I presume it is.

And also, only the filter is at a gain of 1, and that seems to be working ok...

Thanks