The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> The oscillation of one high-pass filter and its loop gain
https://designers-guide.org/forum/YaBB.pl?num=1364653420

Message started by ywguo on Mar 30^th, 2013, 7:23am

Title: The oscillation of one high-pass filter and its loop gain
Post by ywguo on Mar 30^th, 2013, 7:23am

Hi Guys,

My high-pass filter begins to oscillate, but I cannot explain it in the Bode plot. May I have other simulation tool to find why it becomes non-stable.

The following picture is the output of the high-pass filter where it begins to oscillate.

Any comments are appreciated.
Yawei

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on Mar 30^th, 2013, 7:26am

This is the loop gain acquired with stb analysis.

The phase margin is 58 degree and the unity gain frequency is more than 20MHz. The oscillation frequency is ~100MHz.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by raja.cedt on Mar 30^th, 2013, 7:46am

Hello Ywguo,
You can't judge stability from bode, since you have non-monotonic Phase plot , use Nyquist to judge. Looks like you have some encirclements. Why don't you show schematic or try to plot Nyquist in cadence?

Regarding Oscillation Frequency i am not sure why it depends on UGB. May be Frank Wiedmann can explain better.

Thanks,
Raj.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on Mar 30^th, 2013, 8:26pm

Hi Raj,

Thanks very much. I understand Bode plot is not enough for my case because the gain does not decreases monotonically. Plotting Nyquist is what I need. But I don't know how to plot it with Cadence tool.

Best Regards,
Yawei

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by raja.cedt on Mar 31^st, 2013, 2:45am

hello,
plot real vs imaginary of the loop gain, this will give you half of the Nyquist plot means for +ve frequency's. Just imagine other symmetric half if you want total Nyquist plot better export real and imag data into matlab and do some processing(but it is bit tricky).

But looks like you have serious problem with small signal approximation, because your oscillation magnitude is around 1V which can create large shift in the lop dynamics.

Thanks,
Raj.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on Apr 2^nd, 2013, 5:47am

Hi Raj,

The oscillation magnitude is around 1 V because that is half of the opamp output. I have a divide-by-2 with resistors at the opamp output.

I read the section about Nyquist Stability Criterion in Oppenheim's book, Signals and Systems, Second Edition. It seems that I need to figure out if there is RHP poles of G(s)H(s). It states that the net number of counterclockwise encirclements of the point -1/K by the Nyquist plot of G(jw)H(jw) must equal the number of right-half-plane poles of G(s)H(s).

What can I do if it is hard to identify an analytical expression of the transfer function?

Best Regards,
Yawei

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by raja.cedt on Apr 2^nd, 2013, 8:20am

dear Ywguo,
i think you don't need to calculate exact TF, very hard job for a transistor based problem. Nyquist criterion says

clockwise encirclements= closed loop RHP- openloop RHP. So please plot nyquist plot to check how many encirclements, based on your schematic you may guess open loop RHP, so now you can estimate stability.

Thanks,
Raj.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on Apr 2^nd, 2013, 11:03pm

Hi Raj,

I run stb analysis and plot imaginary/real of the loop gain in the following picture. It encircles around (-1, j0) in clockwise direction. That is a regulator, stable. I do not see any RHP pole in Bode plot. What's wrong with my Nyquist plot?

Thanks
Yawei

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on Apr 2^nd, 2013, 11:13pm

I zoom in the picture to show the plot near origin.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on Apr 2^nd, 2013, 11:16pm

This is Bode plot.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by raja.cedt on Apr 3^rd, 2013, 1:11am

hello ywguo,
you have one clockwise encirclement and there is no RHP in the open loop means 1 RHP in the closed loop(please refer my previous post).

Thanks,
Raj.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by wave on Apr 3^rd, 2013, 6:20pm

Can you post a schematic and show where you break the loop for STB probe?
Perhaps you are not including all the FB paths.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by Aman on Apr 10^th, 2013, 2:08pm

Post the schematic. Seems like you have complex poles after the zero, there is a 180 phase shift around 40MHz.
From the looks of it this happens well below the 0dB gain, so system should be stable unless your not breaking
the loop correctly(you definitely have multiple loops or resonance happening).

Regards,
Aman

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on Apr 21^st, 2013, 6:53pm

Hi Guys,

I am sorry to reply late. Here is the test bench. I ran stb analysis to plot Nyquist plot in the previous posts. I am sure this circuitry is stable. But why does its Nyquist plot look like there is one RHP pole in the close loop response?

Thank you.
Yawei

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by raja.cedt on Apr 21^st, 2013, 10:32pm

hello Ywaei,
this circuit is more or less a voltage regulator, why do you say it is high-pass, your transient results shows oscillations and Nyquist confirms it by having 1 closed RHP.

Thanks,
Raj.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on Apr 24^th, 2013, 8:08pm

Hi Raj,

I am sorry that I make this thread confused you. I tried to analyzie the stability of that high-pass filter with Nyquist plot. I am not familiar with this method. So Nyquist plot of the regulator is plotted for comparison. It would be clear if I had started a new topic about the Nyquist plot of the regulator.

Best Regards,
Yawei

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by Pravesh Kumar Saini on May 27^th, 2014, 4:43pm

Hi
1. I just want to understand in conclusion, was the ckt. with above bode plot and nyquist plot is unstable? I am having the similar non-monotonic loopgain. My ckt. is second order MFB biquad LPF.
2. Does the placement of spectre iprobe at any point matter in loop.
3. Can we have different stability conclusion if loop is low pass and not high pass as we discussed in the case presented.
Regards,
Pravesh

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on May 27^th, 2014, 7:05pm

Hi Pravesh,

I need to clarify that the first and second post in this thread is the simulation result of one high-pass filter. The other plots are results of one regulator. The regulator is stable.

Of course the placement of spectre iprobe matters. It must in a point that shared by all loops in the circuit.

Best Regards,
Yawei

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by Pravesh Kumar Saini on May 27^th, 2014, 9:29pm

Hi Yawei,
Thankyou very much for your prompt reply.
Now It seems that I understand your post and issue that you are raising in Nyquist plot for a stable regulator. In my MFB biquad filter I do have the nyquist plot encircling the (-1,0) point once in clockwise. But my bode plot which is having non-monotonic phase response in which phase rises before 0dB shows ~90degree PM at 0dB point and I verified the same phase margin using step change on supply voltage. But I worried about stability because of my nyquist plot and non-monotonic phase response, what's your opinion here.

Title: Re: The oscillation of one high-pass filter and its loop gain
Post by ywguo on May 28^th, 2014, 7:41pm

Hi Pravesh,

I did try to study Nyquist plot, but I have never analyze stability of my circuits successfully with Nyquist plot. :( So it's hard for me to say that is enough for your MFB biquad filter to be stable.

One more thing is that I check the stability with a step change at input, NOT on supply voltage.

Best Regards,
Yawei