The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> stability design with stb analysis, pz analysis and Nyquist plot
https://designers-guide.org/forum/YaBB.pl?num=1640645497

Message started by Buyilehu on Dec 27th, 2021, 2:51pm

Title: stability design with stb analysis, pz analysis and Nyquist plot
Post by Buyilehu on Dec 27th, 2021, 2:51pm

Just wondeirng if anyone sees this "strange" behavior:

The bode plot generated from stb analysis on a design (a regulator) shows PM > 85 degrees. But a load dump transient simulation shows there is oscillation at around 1.5MHz.

Then I did a pz analysis. The result shows there is a RHP pole with real part=26KHz, and imaginary part=+/- 1.56MHz. So it roughly matches transient sim result.

But the Nyquist plot generated from the stb result shows no encircelement of (-1,0).

My understandubg is Nyquist is the ultimate tool that should be used to evaluate the stability. Buy why the Nyquist plot here doesn't show problem?

Thanks

Title: Re: stability design with stb analysis, pz analysis and Nyquist plot
Post by Ken Kundert on Dec 27th, 2021, 7:30pm

Each of the methods you describe can be fooled.  Generally, if any indicates instability, then your circuit is unstable.  But each may indicate stability on a circuit that is not stable.

Stb and PZ are fooled by circuits that are small-signal stable but large-signal unstable.  And a circuit may not oscillate in transient if you do not excite the instability.

Stb also requires that you accurately identify the loop by proper placement of the probe element, and it only works for circuits with a single loop.

-Ken

Title: Re: stability design with stb analysis, pz analysis and Nyquist plot
Post by Buyilehu on Dec 28th, 2021, 9:44am

Ken,

Thanks for responding.

I agree. These methods all have their own limitations. In this case tran and pz show the problem but bode and nyquist don't.

Regarding the nyquist plot, as I said it is generated by plotting RealvsImag from stb result. But we know stb uses simple sinusoid (s=jw) and sweep frequency. So seems to me it is not a good idea to generate nyquist plot from stb result since stb doesn't take care complex frequency.  Not sure if my understanding is correct? From my research people use stb to generate nyquist in Cadence. What do i miss?

Title: Re: stability design with stb analysis, pz analysis and Nyquist plot
Post by Ken Kundert on Dec 29th, 2021, 11:08pm

The issue with stb analysis is not that the evaluation frequency is constrained to the jω axis.  Rather there are two limitations with stb analysis.

1. It applied to the linearized circuit, and so can only be used to determine the small-signal stability of the circuit.

PZ analysis has this same limitation.

2. It assumes the circuit has a single loop and you have properly identified the loop by the placement of your probe.

You are probably not getting the results you expect because you have not placed the probe correctly.  Perhaps the circuit has multiple loops, in which case there may exist no proper place to insert the probe.

-Ken

Title: Re: stability design with stb analysis, pz analysis and Nyquist plot
Post by A Kumar R on Dec 30th, 2021, 2:28am

Hi Buyilehu,

Does your circuit show sustained oscillations at around 1.5MHz?. Are they decaying or growing with time?.

Thanks.

Title: Re: stability design with stb analysis, pz analysis and Nyquist plot
Post by Buyilehu on Dec 30th, 2021, 11:48am

Ken,

The iprobe is inserted in the resistor feedback path to the negative terminal of the error opamp. I believe the design has only one loop.

Interesting thing is the oscilaltion frequency is much higher than the unity gain frqeuncy.

Seems to me the best is to evaluate the stability from all avaibale methods.

Thanks

Title: Re: stability design with stb analysis, pz analysis and Nyquist plot
Post by Buyilehu on Dec 30th, 2021, 11:54am

A Kumar R,

The nominal output is 2.7V. The oscillation peak-to-peak voltage is steady at ~ 120mV around the nominal. In theory if there is a RHP pole then the response would be a growing sinusoid. So that's interseting. The simualtion time is 30ms. From what I can tell the oscilaltion is pretty steady.

Thanks

Title: Re: stability design with stb analysis, pz analysis and Nyquist plot
Post by A Kumar R on Dec 30th, 2021, 9:28pm

Hi Buyilehu,

I saw that there are 37 poles listed in the PZ analysis, is this correct?. i don't know your circuit, but 37 poles!...where all they exist?.

Thank you.

Title: Re: stability design with stb analysis, pz analysis and Nyquist plot
Post by Ken Kundert on Dec 30th, 2021, 10:57pm

If you want help with this, you will probably have to give us more information.  The schematic might be helpful.

Title: Re: stability design with stb analysis, pz analysis and Nyquist plot
Post by Buyilehu on Dec 31st, 2021, 10:14am

Thank you all for looking at this.
Happy new year!

The Designer's Guide Community Forum » Powered by YaBB 2.2.2!
YaBB © 2000-2008. All Rights Reserved.