The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Cystal osc which satisfies start up condition cannot sustain oscillation
https://designers-guide.org/forum/YaBB.pl?num=1640854729

Message started by unaffected on Dec 30^th, 2021, 12:58am

Title: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Dec 30^th, 2021, 12:58am

I designed a pierce oscillator oscillating at 25MHz, the figure attached is the stb analysis result, the loop gain is 20dB at 0deg.

I tried everything I can do to start the oscillator. I set the maxstep to 10ps or 100ps, set initial condition of Lm to 1uA or 10uA， errpreset=conservative, integration method is trap or traponly. tighten reltol, vabstol, iabstol, using transisent analysis or hb analysis, at t=0+ the oscillations starts, but then its amplitude becomes smaller and smaller, and the oscillation stops.

My question is:
1. From the loop gain and phase plot in the attached figure, I think this oscillator can oscillate at about 25MHz. Am I right?
2. Why this cystal oscillator cannot sustain oscillation? Is there anything I can do to simulate it correctly?

Thank you very much.

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by Ken Kundert on Dec 30^th, 2021, 10:59pm

Seems like you are doing the right things from a simulation perspective. It might be helpful to have the schematic and the transient waveforms.

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Dec 30^th, 2021, 11:23pm

Thank you very much.
I attached the schematic of the oscillator

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Dec 30^th, 2021, 11:30pm

I set reltol = 1e-6, vabstol = 1e-8, iabstol =1e-14, maxstep=10ps, set initial condion of Lm (inductor of crystal) to 10uA, reran the simulation.

please look at the attached transient results(VDD=3.3V).

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by Ken Kundert on Jan 4^th, 2022, 9:40pm

What happens if you keep on simulating? Does the amplitude go to zero, or does it settle to a stable level?

-Kenk

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Jan 4^th, 2022, 10:51pm

The attached file is the transient result of 1ms-2ms, at 2ms the peak to peak amplitude of X2 is 127.6mV.

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Jan 6^th, 2022, 5:21pm

The attached file is the transient result of 2ms-3ms,

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Jan 22^nd, 2022, 5:27am

The data sheet of crystal says the CL=9pF, in former simulation, I add 6pF capacitor to each crystal node(CL=3pF) , and the cystal osc can not sustain oscilation. After I changed the cap at eash cystal node to 17pF, the cystal oscillate oscillate steadily.

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Jan 22^nd, 2022, 5:29am

This is the stb plot of cystal osc with 6pF capacitor at each node

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Jan 22^nd, 2022, 5:32am

This is the stb plot of cystal osc with 17pF capacitor at each node

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Jan 22^nd, 2022, 5:47am

1. In the stb plot of crystal osc with small load cap, the lowest phase is -15 deg.
2. In the stb plot of crystal osc with 17pf cap at eash crystal node, the lowest phase is -53 deg.

As far as I'm concerned，in 1, the lowest phase -15 deg do not provide enough margin，perhaps this is the reason the crystal osc can not sustain oscillation.

I am not sure if my explanation is correct, any comments and suggestion are always welcome.

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by Ken Kundert on Jan 26^th, 2022, 4:55pm

It is an interesting question. I don't know what is going on. Presumably you are running the stability analysis about the DC operating point. I noticed that in the first case you have only 15 degrees of surplus phase shift. It is conceivable that as the signal grows the you lose all of your surplus phase shift and so the Barkhausen criteria cannot be satisfied because the phase shift around the loop cannot be 0 degrees. In your second simulation you have more than 50 degrees of surplus phase shift. Presumably in this case that extra room is sufficient to allow you to reach 0 degrees of phase shift around the loop even with the variation in the effective loop gain as you transiton to the large signal operating point.

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by unaffected on Feb 22^nd, 2022, 12:19am

Thank you. I am running the stability analysis about the DC operating point

Recently, I drew the nyquist diagram of crystal osc with small load cap, nyquist criterion shows this osc can not oscillate.
I also drew the nyquist diagram of crystal osc with 17pf cap at eash crystal node, nyquist criterion shows this osc can oscillate.

The above results are in accordance with the simulation results.

Backhausen criterion is necessary not sufficient condition for oscillation, for some oscillators, nyquist criterion must be used to check the stability.

Title: Re: Cystal osc which satisfies start up condition cannot sustain oscillation
Post by Ken Kundert on Feb 22^nd, 2022, 10:57pm

The stb analysis performs a Nyquist analysis. A Nyquist analysis is very tricky to perform on your own because of loading issues. If you attempted do a Nyquist analysis without an stb analysis, the results are probably incorrect. The stb analysis was created specifically to overcome the near impossibility of doing an accurate Nyquist analysis on non-ideal circuits.

Furthermore, the Barkhausen criterion and the Nyquist criterion are equivalent. The only difference is that when we want the circuit to oscillate we reference the Barkhausen criterion and when we don't want the circuit to oscillate we reference the Nyquist criterion, but they are both the same criterion.

-Ken