Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
Your system is not stable if you define stability as havining a minimum 45 Deg Phase margin.
Your graph is a great example of where Cadence's phase-margin gain-margin reporting doesn't work. Taking the phase at gain=0 or gain at phase=0 is not enough. You have a wide range of roughly 15 dB gain where the phase is only -30 Deg! The amount of gain combined with the sub-45 Deg margin will result in oscillation.
A better phase margin algorithm is to report the worst case phase up to the frequency where gain=0, not just at the frequency where gain=0.
Your system is not stable if you define stability as havining a minimum 45 Deg Phase margin.
Your graph is a great example of where Cadence's phase-margin gain-margin reporting doesn't work. Taking the phase at gain=0 or gain at phase=0 is not enough. You have a wide range of roughly 15 dB gain where the phase is only -30 Deg! The amount of gain combined with the sub-45 Deg margin will result in oscillation.
A better phase margin algorithm is to report the worst case phase up to the frequency where gain=0, not just at the frequency where gain=0.
We should know if the question was intended in a general and theoretical aspect or if it was in a particular and practical one.
In the first case we should analize it comparing the gain for phase -180 and the phase for gain =0, In the second one the particular situation should be stated.
why 15dB gain with -45 Deg phase will cause oscillation?
gszczesz said:
Your system is not stable if you define stability as havining a minimum 45 Deg Phase margin.
Your graph is a great example of where Cadence's phase-margin gain-margin reporting doesn't work. Taking the phase at gain=0 or gain at phase=0 is not enough. You have a wide range of roughly 15 dB gain where the phase is only -30 Deg! The amount of gain combined with the sub-45 Deg margin will result in oscillation.
A better phase margin algorithm is to report the worst case phase up to the frequency where gain=0, not just at the frequency where gain=0.
For a simple double-pole loop response, when the phase margin drops below 45 degress then the amplifier tends to ring when settling. The larger the gain, the longer the ringing. At 60 degrees phase margin, you get optimal settling time with no ringing.
So although your system needs 0 degrees phase margin to oscillate, you can get ringing that is so long that it's virtually oscilalting with small phase margins and large gains. Secondly, once the ringing reaches large enough amplitudes that it starts to saturate the amplifier, the non-linear effect will cause frequency shifting and create a permanent oscillation.
I understand that Phase Margin should over 45 Deg (> 60 is better) to avoid oscillation. But I think in the range gain= 15dB, the phase = -30, does it mean the phase margin = 150? So the system is very stable, right?
gszczesz said:
For a simple double-pole loop response, when the phase margin drops below 45 degress then the amplifier tends to ring when settling. The larger the gain, the longer the ringing. At 60 degrees phase margin, you get optimal settling time with no ringing.
So although your system needs 0 degrees phase margin to oscillate, you can get ringing that is so long that it's virtually oscilalting with small phase margins and large gains. Secondly, once the ringing reaches large enough amplitudes that it starts to saturate the amplifier, the non-linear effect will cause frequency shifting and create a permanent oscillation.
My appologies, I'm used to reading Spectre STB analyses where 0 degress (or 360) is instability. I guess you're giving the bode-plot of a negative-feedback system before the feedback????? Normally, if you break the loop, when the signal comes back in-phase you're in trouble. Thus 0 degrees is oscillation. I trust it then that that's not what you did????
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
