OPA855: consider the lowest phase shift or only consider the crossover frequency

Amelie Zheng

Part Number: OPA855

Hi Team.

According to customer's simulation, the phase shift of our TIA circuit is not monotonous, that is, before the crossover frequency (Aol*beta = 1), the phase shift changes are not monotonous.

Therefore, although the calculation of the phase margin is greater than 45 degrees, there is a greater phase shift in some areas where Aol*beta>1. When customer consider the ringing design, they should consider the lowest phase shift or only consider the crossover frequency？

Could you kindly give some advice?

over 5 years ago

0 Michael Steffes over 5 years ago

Guru 48395 points

Morning Amelia,

Yes, the typical transimpedance design will show a loop gain phase that dips towards -180deg midrange as the noise gain starts peaking due the feedback resistor and source C. Since that is in a region of high loop gain magnitude, that is not a stability problem and the overall response is more determined by the phase shift when the LG crosses 0dB.

And, for some peculiar english nuance, the word you want is non-monotonic meaning it changes slope in this case. The word monontonous that you used means boring, so perhaps that is also applicable here.

0 Michael Steffes over 5 years ago

Guru 48395 points

Also Amelia, I had this vague feeling I had shown this issue in one of my articles and yes, Figure 11 here shows exactly what you are talking about and why it is not an issue. This is the 3rd example in the article of why you would use a decomp VFA, transimpedance applications

https://www.planetanalog.com/applying-high-speed-de-compensated-vfas-hitting-performance-targets-while-tuning-phase-margin-insight-10/

0 Amelie Zheng over 5 years ago in reply to Michael Steffes

TI__Genius 10865 points

Hi Michael,

Thanks for your reply. The pictures in the link is not very clear and they can't be zoom in.

Do you have the original document about this article?

0 Michael Steffes over 5 years ago in reply to Amelie Zheng

Guru 48395 points

Sure Amelie, that was an issue on Planet Analog when they migrated platforms and lost the links on some of the articles to the enlarged figures.

To summarize though what you are looking at is the LG/(LG+1) term or 1/(1+1/LG)).. If that LG term approaches a -180deg phase shift it can be a problem if the LG is at 1, if the LG is very large, as it is midband in the transimpedance design, if the phase shift were -180deg it looks odd on the loop gain plot, but that is just a small negative number added to 1 in the denominator.

Applying High Speed DeCompensated VFAs July1_2019.docx

Amplifiers

Amplifiers forum

OPA855: consider the lowest phase shift or only consider the crossover frequency