high frequency compensation implementation question

[yefj]

Hello,I have A YIG oscilator whose being corrected using a servo DC amplifier. i have built a two stage amplifier as shown below with 5MHz BW. but i am not sure regarding the compenstion part In the text bellow the functionality is described. What is high frequency compensation? what is compensation of high roll of responce? If some one could explain to me what is that in simple words?
Thanks.

 

[danadakk]

The basic idea is, for a single stage, inverting, that a loop with fdbk does not exhibit a
phase shift of 180 degrees when there is still G in the loop, otherwise it will oscillate.
Normally one designs this problem by adding a zero (versus a pole) to lift the phase
before G falls below 1. Normally one shoots for a 40 - 70 degree phase margin
when G = 1, eg. a phase of < 110 - 140 degrees @ G = 1.

https://www.analog.com/en/analog-dialogue/articles/techniques-to-avoid-instability-capacitive-loading.html

https://www.analog.com/media/en/training-seminars/tutorials/MT-059.pdf

https://www.analog.com/media/en/technical-documentation/application-notes/an-257.pdf

https://www.analog.com/media/en/technical-documentation/application-notes/an148fa.pdf

Regards, Dana.

 

[yefj]

Hello Danadakk, when we have 0dB and phase -180 the transfer function .
As you can see bellow i got 4dB at 180 degrees.

I do have 0dB and 180 phase at 20MHz, is that bad as shown below?

they say to "roll off the volatge tuning response of the YIG oscilator beyong with 3dB at 2.5MHz"
what is the link between tuning responce of the YIG with the phase marging of the DC amplifier which is a feedback to the YIG oscilator?

 

[danadakk]

Thinking out loud :

A single pole at f >> fp produces 90 degree phase shift. Most fast
OpAmp are 2 pole, so can produce up to 180 phase shift. Now if
this fdbk produces 180, and given opamp inverting = 180 degrees,
then we have a possibility for unstable in that single stage.

Think about walking the loop. First we go thru the first amp, it having
fdbk, and a phase shift then we go thru second amp, more phase shift.

Then in your circuit you have a 0 phase shift element from output all the
way back to input, a fractional G associate with it. But it is positive fdbk.
So if we ground the input, no signal contribution, and walk the loop, and
we find we have 360 degree phase shift and a loop G exceeding 1 at some
frequency we have built an oscillator.

Did you do the bode plot breaking the +fdbk loop, from the non inverting
input to the output of both stages ?


Regards, Dana.