What's the difference between feedback and compensation?

[BackerShu]

These two concepts confuse me. Sometimes I think they're just the same thing.
For example, the miller compensation in the amplifier,everyone should know that for sure. We say it's a compensation to make the amplifier more stable.
But, I think, the miller capacitor also make a voltage-current feedback in the amplifier. Here feedback and compensation are the same.
Am I right? Hope someone could help me to make clear for these two concepts!
Thanks in advance!

 

[darla1]

hi

feedback is what we are taking from output to cmpare with the refernce signal to generate error, where as compensation is used to improve an undesirable response in feedback control for stable operation.

 

[LvW]

You see a problem where, indeed, there isn´t one .

For example, the miller compensation in the amplifier,everyone should know that for sure. We say it's a compensation to make the amplifier more stable.

Correct. Compensation is the result of a certain action with the aim to alter the frequnecy response and to allow universal feedback.

But, I think, the miller capacitor also make a voltage-current feedback in the amplifier. Here feedback and compensation are the same.

And the introduction of the Miller capacitor is the corresponding action.

In short: Frequency compensation is done by feedback via introducing an extra Miller capacitor.
By the way: Exactly the same is done when an opamp has capacitive feedback . The result is the so called "Miller integrator"

 

[chanakya77]

Hello ,

Generally compensation of Amp will be done for good Phase margin .This good phase margine gives better settiling time when Amp used in feedback.

definetly compensation also a feedback ..but this feedback has to apply only on Amps who are having better phase margin .

Ex : In Two stage Amps ,generally miller cap will be placed between First stage output & SECOND STAGE OUTPUT .Since single stage Amp is always stable , safely you can apply feedback around this .

If you see three stage Amps , just placeing miller cap between First stage output & Third stage output can't make sure Amp stable because miller cap has been placed between uncompensated Amp [ If you see second & thrd as a Amp] . In this case first you have to compensate inner Amp [Which is combination of second & Thrid] .In this case ,you have to place a Miller cap between Second stage output and third satge output. This is excatly called Nested miller compensation for higher order Amps.

Just summirizing , If you want to put Amp into Feedback that amp should comepensated first either by miller Cap or any other available 1001 ways. If you want to use feedback as a compensation , that loop [where you are palceing compansation Cap] should be compensated first .

thanks

 

[BackerShu]

Hi everyone. Thank you for your reply!

So I can get the difference between the two concepts. Compensation is a step to make sure the amp will be stable in the close system(of course with feedback).We have many ways to do compensation,and feedback is only one of them.

 

[LvW]

Yes, that´s OK - as far as internal feedback is concerned !

 

[hdmi]

I have a stupid question here.

Isn't the feedback responsable for all the infamous instability and oscillation in the Op Amp, so where is the intrinsic feedback path in the Op Amp ckt, which then requires the compensation ckt to stablize it?

 

[LvW]

Isn't the feedback responsable for all the infamous instability and oscillation in the Op Amp, so where is the intrinsic feedback path in the Op Amp ckt, which then requires the compensation ckt to stablize it?

The opamp alone (without external feedback) is always stable. There is no "intrinsic" feedback destabilizing the IC.
External feedback is necessary for suiting the opamp to specific tasks (finite gain amp, adding circuit, diff. amp,...). However, as a consequence, the whole circuitry may be instable and therefore some kind of compensation of frequency response is necessary.

 

[hdmi]

When we are adding the compensation capacitor, do we always assume the worst case scenario, i.e., beta = 1?

 

[carl_chao]

That's exactly part of the theory of Miller compensation.

With the Miller cap, the dominant pole (at 1st stage output) is slowed down because of the compensation capacitor (~Cc*A2, where A2 is the second stage gain), while the secondary pole (at the 2nd stage output) is pushed out due to the feedback through the Miller cap: at the 2nd pole frequency, the Miller cap looks like a short, so the output device behaves like a diode. Looking back into the output stage, its output impedance is ~1/gm2 at this frequency.

 

[LvW]

When we are adding the compensation capacitor, do we always assume the worst case scenario, i.e., beta = 1?

If a compensation cap is added by you externally (which means: by the user), you always have the choice to partly compensate or to make an universal compensation:
partly: stable not for all gains (only above a certain lower limit)
universal: stable for all gains.