A question about compensation capacitor in the opamp

[bhl777]

Hi all, when I was reading Hans Camenzind's book (http://www.designinganalogchips.com/_count/designinganalogchips.pdf), I have a question about compensation capacitor in the opamp, could anyone take a look of the following question?

In the last paragraph of page 6-12, the author wrote:"It is also important that the capacitor feed back the AC signal from a reasonably low impedance (here the output) to a very high one (the current mirror and the base of Q4) so that we get nearly the full AC voltage swing at this point."

I did not quite understand the last two words "this point",
(1)which point he was mentioning? The low impedance node or the high impedance node?
(2) Could anyone explain why this compensation capacitor has the effect on AC swing?

Thank you!

 

[erikl]

(1) the high impedance node ("this" means the latter one).
Voltage feedback should always occur from a low impedance to a high impedance node.

(2) The compensation capacitor has no effect on AC swing. It serves as Miller compensation cap to separate the poles of the previous and the following stage in order to achieve one dominant pole (could be either of them) - see pole splitting, e.g. Razavi "Design of Analog CMOS Integrated Circuits", Sec. 10.5 , or this IE³ paper: Wing-Hung Ki et al. "Re-examination of Pole Splitting of a Generic Single Stage Amplifier", IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—I: FUNDAMENTAL THEORY AND APPLICATIONS, VOL. 44, NO. 1, JANUARY 1997, pp. 70 ff.

 

[bhl777]

Hi erikl, thank you. I agree with you that Cc has nothing to do with swing. So could you explain why in this book, he said " the capacitor feed back.....so that we get nearly the full AC voltage swing at this point" ?

(1) the high impedance node ("this" means the latter one).
Voltage feedback should always occur from a low impedance to a high impedance node.

(2) The compensation capacitor has no effect on AC swing. It serves as Miller compensation cap to separate the poles of the previous and the following stage in order to achieve one dominant pole (could be either of them) - see pole splitting, e.g. Razavi "Design of Analog CMOS Integrated Circuits", Sec. 10.5 , or this IE³ paper: Wing-Hung Ki et al. "Re-examination of Pole Splitting of a Generic Single Stage Amplifier", IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—I: FUNDAMENTAL THEORY AND APPLICATIONS, VOL. 44, NO. 1, JANUARY 1997, pp. 70 ff.

 

[erikl]

...so that we get nearly the full AC voltage swing at this point" ?

Because a high impedance node doesn't (or just negligibly) load (i.e. burden) the source - the low impedance node. Compare with resistive voltage division!

 

[bhl777]

Hi erikl, I can roughly understand the load doesn't burden the source, but still cannot connect it with swing.

Can I say (1) without this miller cap, the swing at the high impedance node changes and cannot have full AC swing? (2) is it related to that "source" (output stage) has a high AC swing?

Thank you!

Because a high impedance node doesn't (or just negligibly) load (i.e. burden) the source - the low impedance node. Compare with resistive voltage division!

 

[erikl]

Can I say
(1) without this miller cap, the swing at the high impedance node changes and cannot have full AC swing?

Yes.

(2) is it related to that "source" (output stage) has a high AC swing?

It doesn't depend on the value of the AC swing.

 

[bhl777]

Hi erikl, thank you so much for this simulation. I have the last two questions with this.
(1) Your simulation is based on 10MHz source, under this condition, Cap is like short circuit and we know low impedance node and high impedance node both have full AC swing. However, when frequency reduces, for example, to 1kHz, will this statement holds?
(2) Since you say yes to this, "without this miller cap, the swing at the high impedance node changes and cannot have full AC swing", could you explain again why "with this miller cap, the swing at the high impedance node as full AC swing"?
Thank you！

Yes.


It doesn't depend on the value of the AC swing.
View attachment 96517

 

[erikl]

bhl, I think you didn't understand the task of the Miller feedback compensation cap: its task is pole splitting, it is not necessary to reproduce the full swing at the high impedance node! Read and try to understand the pole splitting method, which separates the poles of two successive stages in order to achieve one dominant pole. It is described quite well in the a.m. Razavi textbook, I think.

 

[bhl777]

OK, thank you again. I know the task of the miller cap for pole splitting, but just don't understand why it has something to do with swing. I think you answered my question now, it is not necessary to reproduce the full swing at the high impedance node.

bhl, I think you didn't understand the task of the Miller feedback compensation cap: its task is pole splitting, it is not necessary to reproduce the full swing at the high impedance node! Read and try to understand the pole splitting method, which separates the poles of two successive stages in order to achieve one dominant pole. It is described quite well in the a.m. Razavi textbook, I think.