How can I determine the frequency stability in spectre simulator? (2stage OP-AMP)

[dressler6]

When evaluating the frequency stability of a two-stage OP-AMP using a Miller capacitor, 'stb' simulation was used for the input stage.

Fig1 is an example of the conventional method of determining the loop gain and phase by the 'stb' simulation.

Since the 'stb' instance can be connected anywhere in the loop of Fig1, 'stb' is connected as shown in Fig.2.
However, the loop gain and phase results of Fig. 1 and Fig. 2 are different.
Which one is right? (I think both are wrong)

When a pulse noise voltage is applied to the output as shown in Fig. 2, the fast loop response via the miller capacitor and the conventional slow loop response are superposioned. However, Fig.2 does not include the GREEN loop response.

I wonder how to determine the exact frequency stability of this circuit. How can I determine the frequency stability in spectre simulator? (Maybe 'GNT' simulation in spectre should be used? If so, where and how to use it, please give an example or reference article, thank you in advance.)

c.f] https://sites.google.com/site/frankwiedmann/loopgain

 

[FvM]

The second diagram places stb incorrectly, because the loop is split into two parallel paths in this point. The first gives a useful loop gain result.

Feedback systems can be designed with multiple cascaded feedback loops. In some cases, it's not possible to identify an overall loop with an insertion point for a loop gain probe. But for the shown system, it's possible I think. There are old Edaboard contributions by user LvW discussing multi-loop feedback systems.

The term "frequency stability" applies to oscillators rather than loop gain analysis.

 

[dressler6]

The second diagram places stb incorrectly, because the loop is split into two parallel paths in this point. The first gives a useful loop gain result.

Feedback systems can be designed with multiple cascaded feedback loops. In some cases, it's not possible to identify an overall loop with an insertion point for a loop gain probe. But for the shown system, it's possible I think. There are old Edaboard contributions by user LvW discussing multi-loop feedback systems.

The term "frequency stability" applies to oscillators rather than loop gain analysis.

Your answer is right.
I absolutely agree your opion.
Thank you very much^^

But i mean that method of Fig.1 can not reflect fast response of GREEN path of Fig.2.
So how can i evaluate include this phenonmenon.

As you told me, I search LvW's article.
Thank you very much. FvM^^