Injecting energy into the loop to examine its stability is trustworthy or not?

[zoom8848]

I read from the book "Design with Operational Amplifier and Analog Integrated Circuit_Sergio Franco" that

"One way to assess whether a circuit is stable or not is to inject some energy into one or more of its reactive elements and then observe how the circuit does on its own, in the absence of any applied sources."

I have been taught to use this method to examine my circuit's stability.

Does the results faithfully reflect the circuit's real stability?

If yes, which point to choose to inject energy in?(like capacitive or inductive node, I guess, which can restore energy)
What about the dominant pole, is it necessary?

 

[Warpspeed]

Step (or transient) testing is an excellent way to see any overshoot or ringing.
And in some circuits the results may be quite different in each direction.

What it does not do is tell you what is wrong or how to fix it if the results are unsatisfactory.
It makes a great final dynamic test, once you feel you have obtained sufficient phase and gain margins.

 

[LvW]

If a closed-loop configuration is stable or not can be seen after injecting a short voltage or current puls into a part which is grounded (lift the ground connection and inject the test signal).
If the circuit is stable with a marginally stability margin you will observe a decaying oscilation. In case of instability the circuit will oscillate (rising amplitudesd until clipping) or go immediately into saturation.

However, the claasical procedure is to investigate (simulate/measure) the transfer function of the open loop in order to find the stability margin.

 

[zoom8848]

Thanks for your hint.
I didn't realize the direction problem before!

- - - Updated - - -

Thanks its very helpful.
I used to directly inject into the test node without disconnecting the ground connection and injecting from that node.
I realize your way is more reasonable!

- - - Updated - - -

Is disconnecting the capacitance for main pole compensation more preferable or not?

 

[dick_freebird]

People make devices to do this very thing, for power
supply stability measurements. I'd call it more trustworthy
than any calculation or simulation (which always leave
out anything you forgot to measure or represent).

https://www.ti.com/lit/an/snva364a/snva364a.pdf

 

[zoom8848]

Thanks for your information.

It's valuable, I'll add it to my collection.