How to use a pulse signal to detect a feedback's stability

I have heard from some people that we can add a pulse signal to detect the stability of a feedback loop, I want to know how to do this, Where to add the signal , how much the rise time ,fall time, width and magnitude?
Is there any paper talk about it?
3x!

Re: How to use a pulse signal to detect a feedback's stabili

You can inject the pulse in the closed loop and monitor the output. You can then get an idea about the stability of the system based on the overshoot/ undershoot and the amount of ringing you observe.

The amplitude of the pulse should be sufficient for you to make accurate measurements, but not large enough to cause the system to go into saturation, that is, get out of the linear mode. After all, you are trying to determine its stability when in the linear mode.

The pulse should a pproximate an ideal one as closely as possible, so the rise time should be as fast as possible, but in practice you should just use something that is faster than the system you are trying to evaluate. So if your system has a cutoff frequency of say 1kHz, then the rise time of the pulse should be at least 10 times faster, or 35us, because trise=0.35/f , where f is the maximum frequency.
making f ten times the crossover frequency you get: trise=0.35/10*1kHz.

The point of injection can be anywhere in the system, as long as you can observe the effect on the output. Probably the easiest is changing the reference. If that is not possible, then add it in series with the feedback. In series means you affect the feedback voltage, id does not necessarily mean you have to connect something in series with the feedback network.

I am not aware of any paper describing all this.