Estimating amplifier transfer function...?

[Limhes]

Right now I am in the middle of an EE subject called Adaptive Systems, employing some algorithms to obtain a valid (usable) model of 'any' (linear) system. For example, one puts a 'persistently exciting' input (enough frequency/amplitude content) into a system and measures its output, trying to estimate the transfer function from it, in terms of bode diagrams, impulse response, pole/zero diagram or whatever.

Has anyone ever tried this to get to a good quality measure of his (her) amp, regarding stability/resonance character? To me, it seemed very valuable and worth a try. But, since it needs some (high quality) equipment, I would like to hear some thoughts on this first...

Shoot!

Link to my course reader: http://www.limhes.net/upload/AS1_05.pdf

 

[darkfenriz]

Square wave response gives you most answers what frequency/phase response is. Common, simple and infinite bandwidth in theory.

 

[Limhes]

The problem with this is, that you want to see the response to all frequencies on just a small time period. You think all useful information is in peaking/ringing of a square wave?

 

[Portlandmike]

Truth is I think most of the interest isn't in the linear transfer function. Per you comment on a square wave excitation, which does tell lots, but not for example loop stability per say ( a conditionally stable loop can have excellent transient response, but that's an exception.)

Much is to be learned about the "linear"system in how it responds to say multitone, and what new tones it produces.

If one could actually make a meter that could tell how good an amp would be percieved, then you'd really be on to something.
To date, some amps with rather poor THD actually sound better to the masses then ones with far better THD. Make a meter for that and write a paper.

mike