High performance feedback for "Lowtech" amplifiers

[superR]

I had this idea. Maybe I'm reinventing the wheel or saying foolish things but here it goes;

What if I use a fast high performance opamp in a parallel feedback loop to counter harmonics/mains hum? It's difficult to explain but here goes my attempt;

Take an amplfier.
Measure/take the output, use a voltage divider to lower the voltage and send that into a fast, high performance opamp. In the opamp, compare this output to the source signal. Make the opamp send the difference added with the original signal to the amplifier. This way you can use the specs of a low-voltage high performance opamp and use it in your "high" voltage stage.

Is this used "all over the world" "nothing new here" or have I been in the sun for too long and becoming delusional?

[Ouroboros]

You've just re-invented "feed-forward".

[superR]

Then I wouldn't be using the measured output

[Ouroboros]

It's still a variant of feed-forward, as you are adding a correction signal into the input of the PA stage, where the correction signal is the error between the signal source and the PA output (scaled to suit).

[DF96]

If you are comparing the output with the input then that is feedback. Using two parallel different amplifiers for part of the forward chain, in an attempt to get the best of both, is not new.

Your particular version of this may have stability problems, as parallel paths with high gain can produce more phase shift than a single path.

[superR]

@Ourobos
Feedforward would be like this;
System had fransferfunction G. The controller would have transferfucntion 1/G. That would make the overall transferfunction 1. In other words, the reference and output would be identical.
@DF96
What if my amplfier/plant is stable and the controller is stable, then the overall amp should be stable too but I share your thoughts on this.

I am still looking for examples in which for instance a LM3875 is used as a high gain OL output stage and in which a high performance opamp is used for feedback. Can't find it (yet)

[Ouroboros]

Sorry, I should have said that you had just re-invented 'error correction' rather than feedforward!

[DF96]

Originally Posted by superR

What if my amplfier/plant is stable and the controller is stable, then the overall amp should be stable too

If only servo loops were that simple! An open-loop amplifier is (usually) stable. A pair of resistors is stable. Put them together and you have a loop which might not be stable.

I think you will find that this method is rarely used because it is difficult to get it right, so people use simpler methods.

[jan.didden]

Originally Posted by superR

@Ourobos
Feedforward would be like this;
System had fransferfunction G. The controller would have transferfucntion 1/G. That would make the overall transferfunction 1. In other words, the reference and output would be identical.
@DF96
What if my amplfier/plant is stable and the controller is stable, then the overall amp should be stable too but I share your thoughts on this.

I am still looking for examples in which for instance a LM3875 is used as a high gain OL output stage and in which a high performance opamp is used for feedback. Can't find it (yet)

Google 'paX amplifier'.

jan didden

[jcx]

diyAudio - Search Results

will get you to some of my posts on Black's Feedforward - aka Quad's Current Dumping

which seems to be close to your idea, http://digilander.libero.it/paeng/fe...d_concepts.htm is good high level introduction

there are composite/multiloop series conection of "good" op amp input with global feedeback wrapped around a bigger PA or Chip amp - there the keywords would be "multiloop", "active feedbck", many versions of "error correction"