Tell-tale signs of good feedback usage?

[kegger]

What kinds of things under testing and or simulation would lead you to
believe the feedback you are implimenting is correct and or good?

If it's negative feedback would you look for reduction in gain,distortion as
good signs the feedback is doing what it's suppose to do?

Along the same lines what would key towards bad implementations of feedback?

Any suggestions as to propper exercises for the use of feedback?

[janneman]

Originally posted by kegger
What kinds of things under testing and or simulation would lead you to
believe the feedback you are implimenting is correct and or good?

If it's negative feedback would you look for reduction in gain,distortion as
good signs the feedback is doing what it's suppose to do?

Along the same lines what would key towards bad implementations of feedback?

Any suggestions as to propper exercises for the use of feedback?

Stability would be important. I would feed the amp with square waves (low-passed at 100k or so, we need to stay realistic), and make sure that there is not too much ringing, especially with cap loads.

Also look at clipping behaviour, to make sure the clipping recovery is reasonable fast (say a handfull of uSecs). No sticking to the rails!

This is not really a test of the feedback itself as well as the complete amp with neg feedback. I guess that's what you are interested in.

Jan Didden

[richwalters]

HF instability is more of a problem than LF instability....One of the tricks I use is to put a piezo tweeter across a dummy resistive load with leads # 2meters length or so. Piezo's have a horrible negative impedance characteristic which many tube amps rebel at. A low level square wave between 1-5Khz will show all the ringing and overshoot. Scope probe earthing is also important.

A well designed 3-4 stage amp can have the global nfb loop increased by at least 10dB without showing signs of instability in a standard long-lead load. This (was) is the old industry standard test.

LF instability (i.e motorboating) can be caused by poor supply decoupling and overvalued interstage coupling cap(s). As with the global nfb comments, if the loop nfb is increased by 10dB and motorboating occurs, reduce the value of the interstage coupling caps by a 1/3. The bass will be somewhat fuller.

richj

[Brian Beck]

Piezo's have a horrible negative impedance characteristic which many tube amps rebel at.

I suspect you meant negative phase angle of impedance, as in capacitive. I doubt that piezo speakers have a negative magnitude of impedance. Using a piezo as part of a test load seems like a good idea, though.

[kegger]

All good stuff, I'm looking it all over.

I'm working with local NFB as of now and good ways to use it and to tell
if it's working out to be benificial in it's implimentation.

At the moment I'm trying to stay away from global/loop NFB.

Just experimenting here.

[janneman]

Ahhh well, most of the problems discussed here do not occur with local feedback. You should have said that you only do local fb....

Jan Didden

[kegger]

I'm interested in all forms of NFB, but at the moment experimenting with local NFB.

[Wavebourn]

Originally posted by kegger
I'm interested in all forms of NFB, but at the moment experimenting with local NFB.

Do you mean cathode followers?

How are your experiments going?

[kegger]

Basically my experiments have been plate to grid feedback around single stages.

Douglas here, AKA Bandersnatch clued me into an interesting way to do it on the
Schmidt/LTP type phase splitter stage which worked for what I wanted very well.

Now I'm working on first gain stages trying the same thing and like what I see quite
a bit, now I just need to get it into the real world and see how it actually performs.
(it's all on spice simulation at the moment for this stage)

My main intent was to be able to do away with (global/loop/from the output trannie)
back to the front of the amp while keeping the sensitivity of the amp low enough to
still be able to use active linestages while in the process hopefully making the amp
perform and sound better, so far so good! More along the lines of modifying things
then actually builds from scratch. But as things tend to go if I get what I want they
will lead to implimentations into my own creations, which some have.

I'm not trying to say that global/loop is something that is to be definatly avoided but
at the moment experimenting without it and believe I like the results without using it.

[richwalters]

Originally posted by kegger
Basically my experiments have been plate to grid feedback around single stages.

I'm not trying to say that global/loop is something that is to be definatly avoided but
at the moment experimenting without it and believe I like the results without using it.

I've also tried 1st stage plate to grid with direct resistor and pulling the grid g1 with a negative bias the opposite way if you see this. This will lower the input Z and due to DC shift will place a cap on the input. In a way this imposes a grid bias condition and away from autobias.
I deliberately avoided a coupling cap with resistor from plate to grid....project rethink.... the long time constant created a hangover effect but has to be optimised to avoid another LF pole in the LF response below o/p tranny cutoff. When the first stage has a small signal pentode some screen grid interaction occurs, complicating matters.
A drawback with direct coupling is that a consistent DC performance with the same tube type from different vendor cannot be acheived. This implies the AC thd harmonic performance esp with pentodes will be inconsistent and I've found this has an effect on sound quality.

Global feedback say 15-20dB is a desirable asset to correct LS deficiencies.


richj