• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Negative Feedback

Jump to Latest
#244
I accept that there can be no difference in kind between various feedback loops, but just wonder if there's some significance to the propagation time. At first blush, it's obviously too small to matter. But, after integration? Probably still not. But I don't have the skills to weight its importance to the integrated output. I feel like the topic keeps veering back towards the class-D issues of timing and its relation to linearity, but how much matters?

update: any internal feedback loop has its own internal delay

All good fortune,
Chris
 
#245
On the practical differences Jan mentioned (though not relating to stability), the physical area of the FB loop might be be considered too.
It affects the magnetic pick-up of extraneous "noises". The lower the impedance of the circuit and the larger the loop area, the effect increases.

George
 
#248
jan.didden said:
I believe that the difference between the very tight internal triode anode to grid feedback and a larger loop feedback is that the tighter and smaller the loop, the smaller the phase shift, and the less impact on stability.
Those are practical differences, I don't see any principle difference.
Click to expand...
There's certainly correlation between stability and integrator time constant, and between that time constant and open loop phase shift. But are we certain that the propagation time, which smears this clean model (a little) doesn't matter? It should be a matter of numbers, but I don't know how to calc the numbers. Propagation time smear must add something calculable?

Much thanks,
Chris
 
#250
Black's feedback was initially ment to correct the triode's nonlinearities, little did he know that 100 years later we'd still be unable to have a perfectly linear device and his math approach would still be useful...the mu-follower and white cathode followers were though successfull attempts to improve circuits based on triodes, one with gain higher than 1 the other with lower than 1.
https://www.diyaudio.com/community/...ower-analysis-and-design.299609/#post-4898706
Besides they do solve some of the problems Bruno mentioned where you can't improve everything with global feedback and you need to adress the first stage nonlinearities before gnfb. Arto Kolinummi had simillar thoughts where he went to apply huge amounts of local feedback and no global feedback, but the total amount of feedback is similar gor greater linearity throughout bandwidth.
Pentodes are more like cascodes, so feedback term is not proper unless you're using a mid tap of a transformer to drive that screengrid and then you have a feedback system.
 
Last edited:
#251
MvdG, Then, maybe we should separate capacitances from other active device internal feedbacks (if any !), as they effect phase but not time delay. Those effects fall well within FFBP. I'm particularly interested in finding any possible mechanism that would distinguish very local, very fast, very low time constant, very small delay feedbacks from slightly larger delays, and no other time constants need be changed for stability in practical amplifiers of that topo.

Might be a silly idea, but maybe somebody has something to make a connection?

Much thanks, as always,
Chris
 
#253
Chris Hornbeck said:
I accept that there can be no difference in kind between various feedback loops, but just wonder if there's some significance to the propagation time. At first blush, it's obviously too small to matter. But, after integration? Probably still not. But I don't have the skills to weight its importance to the integrated output. I feel like the topic keeps veering back towards the class-D issues of timing and its relation to linearity, but how much matters?

update: any internal feedback loop has its own internal delay

All good fortune,
Chris
Click to expand...
I believe that the phase shift, accumulating with more stages withing the loop, has orders of magnitude more effect than the accumulation of tiny delays.

Jan
 
#254
I keep coming back to Jan's thoughts in post #240 about the relation of the integrator to stability. Is there a threshold area, some non-linear transition, between our ability to recognize the (whatever) differences between small, local, "tight" feedback loops, for whatever reason, and longer, still very fricking short delays, but bigger time constants, loops?

Stability and integration time are linearly related in a simple single pole integrating system, and both are independent of delay (for audio modelling purposes). My question is: could the (ignored, for usually a very good reason) propagation time be significant to the (possibly even true!) observations about local degeneration, and the easily observable fact of triode curves approaching the Child simplification (that assumes infinite parallel planes, not space charge limited, etc.)

Might be nowhere to go with it, but even a way to calc the contribution of some tiny time delay to the integrators' output is so difficult that I despair of understanding it.

"Camelot!".
"Camelot!"
"It's only a model."

All good fortune,
Chris
 
#256
When you wrap a fb loop over more than one stage, you most probably no longer have a simple 1st order system. Each stage adds a pole, so at a certain frequency the ol roll off changes from a 6dB/8ave to 12dB/8ave or more. That's when instability starts to rear its ugly head.
If your fb gain curve cuts the ol gain curve at a point where the ol rolloff is 12dB/8vave or more, your phase shift turns the nfb into pfb and you've build a fine oscillator.

Jan
 
#257
I'm not sure that would be true. Signal voltage would still appear at the anode, and capacitance would still be there, so lots of the model is still in place. (Nearly) shorting the anode, like with a high Gm cascode stacked on top, might be a better analogy for your model.

All good fortune,
Chris
 
You must log in or register to reply here.
Top Bottom