Reading material regarding feedback loops

[KarenColumbo]

I breadborded this great thing recently ...
http://www.runoffgroove.com/professor.html
... and realized that tinkering with the values of C and R in the feedback loop between Q2 and Q3 altered the sound spectacularly.

Where, fellow travellers, can I read more about feedback topology, origins and explanation of feedback?

[PRR]

At a glance, the voltage gain tends to approach 22K/1.5K. (Actually (22K+1.5K)/1.5K or 15.6.)

This is so as long as the available gain is much larger. Guessing Gv=20 in each stage, we have gain near 400, which is much greater than 15.6, so the feedback has good effect.

Oh, but the 1uFd makes the 22K act "large" at low frequency. How low? 1uFd is 22K at 8Hz. So, little effect in the audio band.

This is often taught as "op-amp" theory.

The origin of NFB in electronics is often credited to Harold Black, though the idea was widespread. But NFB to control engine speed was known to Watt, and not new to him.
https://en.wikipedia.org/wiki/Negative_feedback#History

[thermionix]

More interesting reading:

http://www.valvewizard.co.uk/localfeedback.html

[KarenColumbo]

Thank you! I read these texts and am beginning to get the idea. What I can't seem to understand is that @ valvewizard's ruminations (and e.g. @ TubeCAD journal) and in some of my amplifier books Negative Feedback is applied by "pulling" some of the output signal off from the drain/anode to gate/grid. In ROG's professor tweed (and also, just as another example, legendary Doug Hammond's "Meteor") there is a portion fed off one gain stage's output (drain/anode) to an earlier gain stage's source/cathode.


So it doesn't "mix" with the input signal but somehow influences the gain that's set by Rk(/Ck), I gather? Or is one thing LOCAL feedback, the other thing GLOBAL feedback?

[thermionix]

Or is one thing LOCAL feedback, the other thing GLOBAL feedback?

That's my understanding of it.  It generally does the same thing, just a portion of signal is fed back to an earlier point in the circuit where it is opposite phase, so as to partially cancel out.  I'm probably oversimplifying by a great deal, I tend to do that.

[KarenColumbo]

Or is one thing LOCAL feedback, the other thing GLOBAL feedback?

That's my understanding of it.  It generally does the same thing, just a portion of signal is fed back to an earlier point in the circuit where it is opposite phase, so as to partially cancel out.  I'm probably oversimplifying by a great deal, I tend to do that.

I see! I was just flummoxed because of where the feedback is inserted: At the source, not at the gate. I always thought that the source doesn't "play" in the signal path but is just there to set a virtual 0-point for the actual signal to wobble around. While local feedback gives back the output signal to the gate to be mixed with the input.

[anotherjim]

Watch out for global feedback. Interstage phase shifts can turn it towards positive feedback at some frequency. Local feedback is usually too simple to introduce enough phase shift to cause that trouble, but that's exactly what we do to make a phase shift oscillator.

[KarenColumbo]

So where do I introduce global feedback? And one question escapes me: why looping it back to source instead of gate?

[amz-fx]

So where do I introduce global feedback? And one question escapes me: why looping it back to source instead of gate?

The source of a jfet can act as a non-inverting input. Any signal sent into it will appear at the drain of the same transistor with the same phase as it was sent in, though attenuated by the source resistor (unless driven by a very low impedance signal).



regards, Jack

[R.G.]

The same is true of the emitter of a bipolar transistor, source of a MOSFET or JFET, or cathode of a triode or pentode. In all of these, the active signal to the device is the >difference< between the more normal input (i.e. base, gate, or grid) and the secondary input in the power path, the emitter/source/cathode. In all of them, the base/gate/grid is an inverting input and the emitter/source/cathode is a non-inverting input.

For single stage amplifiers you usually hold the emitter/source/cathode still by some means, grounding it directly or through a capacitor, and put signal in through the low-power input - the base/gate/grid. This produces the highest >power< gain from the input to output, because the signal at the output (collector, drain, or plate) is both a bigger voltage and current than the signal at the input.

If you hold the normal input pin (base/gate/grid) still with a capacitor or grounding and shove signal in through the secondary input (emitter/source/cathode), you get exactly the same current change at the output pin, but at a higher voltage, so this connection gives you voltage gain, but no current gain.

Finally, if you put signal in at the nominal input pin (base/gate/grid) and also signal in through the emitter/source/cathode, you see the difference between the two coming out amplified at the output pin. There are some "operational amplifiers" that use the base and emitter of a bipolar transistor as the + and - inputs of the amplfier; some audio power amps do this directly as well. In the pre-IC era, the secondary input (emitter/source/cathode) was often used as a feedback input for a multi-device gain path.

[KarenColumbo]

Ha! Now I see! Thank you for the explanation - that's food for further experimentation!

[antonis]

@Andreas: Just make yourself sure you can distinguish between "pure" NFB loops and those with also biasing purpose..
(usually, the difference between them is just a single cap but that cap isn't always present..) 

[KarenColumbo]

@Andreas: Just make yourself sure you can distinguish between "pure" NFB loops and those with also biasing purpose..
(usually, the difference between them is just a single cap but that cap isn't always present..) 

Thanks, I will bear that in mind!