Playin' with feedback

[astouffer]

This long weekend has motivated me to take some distortion readings and add some feedback to my PP 429A amp. At 1 watt THD was pretty high at 0.8%. First I wired up a 20K pot in place of R16 to see what effect it had. To my amazement the lower the pot value the more the distortion increased! Took me a half hour to swap the transformer secondary around and fix it. I thought having the feedback connections backwards caused horrible oscillations? Now distortion is 0.2%, much better. Of course the gain was now much lower but the addition of C5 brought things back to normal.

[zigzagflux]

C5, if large enough, effectively eliminates the feedback. Something's not quite right, here.

[tubelab.com]

I thought having the feedback connections backwards caused horrible oscillations?

A small amount of positive feedback will increase the gain and distortion. Too much and all sorts of wicked sounds will come out of your speakers.

C5, if large enough, effectively eliminates the feedback. Something's not quite right, here.

This circuit is similar to the circuitry I used in the Simple P-P amp. You have the feedback wired slightly wrong. There are two ways to wire this.

#1 (the way I do it) Swap R13 and R9. Connect the feedback (R16) directly to the cathode of the input tube move the positive end of C5 to the junction of R13 and R9.

#2 (a bit less common) Leave the feedback connected as is, but move the negative end of C5 to the junction of R13 and R9.

The #2 method will require less work, but places an electrolytic cap in the feedback path. I have seen both methods used, and they would be electrically identical with ideal components.

I use the same circuit in the simple P-P except that I leave out C3, R8 and R11. C3 and R11 are replaced with a short and R8 is left open. I use a 12AT7 so this may not work without some resistor value tweaking, but I get distortion readings under 0.1% at 1 watt and under 1% at 10 watts with EL84's.

R16 sets the amount of feedback and you can place a small capacitor across it to smooth out the high frequency response. The exact value depends on your OPT. My cheap transformers need 510 Pf.

[Gordy]

As an aside, may I ask which drawing package you used for the circuit diagram? Thanks.

[astouffer]

Thanks for the tips tubelab. This is my first encounter with feedback and being able to measure the results.

Zigzagflux: The readings were taken before the addition of the bypass cap. So they could have very well changed.

Gordy: The software package is Altium Designer. Its insanely expensive for the hobbyist but its used at my job.

[kruesi]

I'm with zigzagflux- if C5 is large enough, the cathode of V2a is at ac ground- and will eliminate any feedback. I'm not sure what you're trying to do with C5... A small cap across R16 will effectively set the high frequency breakpoint, and to set the LF rolloff, an RC parallel network at the input may be a good way to go.

With feedback, the gain will be lowered compared to open loop. There's no way around this- if you want to have an amp at the same gain as no feedback, and you're using (say) 10dB feedback, you'll need to build in 10 dB of "extra" open loop gain.

10-20 dB feedback is probably reasonable- beyond 20 dB, HF stability may be difficult to maintain

Feedback is a great thing! -Not sure why I'm seeing so many zero-feedback designs these days. Feedback will
* Lower the distortion
* Lower the noise (increase S/N ratio)
* Lower the output impedance (increase the damping ratio)
* Flatten the frequency response
and of course
* Lower the gain

It will also cause the distortion components to be of higher order (you'll see more 5th, 6th, 7th, etc) which are arguably more audible than 2nd and 3rd, but they're of MUCH lower amplitude.

I hate making these kinds of sweeping generalizations, BUT... A well-designed amp with feedback will usually wind up sounding cleaner and more neutral (lacking a sound or quality of its own) than a similar amp without feedback.

-My $0.02.

[richwalters]

Originally Posted by kruesi

I'm with zigzagflux-

10-20 dB feedback is probably reasonable- beyond 20 dB, HF stability may be difficult to maintain

That is the symptom of a poorly designed amp. A well designed 3-4 RC coupled staged Class AB amp should get reach 36dB of total global feedback, that is on the onset of both LF & HF instability. This requires one-heck-of working out and skill; but having got this, one is assured total stability with complex phase shifted crossovers and using loudspeakers with long leads.

If ones amp shows HF instability and ringing on square waves tests then it can be rest assured that these will appear on signal transients to effect sound quality.
On solid state amps.,the cue is mysterious burnt out tweeters. With tube amps, the o/p stage usually ends up stewing red hot.

richy