RCA tube amp (7199 & 7027A) 30W - NFB question

[Tubekwk]

I don't have much to add to this conversation other than to say I built one in the early 1970's and was very happy with it at the time. The RCA engineers were very good. Enjoy your vintage RCA amplifier!
BTW I made many of the circuits from the RCA manual and may well try out their mag preamp.

 

[6A3sUMMER]

1. Local Negative feedback and Global Negative feedback do reduce gain.
So, yes, they do "reduce" noise.

But they do not really change the signal to noise ratio of the input stage.
Gaussian noise is Very Hard to cancel.

It is much easier to produce a circuit that has lower Gaussian noise, than it is to cancel Gaussian noise.

If you figure a way to do that in a general case, like an amplifier, you may become our next Billionare ($10^9).

It is true that some use negative feedback to reduce the hum in an amplifier. Hum is not Gaussian.

2. -15V or -30V? R20 Ohms x total current draw of the input and output stages = Volts.
Documentation disagreement? Who knows.

Just my opinions.

 

[Kay Pirinha]

Why the voltage discrepancy in the two spots?

A drawaing error, I suppose. Maybe because the artist was as meticulously as he was when drawing the wrongly polarized 'lytic in the #22 schematics.

Best regards!

 

[Globulator]

This is exactly the reason why GNFB is taken from the OPT. NFB only reduces the gain of signals that are present and in their proportion. That means if certain frequencies are of a lower amplitude after the OPT then there is less FB to reduce their gain back at the driver. So FB simply works to lower the gain of the stronger part of the audio band compared to the input signal and so flattens the frequency response you get at the output, which is a better representation of the input signal. FB doesn't add anything that isn't there already, it cancels noise and distortion that is created by the amp circuit but not in the original input signal.

Yes, I can see that argument, it flattens the band so it becomes wider.
I suppose Thorsten's argument was that these lower feedback points were the ones that required the most feedback, for more linearity.
I wasn't sure what to expect with mine, but I converted the GU50 triode to Pentode to get max gain, and added as much feedback as I could. There's not a hint of 'bandpass' to the sound, and the OPTs are pretty cheap too: nothing fancy.
I think it's something you have to try for yourself, it's not a winnable argument on paper, as there are, IMO, too many variables.
But the transition definitely 'freed' the sound. Perhaps the phase shift alone.. before, which not enough to oscillate - is still wrong.. and affects the feedback at that point. Perhaps also there's a synergy between an unfettered OPT and the speaker - both magnetic, that cancels some distortion, which the GNFB with OPT somehow disturbs.
Anyway, just my speculation, as to why it sounded so much better.
I suppose the concept is to 'leave the OPT alone to do it's own thing', just to drive it as well as possible, and let nature work out the details. 😉