Re: 'PC Speaker' Tube Amp

On Fri, 06 Jun 2008 18:30:13 -0500, flipper <flipper@[EMAIL PROTECTED]
> wrote:

.....snip!........
>>   Here I am replying to an old thread. 
>>   I tried using a positive internal feedback loop in a SET amplifier
>>that has a global negative feedback loop. By coupling two unbybassed
>>cathode resistors to each other through another resistor, I was able
>>to reduce the 2nd and 3rd harmonics by about 10db, with a small change
>>in overall gain. I'm impressed with this "trick"! Really, it is a very
>>small modification, and distortion is reduced by a factor of 3!
>>   Too much positive feedback, and the circuit would oscillate at
>>about 1-2 Hz.
>
>Are the two stages DC or cap coupled? AC coupling should break the LF
>oscillation since it wouldn't propagate to the next stage to then be
>fed back. (That's why I cap coupled mine in the PC Speaker amp)
>
>Dynaco used a similar circuit (DC coupled) in their ST35 but AC
>coupled the (positive) feedback resistor, which should also kill the
>motor boating since the PFB would 'go away' at the low frequencies.
>
>That is, unless it's coming through B+.
>
>In both cases you need to watch out for the negative and positive
>feedback time constants or else you may end up with net (excessive)
>PFB at the low frequencies (or you've made a 'bass boost' tone control
>;) ).
>
  
   It would start to oscillate even when I had the global feedback
disconnected. It is AC coupled. The amount of positive feedback that
would start it oscillating was a resistor about 1000 ohms (instead of
the 3300 I actually used). I assume that it is due to the B+ not being
sufficiently filtered, and the 2 front stages would feed back through
the B+. The output tubes run off a different supply, there is little
effect from them. 
   Since the frequency of oscillation is around 2 Hz, my guess is that
I can simply reduce the value of the coupling capacitor (0.68 uF into
a 220K grid resistor) between the two stages of the positive feedback
loop.  The -3db point of those values works out to about 1Hz. That's
really TOO low.
    There's probably an "effective" zero or two in the B+ (like a
large capacitor in series with the plate resistors, and common to both
plate circuits), and they are likely causing stability issues with the
feedback. I don't usually consider the stability, zero's, and Nyquist
Criteria at the low end, and here it's bitten me!  One zero is OK, but
two or more is trouble!
   By reducing the 0.68uF to 0.1 or 0.2uF, I should be able to make a
"dominant zero" just like you use a dominant pole at high frequencies.
The poor thing is still gutted open on my bench, so maybe I'll see
what effect that has.

-Paul