[AMRadio] Push-Pull Parallel Tubes In Modulator

[patrick jankowiak]

One thing of importance in hi-fi amps is separate bias adjustment for
each tube. I know we are discussing modulators, but also I think that
the best possible fidelity is important. So I'd like to offer some
comments about turning PA amps into hi-fi amps from my own experience.
I regret these are all pentodes (beam power tubes), so I cannot offer
much on triodes, although I have toyed with the idea of push-pull
8025's zero bias class B with 300VDC on the plates and a 6-8K p-p
load. No sockets needed with these either, but I digress.

In regards to AB1 operation, Another handy circuit enhancement is a
small (1-10 ohm) resistor in series with each cathode to ground (not
so easy with directly heated tubes). Such a resistor lets you measure
not only the resting current of each tube, but also the cathode
current when driven hard. At near-clipping, you can check the current
of each tube in the quad, and match them up, so each side of the
OutPut Transformer has the same power input. 

I wrote a short article about these and other mods to an old RCA
theater amp using PPP 807's. That old amp was originally class-A,
tube-hungry, and sounded more like a cheap tin can than a hi-fi amp,
but after the mods, it was pretty much an easy-listener, and the
necessity for matching tubes and high bias currents was eliminated.
Values are also shown for parasitic resistors, etc.

I think the best improvement I have ever seen in a tube amp for
linearity in the face of odd loads and funny waveforms, is the cathode
feedback system where the cathodes each return through a winding on
the output transformer, where the winding bucks the individual cathode
current with a voltage equal but opposite to about half the grid drive
voltage. For whatever reason, this mod seemed to make the amp behave
as though its internal source impedance was very low. The output
seemed to regulate better. It works best when the output tubes can
physically deliver twice the normal current required. (more after this
URL) The best winding for this on 20-150W amps has been the 25VCT line
winding, or the 0-4-16 ohm taps (the '4' tap now becomes GND and your
8-ohm speakers float).

http://www.montagar.com/~patj/mi12188a.htm

That said, the 807's can't deliver twice the current in that RCA amp.
So I got a great sounding amp with 40 watts instead of a mediocre but
loud amp with 70 watts.

The following circuit shows a 50W amp modified the same way, but with
much larger tubes substituted for the originals, and after the mods,
that amp provided the same output to strange loads as stock, but with
much better linearity and fidelity. Also of note is the output iron
which was rather large on that amp. The larger iron might have made
better use of the extra current. In both cases of using a cathode
winding on the OPT, more voltage is needed from the driver to overcome
the error voltage. Also, the overall feedback loop (the RC network
which surrounds most of the amp) will need some changing, as less
overall feedback is needed once the cathode feedback is implemented. I
measured the gain of the amp stock, and when done, set the overall
feedback network's resistor to a value giving the same overall gain.
The network's capacitor was set to reduce ringing on square wave test
signals or to flatten or adjust high frequency response as needed.

http://www.montagar.com/~patj/aph1050.htm

About the need for extra current, The extra current takes place during
times when there might otherwise be amplitude distortion due to
irregular input waveforms and transients. i.e. the output tube's grid
drive says "go from 0 to 40V in 1 microsecond", but the OPT says "no
way I'm reproducing that!". Well, since the OPT lags due to its
inductance, there's no immediate bucking of the cathode of the tube
and the output stage power gain is not reduced until the OPT current
actually rises. Here's where the extra surge of current comes in, for
those short time intervals. This might all sound very strange, but you
can hear the difference in quality and you can see the action on a
scope, watching the cathode current vs. plate voltage during a
repetitive irregular waveform. (how this can be applied to a high
power modulator with no special OPT windings I have no idea yet) I
suppose in theory, the overall feedback circuit should do this but it
is not effective. This mode corrects things around the output
transformer only. In the 50 watt amp last mentioned, when you put a
20Hz pulse with a duty cycle of 10-20% in, you can really hear the
transformer 'thunking', a testament to the correction going on there.

One comment about the parasitic supperssors, it was at the peaks of
the audio cycle where I noticed most of the oscillation in my amps,
and the suppressors on plate, screen, ad control grid nailed it in
some cases, and in other cases, the value of the cap in the overall
feedback loop nailed it. Oscillation during these peaks can ruin an
OPT, and usually cause a large increase in tube current.

Patrick