Paul,
your case with that amplifier is interesting.
Could it be that your plate choke was arcing over between those turns
where the insulation flaked off? Or perhaps those turns tended to get
shorted via creepage over those blackened areas, without visible arcing.
Any such shorting of turns would result in a drastic reduction of the
choke's inductance, bringing it into a range where resonance with bypass
caps and the like makes the amplifier unstable in the low frequency range.
It might even be that the negative resistance behavior of such an
arcing or creepage between turns contributed some gain at the
oscillation frequency!
Upon reaching about 700 watts out, the power output suddenly dropped
off sharply to less than 200 watts. It wouldn't cone back up without
reducing drive to a very low level and then bringing it back up. When
drive was brought back to a level sufficient to produce 700 watts
out the sharp drop would happen every time.
That's a classic sign of the amplifier breaking into oscillation.
I am somewhat paranoid the plate choke wasn't the whole story, but I
can find nothing else wrong.
It's quite likely that the only other "fault" would be that the
combination of bypass capacitances, stray inductances, and so on,
including those in the tube socket, are giving you a rather marginal
stability, so that even a small change in the plate choke's inductance
is enough to make the beast oscillate.
Some hints to fix this:
- Load down the grid with resistors, as much as you can until being left
with just the necessary gain. Those resistors should be
broadband-coupled to the grid, all the way from very low frequencies to
VHF. For example if the resistors go in series with a capacitor, and the
group sits between the grid and ground, then the capacitor should have a
low reactance even down at very low frequency - that calls for at least
about 1uF. And to be effective up at VHF, the series inductance of the
resistor/capacitor combination, including connections, should be under
15 nanohenrys or so. In practice, a good way to achieve that is to use
several physically small, "noninductive" resistors, connect each in
series with a physically small 0.1uF cap, with short leads, and then
distribute these RC groups conveniently all around the grid connection
to the chassis. That way you reduce the chance for VHF oscillations too,
while providing enough total capacitance for VLF effectivity of the
loading resistors.
- If you happen to have another choke in the circuit, for example at the
grid, or perhaps the screen, make sure that it is totally different from
the plate choke, in terms of possible resonances by itself or with other
circuit components. This is to avoid creating a tuned plate, tuned grid
oscillator, which can oscillate with just a very tiny amount of positive
feedback, which you will always have at some frequency!
Since the plate choke is air cored and has relatively low inductance,
any other choke could be ferrite cored and have a large inductance, much
larger than really needed for other reasons, so that in the frequency
range where the plate circuit loves to oscillate, the grid circuit will
just be lossy, and definitely not resonant! That tends to take the wind
out of the sails of any would-be oscillation!
If you want, send me the schematic, just to see if I can catch any
further reasons for oscillation.
Manfred
========================
Visit my hobby homepage!
http://ludens.cl
========================
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|