Below is a circuit that I am using to substitute for the interstage audio coupling transformer of
my Ranger.
I carefully removed the transformer (saved for possible reuse) from my Ranger
I put this circuit in its place.
Notes:
A. The FQA 11N90 FETs I used are the same FETs as I use in my Class E transmitters. They can
handle 900 volts drain-to-source. They can deliver 11 amps (obviously overkill for this application).
The FETs are only dissipating about 1 watt each... but they require some heat sinking. I'm using
a small, copper, sheet metal, right angle bracket to attach them to the chassis where the transformer was mounted. Note
that these FETs require an insulating, thermally-conducting gasket under them, because the metal back side of the FET is connected
to the drain of the FET. The purpose of the 12 volt Zener diodes is to prevent transients from destroying the FETs.
Using FETs with a smaller current rating, and a correspondingly lower gate-source capacitance would
make it easier for the phase splitter to drive the source followers at higher frequencies... allowing the source and drain
resistors of the phase splitter to be larger... and therefore reducing the average heat dissipation of the circuit. The FQA11N90 FETs
have about 3000 pF of gate-source capacitance, which corresponds to around 10.6 kohms of capacitive reactance at
5000 kHz. There are 600 volt / 1 amp FETs available, which have a gate-source capacitance that is less than 1/10th as
much as that of the FQA11N90s.
The phase splitter FET, and the source follower FETs are capacitively-coupled at their
respective inputs*. The phase splitter is biased at ~100 volts with respect to ground (1/3 of the supply voltage) and the
source followers are biased at ~ 75 volts with respect to ground (1/4 of the supply voltage).
*Originally, I didn't use a capacitor between the source of the phase splitter, and its source
follower... but, to keep the ac loads on the phase splitter's source and drain the same, I decided to use capacitive coupling
and a voltage divider (for biasing) on both source followers.
Both source followers are capacitively coupled to the grids of their respective modulator tubes with a 100 uF electrolytic capacitor. The modulator grids are biased using a pair of 10k ohm resistors (not
shown).
The combination of the 100 uF coupling capacitors and the 10k ohm modulator tube biasing resistors
(not shown) act as slow-acting bias-adjusting circuits. They tend to keep the modulator tubes biased in class AB1. I.e.,
if the modulator grids draw current, the 100 uF capacitors will tend to charge up... and, therefore to lower
the grid bias voltage.
In keeping with the above, I am using 6550s for modulator tubes. They provide plenty of audio power
in AB1 operation. The plate voltage is ~ 600 volts. The screen voltage is ~300 volts, the grid bias is ~ -33 volts, and
the resting current is ~ 100 ma.
