gain control in differential tube amps

[olafmatt]

[..]

 

[Boswell]

Yes, the principle would work. It's the standard instrumentation amp with variable gain configuration. The AC and DC paths are separate, so set the original source resistors for the minimum gain required when paralleled with half the pot. Put a resistor in series with the pot to set the maximum gain.

You would have to adjust the grid bias voltage to cope with the changed source resistor values and still leave working headroom.

You're left with two problems:
(1) What to do with 2 (or more) FETs in parallel.
(2) What do do about d.c. imbalance between the + and - FET paths.

Take (2) first: since the circuit is AC coupled anyway, you can use a large non-polarised capacitor in series with the pot.

So for (1), and this is a bit iffy: split the "max" resistor referred to above into two equal values. Use 4 resistors of this split value, one from each FET source with a blocking capacitor to the relevant end of the pot. In this way, the FET sources are not directly shorted together.

 

[Boswell]

It's the CCS that would need the extra headroom, so I should have said the -50V source. I think that in a practical design you would need to have the CCS sitting on ground with everything else perched higher up and the HT correspondingly higher.

A small value balance pot at the tail of the source resistors could be used for equalising the static potentials at the FET sources, but it would not guarantee an a.c. balance.