[Morris Slutsky]

There's been a lot of discussion on AGA lately about the Marshall 2204
"Cascaded" preamp circuit and whether or not it sucks and how much. A
lot of people really do like the high gain preamp. But what sucks
about the cascaded Marshall circuit is that there's no way to switch
the high and low gain besides physically moving a cable from one input
jack to the other. That circuit was a long time ago, though, these
days everyone likes channel switching amps which use relays or other
switching logic to push the signal through different tube stages and
recombine them at the power amp. Which is complicated and no fun to
homebrew.

Anyway I came up with this circuit, which allows easy footswitching
with no relays or anything between a 'clean' input stage and a 'high
gain' input stage, with appropriate frequency response voicing. It's
built and working, as part of a homemade amplifier, and I really do
like the sound so far.

I don't know if it's original or not, in the context of a guitar amp,
but I think it's pretty neat.

http://img718.imageshack.us/img718/8...edfeedback.png

This circuit pretty much looks like a Marshall "cascaded" preamp
here. But with the footswitch open, the 47K resistor and 0.44 uF
capacitor form a feedback loop between the plate of the second triode
(half of a 12AX7) and the cathode of the first triode. The 0.44 uF
capacitor really doesn't do anything here but just pass the feedback
signal through - the DC is already blocked by the 0.022 uF coupling
capacitor between the second triode's plate and the output load, and
any high-pass filtering is going to be dominated by the 0.022 uF cap
as well. The result of the feedback loop is a stage with a moderate
amount of gain, pretty flat frequency response, and low distortion
unless driven to the limits where it'll hardclip, which a typical
guitar pickup won't be able to do. So it's a typical 'clean' input
stage.

Now close the footswitch, this grounds the middle of the feedback loop
and breaks it. The 0.44 uF capacitor now becomes a cathode bypass
capacitor for the first triode, bringing in a gain boost above about
300 Hz. The 47 K resistor becomes just a dummy load hanging off the
output. Without feedback, the stage distorts much more readily, gain
is much higher, and the frequency response is not as even. Lows are
rolled off by the interstage coupling RC network and the treble rolls
off due to Miller effect capacitance. It ends up being a typical
'high gain' input stage with appropriate frequency response and
distortion. Crunch!

Small-signal SPICE simulation of the frequency response and gain of
each stage is he

http://img20.imageshack.us/img20/379/freqresponse.jpg

What makes this really amenable to footswitching is that although the
gain at the switching point is high, the impedance is very low, so
noise pickup isn't much of an issue. The feedback loop keeps the
impedance low to such noise signals, it's sort of a virtual ground.
So I just think this circuit is really neat.

There's a lot of talk on AGA about the differences in voicing a clean
and distorted channel. It's true that you can't totally tweak
everything here, but you can probably do quite a bit to change
frequency response and gain. The feedback resistor could probably
even include a variable pot.