I never used the normal channel on my Twin Reverb.
So, for the sake of variety, and because I enjoy
tinkering, I gradually Marshallized it. These are the
steps towards Marshallization:
 
1. Put in Marshall tone stack. A no-brainer.
 
2. Disconnect tone stack from 1st gain stage, connect to
2nd gain stage. This involved switching the plate leads
around. I think I also used that coupling cap on the left
side of the channel's components to couple the 1st gain
stage to the volume pot.
 
3. Cathodes: 1st, I changed the cathode circuit on the
1st gain stage to Marshall bright channel specs. I can't
remember if that's 2.7k / .68 uf or 1.5k / .68 uf. Then,
I used one side of the bright switch to add in a 220uf
bypass cap in parallel when the bright switch is off. I
connected the 2nd gain stage cathode to a cathode in
the reverb driver tube, cuz it has an 820 ohm cathode
resistor. 
 
4. Bright switch: A .001 (1000 pf) bright cap.
 
5. Cathode follower - the final touch. I used the mosfet
recommended in this article, to which one of y'all
referred me:
http://www.geofex.com/Article_Folders/mosfet_folly/mosfetfolly.htm
 
In order to fit the mosfet in there, I disconnected the
33k resistor from the plate load resistor, and angled it
upwards so that it directly connects the treble cap to the
midrange cap (leaving out that extra space below the
treble cap). Then, I attached the source to the junction
of the 33k r. and the treble cap, and further added a
100k resistor between this point and ground. I arched
that 100k resistor over some other components to the
top of a cathode circuit, after slipping some insulation
over the leads. I connected the drain to the B+ supply
(before it goes through the plate load resistor). I
connected the gate through a 220-ohm resistor to the
plate of the 2nd gain stage, at the eyelet on the bottom
of the plate load resistor.
 
The normal channel now sounds more Marshally that I
ever expected. Deep, thick, gritty, and uncompressed.
For whatever reason, it doesn't get as loud as the
vibrato channel does, so now with an AB box I can use
the normal channel for rhythm and the vibrato channel
for lead. You might think the normal channel would
need a bigger coupling cap into the pi, but it gets
enough bass for my purposes just using the .001. Also,
I haven't seen any need to reduce the plate voltage to
the normal channel, because it's just dirty enough.
 
Shea

Part of the reason is the tone stack's insertion loss
happening after both gain stages instead of between,
which causes the second gain stage to see a hotter
signal, compress and/or distort, and *then* have its
level knocked down by the tone stack. BTW, I've had
really good results without a cathode follower (or
source follower) by just hooking up the tone stack to
the plate of the second stage, just like it would be
hooked to the first stage in the standard Fender BF
layout. Just tweek a couple parts values in the tone
stack and it works great. 
 
The other part of the reason is the loading on the output
of the tone stack by the 220k summing node resistor.
This can be eliminated if you add another MOSFET
source follower to buffer the output of the tone stack
before it hits the summing node. 

     You might think the normal channel would
     need a bigger coupling cap into the pi, but
     it gets enough bass for my purposes just
     using the .001. 


 
The reason that the .001 works OK there is the
"bootstrap effect"--the bottom end of the 1M grid
resistor is not hooked to ground, but to a tap in the
cathode resistance. This means the bottom end of the
resistor is being fed a signal that's *in phase* with the
signal coming in from the coupling cap (positive
feedback), and it raises the input impedance to a fairly
large multiple of the 1M grid resistance, because it's
"helping" the input signal wiggle the grid. Its input
impedance is *very* high, so the RC highpass formed
by the .001 and the input impedance still passes plenty
of lows.... Randall Aiken's tech page has a formula for
calculating it, I believe. 
 
C ya,
Dutch