Well, we could talk about the stages of an amp. We already talked about the power supply, and the need to keep the filter caps from blowing when the tubes are pulled.
How about reverb? Let's talk about that. I love the reverb on a Fender Twin. Lush, very pretty. However the circuit only works with a Fender, the preamp gains have to be correct. With a high gain amp like a Mesa or Soldano, Fender's reverb has insufficient gain and you won't be able to hear it.
I've tried various tricks to resolve "insufficient reverb". You'll notice that in Fender's circuit, there is half a 12ax7 called "reverb recovery", and the output of that gets mixed in with the original signal by another half of a 12ax7. The reverb control is actually the grid bias of this second stage. So if your original signal is very hot, there's not much you can do to boost the reverb level with this circuit.
Another consideration is, if you want to put this circuit into a Marshall, the tone stack comes "after" the reverb. It's a little weird driving Marshall's tone stack from the plate of a 12ax7, which has an output impedance in the 40k range. Tone stacks suck up current, they have a pretty low input impedance. You really want to drive them with a cathode follower, like Marshall does - because the output impedance of a cathode follower is only a few hundred ohms.
So let's say you have a hot input signal, instead of 2 volts like a Fender let's say it's 20 volts like a Bogner. In that case you want a Dwell control on the input of the reverb driver, to keep from overdriving the reverb tank. The most voltage you'll get from "reverb recovery" is about a volt. So to keep the reverb working, you have two options: either reduce the level of the original signal, or boost the reverb output. You can boost the reverb output by adding another gain stage after reverb recovery, but in that case your reverb signal will end up out of phase, and you have to account for that. But reducing the original signal level might not result in enough drive for the PI. Plus we want a tone stack and master volume "after" the reverb.
So here's a couple of tricks. To address the tone stack issue, put a cathode follower after the reverb mixer. In a Fender Twin Reverb this means after V4b and before the master volume. It won't change your gain, and your signal will stay in phase. This way you can put a tone stack between the cathode follower and the master volume.
To address the reverb level issue, you can put another gain stage after V4a and before the Reverb control. However this will invert your phase, so you'll have to switch the output leads of the reverb transformer and then you'll be back in phase again. Another option is to change the grid resistor path on V4b. The original signal is basically going through a voltage divider consisting of the 3.3 meg reverb resistor on top, and the reverb control (and associated 470k grid resistor, for a total of 570k) on the bottom. Increasing the value of the reverb resistor has the same effect as decreasing the value of the 470k resistor. The idea is you want more reverb and less of the original signal. In theory you can do this by changing the value of the 100k reverb pot to 500k, and compensating by changing the 470k grid resistor to 100k. The total resistance stays the same, but now when you raise the reverb pot the reverb signal comes closer to the grid of V4b. Maybe this provides enough additional reverb gain, but in a very hot amp maybe it doesn't.
I've found that a 2.2 meg reverb resistor in parallel with a 50 pf cap is a delicious combination. But it boosts your original signal, which is the opposite of what you want. To compensate, you can get more reverb gain from V4a by changing the plate resistor to 220k and the cathode resistor to 2.2k. Then invert the values in the grid path of V4b. Doing these two things can get you about 8x more reverb signal, which might be enough. If it isn't enough, then we're back to an extra gain stage in the reverb path. One big advantage to an extra gain stage, is you can put an independent tone control on the reverb. Even a passive treble cut will work, it generates a darker and seemingly more lush reverb sound.
Fender's idea of using a low value (100k) reverb control is that the original signal level doesn't change (much) as you move the control through its range. The difference between 470k and 570k is barely audible. The best option if this is a concern, is add an extra gain stage and use a mixer, like in the Twin's PI where the two 220k resistors come together. Hopefully this is intuitive but I can provide a schematic if it helps.
Anyway, gotta sleep, 5am work schedule this week.
