Ron, Kyrill,
I had a long response to Ron's post set up yesterday, but it slipped away during a clumsy keyboard manoeuvre, perhaps because the bloody cat was on my lap at the time..... (I love cats, I often feel the world's problems vanish when you stroke a cat).
Ron, thank you for this seminal work. One of the engineering truths seems to relate to size and quality; if C4 is to be really good, it will be LARGE. I suspect that the Solen will be the go, but if the Wima, with its much smaller size, does the job nicely, then we are quids ahead. Are they 22uF or 47uF?
The quality of C7 is very important, too. This is because the front end of the SS section is highly susceptible to noise injection, and C4 and C7 shunt this noise to ground, preventing it getting into the long tailed pair, T1/T2. Better caps in these positions will a quieter preamp make.....
I don't think C19 can be much improved upon, a polystyrene is just fine here, and in any event no AC current flows because it's driving an ultra-high impedance tube grid (4.3M). C21 might go well if you bypass it with a 47nF teflon, just to pick up on high frequencies.
Aside from that, I can think of no other caps which matter. C18 shunts noise from the tube supply; it could be bypassed with a 470nF teflon, I guess, but it would be very bulky; C20 carries signal, but to the drone cathode follower, which merely serves power supply duties, not in the audio chain; C10/C12 might be replaced with Black Gates but even there I doubt it would have much effect.
Cheers,
Hugh