> soft start recitifiers like the 6/12X4 should have no problems with large cap values then, there is no surge of current with these tubes.
No, hollow rectifiers DO have trouble with over-size caps.
I'd much rather put 6V on a 5U4 (it'll last a good long time) than 500uFd on a 6X4. A good 6X4 can probably pull-up 80uFd thousands of times, and it is not impossible that it would survive 500uFd, but....
Look what happens when a tube rectifier warms up.
Cold, no current passes. Hot, lots of current passes at fairly low voltage drop.
But between cold and warm, there is a time when the cathode is just barely starting to emit a few electrons. At this moment, the AC voltage swings up to 500V, but the cap has no charge yet and is sitting at zero V. Those few electrons are ripped away from the cathode, slam into the plate at full 500V velocity. If this keeps up for long, the cathode is ripped apart. A small cap will quickly charge-up, a big cap will hold the output at low voltage until real damage is done.
As cathode heats up and cathode current rises, but output voltage stays low due to the huge cap, the current can become quite large. Until the cap charges, current is limited only by the voltage difference and the stray resistance. We strive to keep resistance low for good operating regulation. For a preamp, say 200 ohms in the winding. And say 500V off the transformer, cap only up to 100V: the peak current can be 400V/200= 2 Amps! Probably less due to rectifier self-resistance. However 1 Amp peaks are not uncommon when starting into a big cap.
Also: the electrons slam the plate at high velocity. Normal tube drop and electron velocity is 50V-100V; with a big cap the drop slowly falls from 500V. During that time the plate is dissipating 5 to 10 times more power than normal. It gets hot. A hot plate will emit electrons. Not as well as a good cathode. But good enough to start conducting backward; i.e., not rectifying at all, but just cooking both ways. This makes things hotter and you get "arc-back", which make instantly kill the rectifier or transformer (big arc-back will blow power fuses).
Equipment designers of old were only a little sharper than those today. They did sometimes use over-sized caps (though never 500uFd!). In self-defense, rectifier factories quietly beefed-up their parts: it was cheaper to do that than to argue with customers about careless designers.
If this design uses the stock Dyncao transformer (you can get them from TriodeElectronics), that tranny had quite a high winding resistance. With large resustance the starting surge is, of course, lower, and maybe the 6X4 survives very nicely.
Anyway, the first filter cap rarely needs to be better than 1% ripple. Say 4V. Four sections of 12AX7 can't be sucking more than 10mA, confirmed because that 10K resistor makes no sense if we pull much more. 1,000uFd at 1A (1,000mA) gives around 1V ripple, so 10uFd at 10mA gives 1V ripple, and we only "need" 2.5uFd to give 4V ripple. Since caps are cheap and 6X4 can stand 20uFd or 40uFd, I'd make the first cap 20-40uFd and take the <1V ripple as a bonus.
> mic preamp and needs to be quit.
Yes, but brute-force is not the only way. A more elegant path is to start with a little ripple and reduce it in several steps.
How little ripple do we need? I'm not privvy to the schematic in question, how well it rejects ripple. But a pessimistic estimate is that supply ripple should be as low as tube noise. Tube noise is liable to be 2uV at the first grid, but that does not get B+ power. Tube gain is, say, 50, so the tube's self-hiss is 100uV at the plate. Triode stages without overall feedback have PSRR of about 2, so we could tolerate 200uV ripple at the B+ rail.
We have about 0.5V ripple at the first power cap.
0.5V/200uV= 2,500:1 ripple reduction needed.
If caps cost money, you usually get best economics by taking 10:1 reduction at a time. That says three more stages. Big caps are cheap, so we might take 30:1, even 50:1 per stage and not break the bank: 2 R-C stages.
We have a 10K resistor. Cut that into two 5K resistors. We need to be sure that the resistor from the first cap to the first BIG cap is big enough so that the rectifier does not gag on huge charging current: a 400V supply feeding a 5K resistor to a fat cap can only pass 400V/5K= 80mA, which a 6X4 can pass OK for quite a while. We then need a big cap that is about 5K/50= 100 ohms reactance at 100/120Hz.... 20uFd is 80 ohms at 100Hz, and will be fine. 40uFd costs so little more, we could do that instead, just for over-kill.
So a well-engineered plan is:
6X4-40uFd-5K-40uFd-5K-40uFd
This is roughly how the Dynaco did it. Worked for them. A phono preamp is just as susceptable to buzz as a mike preamp. Dyna actually spread the load over several filter stages: a full PAS has subsonic gain over 100,000 from end to end, and needs lots of power supply isolation to keep from motor-boating. No mike preamp needs that much gain.
There may be some audible benefit from having a BIG cap facing the amplifier. At least I like to do that. 6X4-40uFd-5K-40uFd-5K-500uFd seems like overkill to me, but I'm guilty of similar over-kill.