>
I know that 6550s are rated to 450V
FWIW,
GE 6550A is rated 500V in Triode mode.
Numbers on 6550 vary between makers, and some fudge the numbers more than others. While SY is right to stay inside book-spec, especially on the Screen where abuse is hard to notice until the screen melts, I am a wildman and would not be TOO concerned running 11% over in home HiFi. 6550 is an industrial tube, and the ratings assume hard work with economic life in industrial service. They were run harder in some non-industrial amps, no real problem. And 6550 is again in multi-factory production... if you melt a couple, it isn't cheap but it isn't a Big Problem like it was around 1988 when we had to grub the trash to find un-dead 6550s.
>
if I use less than 100k for the 1st grid
Won't do much to reduce screen-grid abuse. However, notice that 6550 grid resistance is
50K in fixed-bias, 250K in self-bias.
Look. If you wanted POWER, you'd run them 6550s in pentode mode. Or at least UL mode. Using triode mode means you want a smaller nicer amplifier than an Ampeg SVT. And that usually points to self-bias. And that means "wasting" some of your precious B+ voltage. In this case you have B+ to spare and you
want to waste some.
Your underloaded transformer will end up around 500V DC. Pencil 450V across the tubes. 6550 plate rating is 35W or 42W depending who you believe. 75mA gives 34W, a nice safe first-guess. Squinting at the average Triode curves, this needs -50V grid bias. So drop 50V in a cathode resistor, 450 in the tubes, everything is groovy. Your transformer is in fact "exactly right" (except for being heavier and more expensive than it had to be). Figuring 150mA for two tubes, the cathode resistor is 333 ohms 7.5 watts, which could conveniently be three 1K 5W or 10W resistors in parallel. Or twelve 33 ohm 2W resistors (cheaper by the dozen) in series, and you short-out 1, 2. 0r 3 resistors to trim the bias for the specific tube you use (remember tubes vary +/-20% in grid bias for the same current).