Step 7: The Heater Power Supply(s)
The trannie also powered about 12 tubes, and the filament voltage is ~7v, and doesn't drop enough under load to get near 6.3V (load isn't big enough.) In fact, one 12AX7 "went nuclear" and burned out (~$25 "down the tubes".)
Placing two large diodes in parallel but opposite directions in an AC supply limits the voltage by the voltage-drop amount (.5 to .7V), just the same as one diode in a DC supply. That dropped the filament voltage right to 6.3V, and the tubes were happy.
The two-diode trick works only for AC--current flows through one diode at a time, dropping that half of the waveform by the diode's voltage drop amount. One diode would do the trick for DC.
However, they weren't quiet. You really need separate windings to setup a false center tap, which can be used to quiet the heaters.
After trying various solutions, I decided to light just the power tubes with the main transformer, and use a cheap "wallwart" for the 12AX7 preamp. Now the preamp has it's own, "dedicated," DC supply This was very quiet, indeed. The wallwart was already on hand.
A ground reference (false center tap) was provided by bridging the 6V by a pair of 180 ohm resistors, tied to the chassis ground. It does make a difference.
For some power transformers, the required 2.7A @ 6.3V is a little much. Many are rated for 2.5A max. Of course, an extra 200 MA might be fine, depending on the transformer. But a separate DC supply for the preamp isn't a bad option.
OK, this is a little, ehem, unconventional, perhaps even ghetto. But it works well.