The hum is at 120Hz. It is low enough so that on a typical bookshelf-monitor, it is below the level of typical phono hum when the volume on the preamp is set at normal listening level. However, on high-efficiency speakers it's audible at the listening chair. It occurs even when the input jack is shorted.
I first swapped the 6AN8 tube for one that was in one of my "quiet" amps. No change.
I then tried a set of EL34 output tubes from a "quiet" amp. No change.
I tried another 5U4 rectifier. No change.
I lifted the coupling capacitors from the triode section of the 6AN8 to the grids on the output tubes. No DC leakage on either one (the originals have been replaced with Sprague Orange Drops). So, now the only thing the grids on the EL34s were seeing was the bias voltage. No change, and this would, I think, eliminate the input board as a cause.
I swapped in a new silicon diode and filter capacitors into the bias circuit. No change.
The power supply has been beefed up considerably on the underchassis, with each section at about 150-200uF, even the first one, which is too large as per "conventional wisdom". Each section seems OK, no excessive leakage when checked with an ohm meter, and no improvement when more capacitance was jumped across them. I lifted the first section and tried 30uF, as per "conventional wisdom." The hum got WORSE.
I installed resistors, 100 ohms/1 watt in series with the screen grid leads from the output transformer, as a suggested stability measure. No change. I then used the same resistors to connect the amp in triode mode. No change. I put them back in series with the screen grid leads.
I installed separate bias controls for the two output tubes, since I was planning to do that anyway; and separate cathode resistors at the output tubes. Adjusted the EL34s for balance by meter AND tried to do so for minimum audible hum, but there was no improvement when adjusted optimally.
I reconfigured the grounding scheme to a star ground at the quad filter capacitor, which is no longer in use but is still mounted to the chassis. No change.
The amp already has a three-wire power cord. I lifted the ground. No change. I swapped the polarity of the cord. No change.
I can't think of anything else but an unbalance of some sort in the output transformer, which would obviate the entire project. Any more ideas out there?