Bias for 6L6-GC in Super Reverb

[Fresh_Start]

Proof positive that you should not work on an amp when tired, let alone waste other people's time on this forum.  I am sorry!
Turns out the Intensity should never be turned to 10.  The tremolo signal is too strong.

AB763 Super Reverb clone built in 2009.  Apologies in advance - I'm very rusty at this since I haven't touched a soldering iron in several years.
Came back to me with the problems described here:
https://el34world.com/Forum/index.php?topic=23917.msg257233#msg257233
I replaced the GZ-34 rectifier with a Sovtek AR4 from Hoffman.  That lasted about 2 hours of playing time then appeared to fail "open".  IOW no voltage at all on the plate supply node of the power rail.  That's been replaced by a JJ GZ-34.
The voltage from the rectifier is steady with the amp on standby.  However, when I take it off standby voltages on the plate node and screen node are fluctuating a lot.  So is voltage on the cathode of the power tubes. 

There is a noticeable pulsating of the background noise now that wasn't there yesterday or even earlier this evening.  That's with all knobs at 1 and reverb disconnected.  The amp sounded fantastic until I started cranking the volume up gradually to 10.  (I was using an attentuator so I'm not deaf - thing is LOUD.)  That's when the pulsating started.
Edit to add:  It's not the power tubes or the rectifier tube.  I put the old ones back in, power tubes first, then the rectifier tube and the pulsing is still there.  On standby, voltages are steady.  Without power tubes in, voltages are steady all along the power rail and the bias supply.  Put in power tubes, either set, and the pulsing noise and fluctuating voltages are there.  The only thing I can think of is the plate node filter caps.  They did get a hard short when the standby switch rotated into the chassis.

Any other ideas?

The replacement 6L6 power tubes are NOS Sylvania, US manufacture.  Cathode current was 46/42 ma so a reasonably well matched pair.  Plate voltage 470, so cathode dissipation about 22 watts on the hotter of the 2 tubes.  Screen grid dissipation about 5.5 watts. 

[So did I have the bias too hot?  I thought that dissipation equal to 70% of rated wattage (30w for 6L6-GC) was a reasonable level.  After subtracting the screen grid dissipation I'm well below 21 watts.]

It's not the power tubes.  I bumped up the bias voltage to -60 and put the old tubes back in.  Voltages on power tube plates and screen grids are bouncing around.  Maybe 30 volts?  Hard to tell with DMM.  Voltage across the 1 ohm resistors at the cathode is bouncing by about 10 mv.  Bias voltage is steady at the "supply" before it goes to the Intensity pot; however, I'm not reading any voltage on the control grids.  I don't think it's zero volts, but it's really odd.
Sorry for rambling, etc.  The amp's owner is coming tomorrow to pick it up.  I thought I was all set and just wanted to run it up to 10 as a final test.  CRAP!!! 

[John]

The only thing I can think of is the plate node filter caps.  They did get a hard short when the standby switch rotated into the chassis.

That would be what I'd change out first, if it was me.

[PRR]

+1 to filter caps.