Strange 6L6 bias problem on '59 Premier 50

[terminalgs]

Here's the schematic:

http://www.dougcircuits.com/premier/premier5-6L6version.jpg

This was an ebay "AS-IS" project amp.   arrived missing tubes, speaker, as well as C2 and C5 coupling caps.  After doing a lot of testing of the circuit (checked all resistors values, drawing the schematic from the actual amp (reverse-engineered),  check the transformers, checked all grounds,  etc...)   I did the following:   

1. created an artificial center tap with a pair of 470ohm resistors, and elevated them to the cathode of the 6L6. (originally, one leg of the 6.3 was tied to chassis ground).  I know 470 is high, but it should be okay, right?

2.  installed .1/300v caps for C2 and  C5 (I found someone else that reverse-engineered the schematic, and he indicated .022's, but I wasn't entirely sure *he* was sure about that).   I stuck .1's in there  and calculated the bass roll-off freq. at 2Hz for C2, and 5Hz for C5 (with .022's they'd be about 10Hz and 24Hz respectively),, so I figured .1's would be okay.

3. installed tubes and speaker and fired it up.

The amp was remarkably quiet (I expected the old cap-can lytics to be dead, and figured it'd hum and motor boat).    the voltages on the schematic are as measured. 

Here is where it gets weird: with volume and tone at 100%, the amp is stable, and sounds fine.   if I roll the tone knob back to 50%, the amp becomes unstable:  6L6 cathode voltage goes from a steady 17V to a fluctuation between +30VDC and +15VDC,  and grid goes from a stable 0VDC to a fluctuating -40VDC to 0VDC.  It sounds like a tremolo.   The voltages tell me the tube is going into cut-off and then recovering.

here's a clip:

Premier 50, strange bias problem???

Really, if the volume is rolled off 100%,  the tone control can be anywhere and the amp is stable.   

If both controls are at 100%, its stable.   Its only when volume is at 100% and tone is rolled off that is becomes unstable.   There are two way to re-stabilize:   turn tone back to 100%, or roll off volume.   You'll see it in the clip.

The  problem occurs at approx. 30sec. mark.

I've tried a couple different 6L6's.

I think its a 120Hz cycle, but I'm not sure..  maybe the 6L6 cathode bypass cap?  maybe the elevated, center tapped heater supply?


more intriguing,,why does this happen with regards to the tone control setting??

[PRR]

It has a supersonic oscillation.

Probably from 6L6 plate to the volume or tone pot _wiper_ lead.

If the wiper is all-the-way to either end, that lead is damped enough to kill the oscillation. But at mid-point it will accept stray signal. Stray from the BIG voltage at 6L6/OT will oscillate.

Once working more toward the radio band than audio, the BIG swings are upsetting bias, and may also bring out a 120Hz from the overloaded power supply.

Look how that wiper lead runs. Chopstick it. Keep it away from the plate/OT lead.

Don't get shocked in there. Insulation is older than my plowtruck.

[terminalgs]

I played around with this a lot.  the pots do sit right above the 6L6 base and the PT.  after a lot of unsuccessful chopsticking,  I took to repositioning the top panel with the pots, trying to get a change out of the oscillation.  fruitless.  so, I took it out of the equation completely by removing it, and wiring a vol pot between the two stages near to the 12ax7 (vol pot to C2 in schematic).   No change to oscillation,  but now it would come on with about 80% vol pot (before, it'd come out with tone pot fiddling, but with no tone pot, itd just come out at high gain).

Next I tried a grid stopper on the 6L6, first a 1.5L and then a 4.7K.  no change.

finally, I thought the bargin bin Sprague .1/300V 410P's might be the issue, so I yanked them in favor of some 715P .022/600V's.  this took care of the problem.

Don't get shocked in there. Insulation is older than my plowtruck.

That is, what insulation happens to be in there!   Its sort of the amp version of that tree trunk with the scorpion in Flash Gordon.