KT77 Bias problem in JCM800 (redplating)

[figuringstuffout]

ok, so I have another problem that is destroying my mind.  Here is the sequential progression of events.....
-I built a JCM800 (100w) a few months back. 
-Running EL34's, everything is great. 
-I get the brilliant idea of running some KT77's. 
-Install, rebias (70% of max  dissipation), crank up, drool cause it sounds amazing. 
-Record and smile lots cause it sounds amazing. 
-Hear a loud pop followed by a significant loss of volume.
-Immediately look at the tubes to find that the two closest to the PI are redplating like nobody's business.
-Turn the amp off and take it apart to figure out what went wrong.
-Check bias with standby on and find the bias supply is acting normal before it gets to the 220k resistors feeding the grids of the KT77's.  The two tube sockets away from the PI have correct bias numbers, but the two next to the PI (the ones with the formerly redplating tubes) was down at -10v...way too low.
- Immediately suspect the cap coming off the plate of the PI (right side) and replace it. 
- Bias problem goes away temporarily.  EL34's still work great.
- Get some new KT77's....everything sounds great again.
- They suddenly redplate again after a few uses (same tubes next to the PI)....not on long enough to damage them cause I am watching closely and quickly shut it off.
- I now am very confused because sometimes the bias is way too low for the same two KT77's and sometimes not.
- I check every resistor anywhere close to the bias supply and in any way close or connected to the KT77's and find no faults.
- Install the EL34's again, rebias.  No issues.
- I read somewhere on the internet that JJ KT77's have a little problem with the size of their pins....they are a touch too small and could be losing contact with the tube sockets.
- I remember that I bought Weber tube sockets for this amp (little cheaper in quality potentially) and realize that indeed the JJ KT77 pins are looser in the sockets than the EL34's were.
- Tighten the socket holes until the KT77 pins are much tighter.
- Fire the amp up again and it sounds great.  Bias is fine.
- Two weeks later, same pop and drop in volume...tons of redplating happening in the two KT77's next to the PI, yet again....the other two KT77's are fine as always
- Tons of profanity and bashing everything close to me.

Sooooo.....I thought I had the problem fixed, and am pretty annoyed at JJ for the stupid pin issue, but obviously I have either not discerned the correct problem, or didn't tighten the pins enough (which is hard to believe because they were tightened a lot).  I guess my next step is to rip out and replace the entire bias circuit, all connections (resistors and caps) in the PI and around the output tubes, and replace the tube sockets, but I'd rather not do that if it isn't necessary.

Any ideas I have not considered?
Thanks for the help....

 

[abbey road d enfer]

Have you measured the same wrong bias voltage on the grid resistor (the 5k6) not on the tube pin. It seems impossible that the voltage be so different from one set to the other.
Have you checked the value of the 5k6 res?

 

[figuringstuffout]

Have you measured the same wrong bias voltage on the grid resistor (the 5k6) not on the tube pin. It seems impossible that the voltage be so different from one set to the other.
Have you checked the value of the 5k6 res?

thanks for the quick response!

I have not checked the bias voltage level on pin 5 directly actually, but on the connection of the PI plate cap and the 220k resistor from the bias supply that runs to the 5k6 resistor.  I actually didn't check directly on the pin because man it is crowded down there...lol

I have verified all 5k6 resistors repeatedly.  Could resistors fail intermittently?

 

[abbey road d enfer]

Have you measured the same wrong bias voltage on the grid resistor (the 5k6) not on the tube pin. It seems impossible that the voltage be so different from one set to the other.
Have you checked the value of the 5k6 res?

thanks for the quick response!

I have not checked the bias voltage level on pin 5 directly actually, but on the connection of the PI plate cap and the 220k resistor from the bias supply that runs to the 5k6 resistor.  I actually didn't check directly on the pin because man it is crowded down there...lol
I have verified all 5k6 resistors repeatedly.  Could resistors fail intermittently?

Not very likely. My understanding is that one side doesn't get the correct bias. It can be many things. since the problem occurs with the KT's and not the EL34's, it may be grid current, but that would change side if you swap the tubes. Have you actually tried to move the tubes around?
On the EL34, g3 is connected to pin 1, on KT77, it is internally connected to k; shouldn't make any difference...puzzling.

 

[Sorr]

If one valve of the pair loses bias through an internal fault then it will pull the other one down.
Having double checked the 220k bias feed resistor and its soldering then I reckon its one of your KT77's developing a fault . Have you measured its Grid to cathode leakage? (this can change over time) I assume its the same two KT77's not just the ones on that side?