[R-390] R390-URR BFO Problem

[g4gjl at btopenworld.com]

Im looking for group wisdom and experience with an intermittent BFO in my R390-URR.

The set is in average condition but is very hot, much better than either of my R390A-URRs.

The problem is that the BFO is intermittent.

The failure mode is a slow fade out of BFO signal, though no audible change in BFO frequency, until just before it drops out completely. This occurs about 30 mins to 1 hour after initial warm-up. I have found that removal of the tube will re-start the oscillator. Power-cycling the BFO switch will not re-start the oscillator.

My actions thus far are as follows:
1 Changed valve (tube) for a NOS British equivalent. Result:  No change in nature of fault.
2 Inspection of valve holder reveals no visible sign of tracking.
3 Inspection of valve holder reveals no obvious dry joints.
4 Inspection of valve holder reveals no obvious bent pins or loose pin sockets.
5 Last night I replaced all three HT feed resistors. Result no change in the nature of the fault.
6 Today I removed the BFO PTO and built a new oscillator to test the PTO on a separate chassis. I used the same valve but all new resistors and capacitors. I ran the oscillator into my HP frequency counter and it remained on and stable at 455.03 kHz  for 30 mins., it then 'failed'. Interestingly the 'failure' resulted in the oscillator squegging at about 84MHz. So I guess that is exactly what is happening when the PTO is installed in the IF strip. 
One faulty BFO  PTO......

Now the questions.
Does anyone have similar experience with this PTO, or any suggestions as to how they might fail?

 How I should set about opening and repairing the PTO?

 Does anyone have any pictures or text description of the opening procedure?

Looking forward to replies and thanks in advance for any help.

Pete
G4GJL