Tube amp problem... I need help!

[R.G. Keen]

...
>four EL34s in the output stage. It came to me wih 'blowing fuses
>continously'. I ran it up on the bench, via a variac, and found two of
>the tubes, the top pair, were running red hot. You know that glow, light
>up a small hall type of thing. Checked the bias voltage, all was OK on
>all tubes. Marshall amps are set to be biased at -40V. According to JMP,
>there is no need to set individual current. 
That's OK if you're not too fussy about sound and can get highly-similar 
production tubes - which you can't do very well these days.

>But after a long run the older pair started to glow again. Replaced all
>four with a matched set. All bias present on tubes and no tube getting
>too warm even under load. So it seems OK now... but... The thing is,
>surely all four tubes can't go at once, or is there another subtle
>failure mode that I haven't understood??
Actually, all four tubes could go in a relatively similar amount of time.
tube life depends on how hard they're pushed and how hot they get, among
other things. If the old set was abused similarly, keep too hot and run 
at maximum warp all the time, their lives would be similarly shortened. The
life shortening would also shorten any differences in life between them. The 
more they were a closely matched set before, the more they'd die together.

Just to be sure, test - or just replace! - the grid leak resistors, screen 
resistors and coupling capacitors that lead to the output sockets. Those 
are the things most likely to be involved in runaway of an output tube 
after the bias supply is right.