The U70 I built a while back had a problem. It was working fine last time I saw it, and left instructions to check the bias every couple of weeks. Then it came back with "glowing red plate" syndrome and a funny sound in the right channel (where the glowing red tube was).
On further examination I noticed that the right channel bias pot was turned up much further than the left. Ah, I got it, replaced the tubes all around (had a fair bit of time and EH 6ca7 are reasonably cheap), turned the pots to about the same, fired it up. Fine. Let it warm up a while, and set the bias. Still fine. Set my VOM to record high and low voltage. As I turned to walk away and let it go for a while, I hit my knee on the table. Immediately it beeped at me. Showed about half a volt as the min voltage. Checked the bias voltage again, it was back to normal. So I tapped it, sure enough I could get the bias to almost drop to zero if I knocked on it horizontally. Then I wiggled the tubes, and sure enough the sockets were loose , screwing up the ground for the bias - since it's terminated at the tube socket.
So I let it cool and removed the screws and slipped some sandpaper in between the "ears" on the sockets to sand off any oxidation and then screwed them back down tight. Set it back up, and now everything was pretty stable. But I got to thinking, and then took it back apart and ran a wire from the ground side of the bias resistor to the central ground point on each side. Now it doesn't care if the sockets are tight.
I figure what happened was that the socket got a weak ground, and the bias was dutifully run up to 1.56V. Then it got jostled, and it grounded solidly, making the bias very high. And then wiping out the output tube. By grounding it with a solid wire I think there's nothing in there that cares about the chassis ground any more and it should be more robust.
Word to the wise.
Brett