EL34 glowing internally with signal...

[liquidair]

All components with leads will initially try to dissipate their internal heat via the leads to the outside world and to whatever electrical conductors the leads are attached to. 

Interesting. I guess that makes sense given the plastic body will act as an insulator. In my case I'm using cement resistors with leads that are more plates to stand them off the board.

I think I figured out what I'm observing and why heat is such an issue in this build...it's the lower voltages. The OT is a 4.2k Ra-a which pushes the load line further down the Vg1 = 0 curve than the normal 3.6k used with EL34s, but then with a Va/Vs of around 400 the load line intersects well below the knee (Vsat~50V). This is the region where screen current should skyrocket. At just 420V Va/Vs, the slope of Is is much flatter and further limited by Vg1=0. That should explain why the last build's screens were much cooler.

I turned the MV down just a bit last night and after 30 min or so the screen resistors were barely warm. Opened it up and within about a 1 minute they were too hot to touch. Back down again and they cooled down extremely fast.

It also explains why in the previous build, I didn't feel comfortable with how hot the usual 1k 5W part was getting. I remember thinking "Marshall's use a 5W part here on amps that have been abused for 60 years without issue, why do I feel 5W is not enough??" Well, Va/Vs of 440V and a 3.6k Ra-a OT intersect Vg1=0 where screen current will be much, much lower than where I'm at with 400V and 4.2k Ra-a. 

Thank you Kevin for all your help!