I tried to feed the screens at 2nd node. I still get 400v there !
But now i can't bias the tubes... At full bias i'm at 20mA. Don't want to alter the bias circuit...
Did you turn the -bias pot both ways? 'Full -bias' is the most negative setting, least mA setting. You need
less negative dcv bias to run the tubes hotter.
Your probably getting 400dcv at the B+ filter cap stack mid point because it also changed the power tubes current draw. Less load on the B+ PT wind.
Lower current through the power tubes the B+ dcv rises, higher current, it drops.
You'd have to measure the B+ at the power tube plates to see how much it rose.
You'd only have to change 1, maybe 2 R's to decrease the -bias so the tubes run hotter. I'd probably try dropping the 47K to maybe 20K or 15K?
You might not even have take the 47K out, just alligator jumper cable in another 47K in parallel with that 47K. That would give you 23.5K to test it with.
Take the power tubes out, keep them out, measure the max and min. -dcv bias at the power tubes grid pin, write it down. Then clip in the 47K and measure max/min. -dcv again, write it down. They should now both be lower, less negative. With the -dcv bias still set on max, most/highest negative setting, pop the power tubes back in, then try to bias them hotter, less negative dcv and see where your at.
If still not enough, and this might be better anyway, double or triple the value of the 15K that's between the 2 @ 50uF -bias filter caps. And leave the 47K alone. You might have to do both? It just easier to clip in across that 47K for a test, so no soldering.