Since posting this I discovered that my "feedback" problem was an NFB phase issue. So problem solved.
I don't have experience you guys have (obviously) but I have been studying the LarMar PPIMV a bit more and discovered that the value of the pots is indeed important. With 250K pots and their 1M5 to 2M2 parallel resistors is an aproximation of the normal 220K grid leaks (aka Bias feed). Using values that would parallel out to much more than 220K would subject the power tubes to more input than they're meant to have, possibly more than they can handle with catastrophic results.
So 250K pots and 1.5M - 2.2M parallels work out fine,
if you have the normal 220K grid leaks. I've got a 250K ganged pot. Unfortunately, the Jana 2204 mods I'm trying use 100K grid leaks. I would think these are for tone issues rather than to safeguard the tubes. But wanting to preserve the tone, I will use 180K parallels instead of the usual higher range. That works out to around 100K.
However, this brings up something else. Jana's mods start with a recommended HT of 375V. My PT is a bit hotter than that so to get the preamp voltages down I added 8K to the dropping resistor between the screen voltage and the PI node. Works great, but now I'm wondering. Are Jana's 100K grid leaks related to the 375V HT? If so, how does my ~425V HT effect that? Should I be using the higher grid leaks? And if I use the LarMar, adjust the values there too?
So my question is: are the grid leak resistor values set according to the HT?
tia,
Ken