partial feedback

[dave dove]

after numerous problems i am knocking partial feedback on the head
here's the articles that can explain the theory much better than me:

http://www.tubecad.com/march2001/

and olsher's t rex:

http://www.enjoythemusic.com/magazin.../0105/trex.htm

for some reason i keep getting a voltage on the 300b grid
it is a gradually increasing phenomenon getting up to over 150V
and popping the cathode bypass cap
there is a distortion which increases with time
presumably as the voltage on the grid rises
takes a few weeks

i thought at first it was the 2uf cap
in a short loop.. 300b anode to grid
-as i'd done it the t.rex way-
but i've tried with different caps
and now i've now tried a 100K feedback resistor
to driver anode and the same happens

on one occasion i could follow the measured voltages
when switching 300bs which suggest a valve problem
but when i remove feedback the problem is gone

i don't know why this is
i can post some thoughts though:

the valves are faulty
but why only when feeback applied?
is it because of the aikido driver configuration?
or the resultant low driver output impedence?
whatever
i've got all these half realised
half understood ideas about this
but it doesn't work in this house