Look up the "OddWatt" self-inverting circuits, these have several references on this forum. The shared cathode "resistor" is an LM317 or an IXYS IXCP10M45S or similar CCS regulator device. With a simple shared cathode resistor, there would be current and voltage fluctuations, and that would resultin some nasty distortion products.
From my experience with paralleling the two pentodes in each channel of a 6DZ7 SE amp, the grid resistor has to be 150K or less, with 200K and above, they get very unstable and start to oscillate at RF frequencies., due to the relatively high combined transconductance. This would need an SRPP driver stage, perhaps 5695 would be good, or a 12AT7, with no bypass cap on the lower triode's cathode resistor.
For PSE operation, I used an SRPP 6CG7 driver, operating at 4-5 mA, with RK+RA = 1K, with very good results, and an 0.47 uF (0.33 uF would also be good) interstage coupling cap. The Self-inverting PP or PPP design would require about 2X the driving voltage to the top 6DZ7's control grid.
These particular dual power pentode tubes (6DZ7 and the similar 6DY7) were intended and designed for PP Pentode service, with a single tube, then the two control grids are driven 180 degrees out of phase, it may be better to connect them as so that each 6DZ7 is connected in PP, rather than in parallel, as you have them, but then, the downside of that would be that you would not have the option of ultra-linear or triode-strapped operation, due to the SGs being internally connected together . Seriously, four separate 6BQ5s / EL84s (or 6GK6s) would be a lot easier to work with, and a lot more circuit-flexible. 6DZ7s ar eharder to acqire,and generallya LOT more
, too.
A 1K grid stopper for each control grid, and 100 to 220 ohms grid stopper for the screen grids is almost mandatory for stable operation.
HTH, hopefully this is not too confusing of an explanation, I could have probably written it alittle better
.
/ed B