I couldn't find this thread earlier (it's been awhile since I worked on this) and posted a new topic "Split Supply for PP Amp" (merged and is now above this post - GM) i. Please ignore that one and use this example instead.
This GM70 thread won't die until I succeed or fail at building a push-pull GM70 stereo amp.
I have a new power scheme I ran across and I have drawn out my latest schematic for comments/criticism. Here's the latest:

I found someone who claims to have bread boarded a similar design that works. His didn't have the CCS on the cathodes and also used a transformer phase splitter. I don't want an interstage tranny and would prefer capacitive coupling. The phase splitter is at the input of the my amp design - the input is split and feeds one inverting audio op-amp and a second non-inverting op amp, at least one of which has adjustable gain to allow balancing the positive and negative driver stage gains. The op-amps will have a split supply with input and output at ground potential and there will be NO coupling cap between the op-amp outs (an OPA134) and the driver tubes as shown in the schematic - those are gone. So the entire amp will have only one cap in the signal path, and can be driven to full power (hoping for 75 watts / channel) with a 1 volt peak-peak signal.
Has anyone built an amp using a split supply like this or does anyone see a problem with why it won't work? The idea here is to reduce the voltage between the OT primary and secondary. Buy using +/-500 instead of +1000 volts the maximum voltage between the primary and secondary of the OTs will be about half what it would be with a single ended PS. The chance of failure (arcing from primary to secondary is my concern) of the Edcor OT I bought should be radically lower and it the amp should be susceptible to fewer of the problems associated with high voltages. At least that's the theory.
I'm excited about this.
If this works it solves the final remaining problem that concerned me greatly. I can't take credit for it though. I found the idea on another website. The heater supplies need to float, and although the output stage isn't self inverting and the amp uses a phase splitter, hopefully the CCS will have the same effect as it does in the Oddwatt to help linearize the operation and reduce the distortion.
Did I screw something up or does this look OK?

It's actually quite a simple design. A split supply is easy to build and it also eliminates the grid bias supply since the tubes are self biasing. I just hope that the 25 ohm pot works as well on GM70s as it does on the KT88s in the Oddwatt to balance the tubes.
Oh yeah, the drivers might be either SRPP stages or pentodes, probably an SRPP but I just drew triodes for simplicity.