Ben,
Let's look at the gain structure this project needs. If you constant current source (CCS) load an ECC99 section, a voltage gain near 20 is a reasonable expectation feeding a triple Darlington emitter follower O/P stage. The driver should stay linear as it pushes the "finals" into clipping. So, the driver has to generate roughly 60 VRMS. Don't forget the NFB loop's requirements. 60/20 = 3. The full O/P of a "standard" CDP is 2 VRMS. Therefore, a line stage with some gain is in order.
I'm attaching a schematic for a 12B4 based line stage, which WILL get the job done. Look the diagram over and feedback on your observations. Pay particular attention to the pseudo choke I/P filter B+ PSU. The exact value of the "fudge factor" cap. has to be determined experimentally.
Some of the "denizens" of this board developed PCBs for use in constructing CCSes. Those or the
Bottlehead C4S will do quite nicely in both the line stage and power amplifier.
Oh yeah, look at the ECC99 data sheet I previously linked. 260 V. on the plate, -10 V. on the grid, and I
B = 10 mA. appears to be a suitable operating point. A B+ rail of "300" V. should be adequate to provide for the plate to cathode differential, the 10 V. of bias, and the compliance the CCS load requires. "Full wave" voltage doubling a Triad N-68X will neatly take care of the B+ PSU requirement. You follow the doubler stack with a LC section. CCS loading provides superior PSRR.
