Regards, Russellc

300b driver
300b output (PP so atleast 2 300b)
300b regulator

should be a pretty expensive amp.

Russellc

An 8k a-a with 20W of core rating would be my choice. 10k a-a is twice the SE load of 5k and cuts maximum power just a bit more than 8k a-a. E-Linear taps at ~10% and B+ ~370. Higher load would allow a higher B+ w/o leaving class A and leave more volts for the front end. Keep it absolutely Class A bias, perhaps even to the point of limiting power a bit. 12 or 14 Watts is hard to tell from 17...

For the front end, a differential pair of 6C45's look quite good. Hard to get a matched pair I am told, but I have heard them built in PP DHT amps to good effect.

Lots of options for sure, and with good OPTx design, it will be hard to make it sound bad...
regards,
Douglas

I have two pairs, bought as "matched", but for the differential "work" they probably need some "trimming" (high gm = small change in Ugk=large change in Ia). Fantastic tube, high µ and gm, 1mm thick Ti-anode, Pt- grid wire, very linear, but needs some attention to avoid "transistor sound" (k5) - better run on lower current (by Pete Millett measurements). Then oscillations are higly possible, maybe best to use "stoppers" on both grid connections, and all four cathode connections.
It`s high on my list too (300B driver), but too many other options to try, and I`m too lazy...
Another not so good feature is that it needs bias in the -2...2,5V area, broadly speaking, and cannot "accept" full 2Vrms signal from CD source... And we need about 2*55Vrms input (min.) in 300B pair...not so easy even for 6S45Pi in diff. driver. But, one stage driver sounds good:-).
At least, if nothing concretly, I posted the picture:-)

With the -2V of cathode-grid bias on them, the input headroom will be twice that assuming a few idealizations( like perfectly linear valves ). apply a 1V signal ( say a DC voltage ) and the cathode voltage will increase by 0.5V There will be a +/- mu/2 plate voltage change on the anodes( the input section will go - and the grounded section will go + ).

The anode-anode voltage change will be mu*input voltage. Accounting for nonlinearities, most of which cause 2nd order distortion, the actual headroom will be slightly different.
regards,
Douglas

Yes, I described my experience with SE driver, and forgot that diff. driver is totally different beast. About half of the input voltage we`ll "see" at the cathodes, you are right. (The last time I tinkering with "long-tail" circuit it was on the guitar amp about 5 years ago:-)).
I forgot one other thing - diff. pair cancels even and sums odd harmonics, probably not a good thing with already "difficult" 6S45Pi?
Well, YOU must try it in the real world, and see/hear
I used rel. low values of grid-leak resistor, too.
But, with 2Vrms input on the grid of V1 we can expect about 40Vrms per anode out, too low for 300B.