Apologies for sounding a little crabby.
I looked again at what the problems with the original were, since it's been a long time. Here they are:
The original used a 3K transformer, the BH-5. A great transformer, but the low impedance makes the damping factor very poor when used with the 45 tube. It seriously restricts the variety of speakers that will work well. I feel that 4K to 5K is the sweet spot for that tube, when restricted to the specified maximum voltage. The PC-3/OT-2 combination at 4K impedance would be more appropriate.
In the original, the biggest available clip-on heat sink (on the C4S feeding the shunt regulator) limits the maximum power line voltage to 120v, whereas most North American urban systems provide between 120 and 130 volts. So for most people, the heat is such that the regulator transistor will fail in a few years. (This is based on our experience with the FP-III, it's not just conservative engineering!) The original added that heat sink in a clumsy way since the PC board did not allow enough room for it but even with the current (v4.4) board the higher line voltages will be a problem.
The shunt regulator current is barely adequate due to the limitations of the (now out of production) PT-2 power transformer. I'd much rather use the more recent PT-7 which can provide another 10mA, nearly doubling the SR current. This will provide an improvement in low frequency power bandwidth.
For an amp of this sonic quality and low power, it seems to me inappropriate to me to use AC filament power. This amp will usually be used on very efficient speakers, so the residual hum will be a problem in many applications. Some old-stock 45s have lower hum due to a design that counters the filament voltage hum with the magnetic fluctuations of the current paths, but that technology has not to my knowledge survived past the Thirties - it's not there in modern production tubes, some of which are good candidates for this kind of amp.
Honestly, the TV vertical sweep tubes used for driver/shunt regulator duties are not free from heater-induced hum. Either an elevated AC supply, or preferably a DC supply, would improve that problem.
For both heater/filament supplies, the considerations of power line variability lead me to wish for regulated supplies here as well as for the high voltage. EML for example insists that the best performance is achieved by keeping the filament voltage within 3%, and - especially when operating the tube at its maximum ratings - I agree.
