> A 6L6 has an Rp of like 25kohms. A 6AU6 has an Rp of 1-1.5 mohms.
> the rp was close to that a 6V6
In Pentodes, the plate resistance is utterly meaningless in audio applications, resistance-coupled OR power amplifier. We just can't use it.
It's like the Collector Resistance of a BJT. We might estimate that for a DC voltage amp, but in audio and Power use it just does not matter, and is rarely mentioned.
> I wonder if the output transformer primary would be the limiting object for maximum power xfer and freq response?
Not really. The transformer is shunted by Rp||Rl (and the ratio). Triodes are transformer-friendly: their low Rp will maintain bass even when inductive reactance falls below load impedance. In Pentodes, Rp is always far higher than Rl, and huge changes in Rp make no difference to the transformer. We design the tranny for Rl, ignore Rp.
Instead of Plate Resistance, in Pentodes it is often useful to know the Screen Impedance. This is equivalent to the plate resistance of the Pentode when triode-wired. If you don't know that, look for the G1-G2 Amplification Factor (usually between 4 and 40, 10 for many audio power pentodes) and the Cathode Resistance 1/Gm. For 6L6 it will be about 2K; good loads will generally be somewhat higher, say 3K to 5K.
> 6AK6 ... 6CA5
These represent two compromises. 6AK6 likes fairly high plate voltage and very low plate current. This means very low heater power, but requires hi-Z loading. The 6CA5 has a much bigger heater demand, but that allows low load impedances. Low-Z windings are easier to do well/cheap than hi-Z windings. So do you want to buy fat heater iron or fancy output iron?
Low power, cost, heat is not a new goal. The kings of this field, in the 1-2Watt range, are the AC/DC radio tubes. 50L6 is not a different 6L6, but is a lot like a cross between original 6L6 and EL34 but scaled-down pint-size. And the "L6" name looks good on guitar-amp bottles. If you want something smaller, 35C5 is a small but mighty tube. Tubes like 35C5 exist at 6V heater. Might be fun to scale a Fender Twin down to 1/10th power: change 450V to 150V or 100V, keep almost everything else the same.
As a general solution to getting "tone" without neighbors playing rhythm on the wall: vary the B+. NYD, you have a variable DC supply(?): wire the little tubes and start ramping from +150V down to +20V. You may find that +50V is plenty loud for the bedroom, even in recording. And at the few-Watt level, you can use a big MOSFET to control the B+ with a pot.
> push-pull pentode or BPT output sections with global negative feedback.
Very light feedback. Maybe 3dB-6dB at nominal load. What it mostly does is control the uncontrolled rise of gain at bass-resonance when a loudspeaker is driven by pentodes. You can also let it be uncontrolled, but that suggests a different speaker choice, and as you say the light-feedback plan has been awful popular.
> Not enough preamp gain
One bottle is hard. There is a 12-pin TV IF tube that might go there.
In two bottles: 60FX5 is king of high-gain 1-Watt tubes. You can build a stereo record player with just a coupla these. A very high output ceramic needle makes almost enuff signal to saturate the output. There is even a 12V version, and I think I've seen it stocked somewhere.
You still need gain of 100 plus tonestack loss. This either means a high-gain pentode and low-loss stack, or good old 12AX7 2-stage preamp. Fender's Champ explored this decades ago. He had most-all the same tubes you have, and motivation to think hard about minimal amps; you probably won't find much that is good that he didn't do before you.
One other idea: bringing the pickup up to a couple volts is not a "tone" job, but it is a noise problem. Hide a 5532 in there. Gain about 10. That can drive a triode into overload, cover some tonestack loss, and still smack a power pentode.