Concerning the "extreme position" placed on the driver stage with the use of output stage plate feedback, the related issue I'm mulling around is the use of fixed bias on the output stage. Fixed bias requires strict attention to the output stage's grid leak resistor size. 6550 data sheet says 50K max for fixed bias, while KT88 says 100K max for fixed bias. This value can be raised quite a bit if using cathode bias, to 250K for the 6550, and 100K for the KT88. The driver tube working into a 250K or even 100K load would be much easier than working into a 50K load. But, as I mentioned, I am not interested in putting a hot cathode resistor under the chassis. So I'm going to have to live with the more difficult 50K load demanded by fixed bias as well as being compatible with either 6550 or KT88 types.
What I really need is a driver tube with a bunch of gain like a 12AX7 but with a bunch of current capability like a 6SN7, and having that tube be a more common 8- or 9-pin type that ideally is available in new production. I don't think there is such a thing....
I've been thinking then about relaxing this requirement, meaning using any tube I can find either new production or old stock that has a mu say of 50 and able to handle 10 mA current. This kills tube rolling choices though, but maybe it's the lesser of the evils.
Another idea. Using a single triode, the flatter I can make the driver stage load line, the more gain I can realize, limited by mu of the tube. Maybe I should consider a CCS plate load. I could get nearly 70x gain with a 12AT7. That would allow the application of more feedback to get to my targeted 10 dB feedback range.
Another idea. I could use a compound driver with a stacked triode topology. Something like a pseudo current source with another 12AT7 triode, or a SRPP, Mu follower, or even a cascode. Any of these approaches would also increase available gain. Don't really like any of these approaches because of the need for elevated heaters, but I'm not ruling them out yet, either.
My very strong inclination is to keep things simple. So I'm guessing I will end up sticking with the common cathode topology, either 12AT7 (preference), or another old stock tube that meets the high mu, high current capability requirement. If I use a CCS, it will be discrete silicon based (thus avoiding the elevated heater issue). Probably not interested in the IXYS 10M45 current source. I've tried it before and I don't really care for it. Just personal preference on that.
So It turns out, I'm finding that designing a decent SE amp may be a lot harder than designing a decent PP amp. Relatively speaking, PP amps are "easy" to design well, in my opinion, because they implicitly solve a lot of the most plaguing problems of SE amps--less expensive power supply, noise and distortion cancellation, higher efficiency, more power with less iron weight, inverter can act as a driver and can give you a crutch of a little extra gain. Affordances that are readily available in PP amps are non existent in SE amps.
Back in the mid 90's I met Bill Conrad (of Conrad Johnson). He came to a brick and mortar re-seller store in Salt Lake City for a routine visit. I happened to be in the store at the time. So I asked him why CJ did not build SE amps. He gave me a long list of reasons th[poat are mostly summarized in the above paragraph. Now that I build my own amps, I can see his point much clearer than I could back then.
So anyway, enough musing. Back to figuring out how I should move forward here...
Here is what "I" think:
The best sound in a SE amp comes from as simple, but as
well done design as possible. Schemes such as SRPPs, Cathode followers, three stage amps, and CCS should be AVOIDED like the
plague, if you want the best possible sound quality. CCSes and SRPPs are only good for people who do not know how to build a good supply to start with !! A crutch, a band aid, that will sound "forced", compared to an optimized simple implementation.
Heat under the hood from a cathode bias resistor is NOT as bad a choice as YOU , or 95% of the people in this world, would make by employing a fixed bias supply. I have never seen a great fixed bias supply !! Anyone who
could design that, will chose self-bias instead.
The fixed bias supply you build will be ON THE GRID of the KT-88 and it will NOT be good enough supply, in its design, and in its parts quality !!!
Rks, heat : My upcoming 6AQ5 DC stereo amp will have 177.5 VDC on each cathode ( elevated due to DCing ) at 35.5 mA., so that is two resistors in a stereo amp, each dissipating 6.30 Watts, , or 12.6 Watts total.. So what ?? I have ALL the advantages of cathode bias in reliability and in sonics !!! Each Rk is 5K, 1%, wirewound, a single 50 Watt chassis mount, with thermal grease, so it is operating at 12.6 % of the R's rated power. Far better than adding a complex, and mis-designed fixed bias supply to the grid !! See Rks below, with three dimensional layout and short signal paths employed, not yet fully evident.
I chuckled when you dismissed my driver tube suggestion, because it is no longer in production. You are an amateur DIY builder, like me, and you are likely older, like me, and NOT going into production. WHY is current production tubes a requirement. You could buy four 12BZ7s ( I like Raytheons) over a few months, and your GRANDKIDS will still be able to use the amp with those four tubes you buy in 2018 !! How about, the BEST tube compromise, as a choice ?
The 12BZ7 IS a compromise., in that it does not have great, linear tube curves, when you look at large signal swing. A typical EE will poo poo it, dismiss it. A good DIYer, on the ball, will not dismiss the tube. Its a great mu of 100 driver, sounds GOOD, robust sounding, nice large plate and cathode !! I like it around 2.2 mA and 190 Ea by ear. I have the cajones to use it, others won't bother. LOL.
Recall, the Roger Sander's QuickSilver KT-88 amp that was SO GOOD at the Azzolina room at RMAF years ago, was 1/2 a 12AX7, cap coupled with a small uF REL Teflon coupler, to the grids of a KT-88 !! You should talk to him, not amateur DIYer me.
All of this is in my opinion, I am sure you realize that, and that we can discuss this on Forum.
Dowto1000