Here is from my thread couple of years ago: http://www.diyaudio.com/forums/tubes-valves/183329-one-more-4p1l-se-5.html#post2495250
Maybe I'm just thick, but what I see in the "shunt cascode" is a common base transistor amp. That's not "all DHT" to me. Perhaps it's nice, but it's not pure DHT.
True and neither is Wavebourne's 4p1l design pure DHT, I think we've shown that a pure DHT isn't going to be a significant improvement on the classic single stage Espressivo or Cooter's pure dht schematic on the first post.
Also where do you draw the line, the cammmon base transistor is feeding the output tube , so does a mu output gyrator on the first stage. The central theme is a DHT output, everything else is auxillary to that. Never said there would be no transistors in the amp, I personally don't want "pure" if it mean a 100lb headphone amp. Heck the coleman transistor regs are just as much part of the amplification as anything, I certainly don't want pure if it means only ac heat or 20 lbs of choked down to dc filaments.
So at this point we have to stray from the project definition if we want something better than 0.2 W. Sounds like you've ruled out a stepup input transformer so either that leaves the common base transistor, idht's, or a three stage amp. Pick your poison.
Pure something can be some kind of a "pure exercise". I prefer to exercise in getting optimal design in terms of set of criteria needed for the end purpose, in this particular case, for the enjoyable music listening. However, "pure no compromise" some selected parameter may be good for marketing purposes, though. 
Running it at full tilt (Pa=10W) you can get:
A1: Po=2W biased at 250V, 80mA, Vg=-20V Vin=40Vpp for a Zaa=5K and THD=.9%
A2: Po=3.2W biased at 250V, 80mA, Vg=-20V Vin=50Vpp for a Zaa=5K and THD=1.2%
CHeers
Ale
If anything can be employed to achieve the goal then you've left out a whole class of solutions: those with no transformer on the output. You can easily get outstanding THD with a DHT input and an SS buffer. A 4P1L into an F4-like buffer or any other good buffer will drive rocks and speakers and easily the HE6.
It seems to me everyone chooses whatever they are comfortable with, and I see nothing wrong with that. You are comfortable to use a three stage amplifier of which one stage is a common-base transistor amplifier. Wavebourn is comfortable with an extra cathode follower loaded on a mosfet based CCS directly coupled to his parallel 4P1L stage. Where do you draw the line? The way you put it above, no more lines can be drawn. Everything goes.
Yes, (I left off a factor of 2 of my calc) but don't discount the copper loss of the secondary, it is a solid 1/3 or more loss formed by the resistive divider from the secondary DCR to the voice coil. This is why parafeed can be such a good solution to headphone secondaries.
If anything can be employed to achieve the goal then you've left out a whole class of solutions: those with no transformer on the output. You can easily get outstanding THD with a DHT input and an SS buffer. A 4P1L into an F4-like buffer or any other good buffer will drive rocks and speakers and easily the HE6.
It seems to me everyone chooses whatever they are comfortable with, and I see nothing wrong with that. You are comfortable to use a three stage amplifier of which one stage is a common-base transistor amplifier. Wavebourn is comfortable with an extra cathode follower loaded on a mosfet based CCS directly coupled to his parallel 4P1L stage. Where do you draw the line? The way you put it above, no more lines can be drawn. Everything goes.
Yes I agree there are many ways to skin this cat, this is what makes this so much fu. I actually have a commercial hybrid heaphone amp now with a class AB Mosfet ouput that gives an honest 4W ito 38 ohms. I considered a 4p1l to F4 buffer but its really a P-P amp. I just prefer a transformer coupled output SET, there are several technical reasons for this but for some reason an OPT just sounds better to me: Valve / Solid-State Amps
One thing I'll throw out there for the 4p1l to 4p1l IT coupled proposition. The Ra is low enough that a good 1:2 IT could be built which would eliminate the sensitivty issue/A2 drive issues: || DHTRob - Tubeamplifiers and fullrange units: A 1:2 interstage by Tribute ||. This would truly get the most outof an all 4p1l SET.
The question would be what is the price and lead time. Ayone have experience with Tribute ?
Link provided (DHTRob) Tribute's Euro 190 p/pair, I don't know is actual prices & lead time. If prices are right can be the way to go for OPTs.
Iko,
From triode models over real curves. I will send you the screenshots when I get home if you are interested
Ale
This is the best approach, I agree!
In every case I choose the best device for a given circuit position. JFETs, MOSFETs, BJTs, triodes and pentodes can all be used for very successful audio design (ie, can make the music enjoyable).
OTOH, any device can sound really terrible, if wrongly applied. The only "secret", (if any secret exists) - is to find out why it sounds bad, and solve the problem of the bad sound - caused by compromise in the circuit design.
There are a many general rules, but designing an amplifier without deep understanding of active devices always runs the risk of ending in confusion.
Cooter's circuit looks like pure DHT to me. That's what I'm doing, just with 4P1L on the in and the out. So I'm not sure what you mean that a pure DHT isn't an improvement over the Cooter circuit. Aren't they they same thing?
OK, sorry - I thought this was a an all DHT design. Perhaps that could mean many things, but to me that would imply no gain from any devices other than DHT. I'm OK with SS filament regulation, CCS, LED bias, etc. Others may not be.
No, not yet, not ruled out. I did say with the right transformers it should work very well. I'll keep trying!
If having a DHT in the output is the only criterion, then that opens a lot more doors. IDH triode or pentode at the input, for example. That seems to be the logical way to go in view of the problems of gain and microphony that a DHT has in the input stage. Shootz, maybe even a transistor at the input. But at that point we could drop the word "All" from the title.
From my point of view I'm now quite satisfied that DHT is NOT the way to go for the input stage. There has been some good discussion about that.
Yes my argument is simply that the 4P1l-4p1l isn't a big jump in power thats all.
Good news Tube Depot has 307A's on sale for $12.95 each, I just ordered 4, sure they will go fast. If that tribute 1:2 IT would work with a 4p1l input we would have a sure winner. Guess its time to start building/trial-error.