Why not? Less is more and if well achieved the task, you will end up with an amazing result. To do this you will need a good quality OPT and the valve which can do this task. It will need high mu where possible and linear. There are just a handful of DHT that could do the job (like EML20/30) or you can look for the high gm pentodes (e.g. E55L, E810F, etc.) which are very expensive generally or Russian valves like my beloved 6e5p which is cheap as chips (still).
Depending on the step down ratio of your OPT you will need likely a step up input transformer to drive the 6e5p to full tilt. Here is what I experimented with many years ago. Unfortunately I didn’t take any distortion measurements but it sounded really nice.
If you look at the performance of one bottle, we can squeeze out a pair of watts easily out of this valve:
Ignore the THD projection, it’s much lower than that in reality. I will hook this up again and measure to contrast. The 6e5p in triode mode is extremely linear valve so you should expect somewhere below 5% at full swing. You can see that we can get out 2W when valve is biased at 40mA/250V. Input drive it’s about 11Vpp or close to 3.8Vrms. So at least a 1:2 SUT will be required.
As you can see on the above diagram, a pair of 6e5p can do 4W. I omitted the SUT as I was driving this stage with my preamp. At the time I used a Lundahl transformer, but any 3K-ish transformer would work well here. The filaments are DC regulated but you can do as you please. The cheap DC-DC SMPS converters can do great job here. Also the HT can be derived with a HV DC-dC converter plus a cap multiplier stage. All it will be powered from a small 12V DC wall-wart supply.
The SiC diode stage performs great in terms of sound and low dynamic impedance. You can use these PCBs to hold them easily.
If you have good gain but not enough (current) drive into the grid of the 6e5p to handle the miller capacitance, you could add the follower stage like this one:
Of course you end up adding an extra stage (follower) and a pair of capacitors, but is worth trying to listen to HF improvement due to slew rate distortion removal.
You could also power the follower stage from a 24V bipolar supply and tweak the R1/R4 resistor divider to provide DC coupling into V1/V2 by removing C4. You will need to ensure the MOSFET is biased to about 0V. This would help with A2 drive if needed. But all this is extra complexity not needed here. I think it goes against the SPUD ethos but wanted to put it out there as many of you like experimenting a bit around the circuits (like I do).
So the above implementation is what I’d do again. The addition of T2 (1:2 or 1:4) provides the voltage gain needed. You may need (or not) to add the Zobel network (Rz1 and Cz1) to keep HF response smooth depending on the SUT of your choice. I could recommend Sowter, Lundahl or Dorin’s Bodea (DVB) transformers depending on where you live.
An amazing stage and no excuses to build this one. Only big expense is the OPT pair. Other than that, you can build this easily and enjoy the 4W out of it. Enjoy!
