Originally, this circuit was driving a 6B4G PP output stage. Transformer-coupled, naturally. I started with cap-coupled Mullard splitter using three 4P1L direct-heated loktal pentodes. Not bad at all, but had insufficient gain. Then I tried a plain Loftin-White configuration with a LL1660 interstage (this one is a one-of-a-kind custom 30mA version - later Lundahl introduced a high-current, low-resistance LL1671). Power supply is tube-rectified LC filter (here's a schematic of original all-DHT amp with Mullard splitter). All capacitors - paper type. Both 6B4G amps had a touch of that DHT-SET vocal magic, and I would probably stay with Mullard splitter.. if only it had enough gain.
Then the amp became an OTL driver. I simply removed all driver tubes from my OTL power amp, and ran unshielded twisted pairs from Lundahl secondaries to CF tube sockets (cathode=hot end, grid=cold end), plus common chassis ground. It was noisy but after relocating bias pots closer to output stage and an additional C-filter on bias network - acceptably quiet. Bandwidth into 8 Ohm resistive load is 40Hz-40kHz (-1dB) at output terminals.
This driver is very good in driving a Cirlotron output stage. There was more body in low midrange (a big problem with my 2.5-way speakers where impedance falls to 4 Ohms under 300 Hz). Overall, sound is less fatiguing. With secondaries in series, gain is 8x at V1 plate, 80x at V2 plate, 100x at each transformer output - 4 times higher than a 6SN7 cascode. Rewiring transformer primaries in series... this was supposed to give wider bass extension, but in fact sounded worse. I would blame it on the OTL's uneasy marriage with ported speakers - more low bass cause distortion below port frequency.
More to come: the circuit can be reduced to a single IDHT tube - 6N30P, or 6N6P seem a good match for both stages. However, I suspect gain will be even higher - so the second stage should be a low-gain regulator tube.