I do not think that this will work, but it was a challenge posted out on RAT that I wanted to play with. I recall once upon a time that there was a flirtation with direct coupled tube technology, but it required an elaborate power supply scheme, and it included resistors also. Here is the schematic I came up with. It useds a current source at the bottom biased so as to have a zero grid voltage referenced to ground. It is differential connected with another pair of tubes SRPP with them. The output is a cathode follower with a current source with bias to have the cathode zero with respect to ground.


    After you finish laughing perhaps someone can actually try it out with spice or something like that. Similar circuits with transistors are done all the time. Here is an alternative that may actually work better.


    Notice I tie the two grids to the plate of the first SRPP. It adds some NFB although it looks forwarded. The signal goes in positive at the first grid, then gets inverted by that triode. It also goes through the second triode via the cathode, which does not invert the signal. The plate output of the second triode "meets" the signal of the first at that point. At worst, this may leave us with a unity gain, requiring a high voltage gain preamp, or oscillation.

    This signal then gets injected to a directly coupled output triode, paralleled for lower output impedance. The output is a cathode follower, biased to zero by the bottom row of tubes so as to have a direct coupled output.

    A third option would be to have the top two grids at the input biased by yet another bias adjustment (or fixed and regulated using tube voltage regulators) from the power supply. This way gain can be attained by the first stage to drive the output stage.

Just for yuks, let me know what you think: