Russian rod pentode circuits

[Mike Watterson]

I then fed the 1j37b with oscillator output (no buffer) to g1a and a sinewave (not perfect) input to g1b

That works poorly. Do read what the expert, Joe Sousa, says on the Radiomuseum

To use 1j37b as a modulator:
Use a centre tap RF transformer to feed G1a and G1b differentially with the Local oscillator and feed audio to the centre tap so it's equal on G1a and G1b.
Or use a 1j24b and feed high level LO direct from LO anode to G2 as a mixer/modulator as it has nearly no gain at 25V Vg2 and maximum at about 55V Vg2, assuming an RF choke load and 55V to 90V HT.

To use 1j37b as a mixer
Use a centre tap RF transformer to feed G1a and G1b differentially with the Local oscillator and feed input to to the centre tap so it's equal on G1a and G1b. There can be a coil and cap (RF tuning) on the centre tap to ground.

1j37b or 1j42 operation.
Same signal on g1a and g1b is same as regular rod pentode gain. Differentially changing g1a and g1b (in either polarity) steers the beams away from the anodes and thus dramatically reduces gain, can simulate a beam switching tube or remote cut off. Thus a suitable level of LO there will work as mixer.

A differential g1a & g1b signal, depending on output load will be full wave rectified (rf detector) or frequency doubled (as ripple of full wave rectification is x2 frequency)