Octavers again...

[bea]

Well, hopefully final version, at least in LTSpice. I added an overdrive stage with an internal positive feedback loop. The thing seems to run better at 30V, but 20V seems usable. Let's see what happens on the breadboard...

[bea]

Well, a first live result. I built the version with the anode follower in the PI and the cathode mixer and did first tweaking of the values of the resistor. Unfortunately, my power supply delivers only 18.7 V - a bit low for the JJ ECC82 i have available.

[PRR]

Interesting.

[bea]

This evening i did measurements on the phase splitter itself. Unfortunately, the 2nd stage has a lot of even harmonics. Its gain is dependent on the amplitude of the input signal. So the phase splitter is not symmetric, and it appears that it needs to be fine tuned to every individual tube.

I also checked the cathodyne variant with limited success. While the output signals at each branch were more symmetric than with the anode follower, their addition was a lot worse: the part of the anode of the cathodyne stage contributes much less than the branch of the cathode, and hence suppression the fundamental was worse than with the anode follower. Mhmm. The simulations tell me different...

Let me note that the emission of the valves seems quite poor - a known property of recent JJ valves at low voltages: voltage drops across the resistors were less than half the computed values.

It looks as if i'd use better suited valves (EHs, for example) and a larger rail voltage. Which means i have to wait until the missing parts arrive.

(If anything else fails, it should be possible to tweak the 1st circuit a bit to achieve a usable result...)

[PRR]

The grid of the cathodyne "should be" about 1/4 of the supply voltage, for "normal" supplies.

Referencing it to ground ensures large 2nd for small signals.

You may not be able to go up to 1/4 and still have ample conduction in the vacuum. But your sim should be easy to bias-up and try.

[bea]

I am aware of that.

Even in my sim i would need to bias all three electrodes. If i just bias up the grids the voltage difference between cathode and grid remains the same. Similarly on the difference stage.

I think i'll also need to add a pot in the cathode branch to simply adjust the levels.

(EDIT: works fine; it is possible to adjust the balance in a way having a "cleanish" octave at small and intermediate levels with the fundamental reappearing fairly slowly when the signal becomes strong.)

BTW: in the sim the circuit works just fine.

EDIT: now its time to concentrate on proper biasing which is an issue in all stages. IMO the input stage id the most critical because an input impedance of 1 MOhm is needed.

(Hopefully) Final edit: after some successful experiments using biasing by pullup resistors proposed by Merlin Blencowe in his article on LoV tue preamps i think it is worthwhile and promising to test this approach here as well. Let's see.