Hi tubeswell! Yes, that was mainly my intent, and primarily my question. But its an interesting proposition, so lets talk specifics:
A. We could bias them differently; one fix, one in the conventional vanilla missionary bias manner. But the whole reason i wanted to abstain missionary bias in favor of fixed was to keep the feral FET tolerances +temperature drift in check, plus avoiding the hassle of biasing each LND sample individually. Also, wouldnt the split loads NFB keep the PIs output identical until one of them prematurely hit the rails??
In my case the EL84s will be fed with about 275v B+, so they should start overloading at roundabout 18-16v grid swing peak-peak. So when the PIs hit the rails with their 250+v supply, they since the're FETs they should be putting out about 2x 120v swing peak-peak (unless severely mis-biased). So could any difference in the AC they now put out enrich the harmonics of the EL84s, which left this realm for square wave paradise a long time ago? I honestly dont know. I suspect not really.
B. If we instead half/double the plate+cathode resistors on one of the cathodynes (both fix bias), the Idiss would in-/decrease, but again, if i understood the function of a SLPI the output as well as headroom (being a FET) should remain basically the same? Though its ability to handle loads would change, so perhaps some differences in signal out from the PI would arise, especially with the EL84s in overdrive. Although i do suspect/hope this possibility would be strongly counterbalanced by the big grid stops, which for mentioned reasons (not having one side of the PI fold in half) im not too keen on ridding.
C. We could differentiate the output grid leaks, which possibly would exacerbate any discrepancies arisen due to point B upon EL84 overload. But also affect what "environment" the output tubes see, and thus how they behave. And perhaps more obviously: affect low frequency content. Dont if it might sound good, but youre definitely differentiating the output tubepairs, if only in a limited freq range, which is also resides in the low end.
D. the LF differentiation could also be had by changing coupling caps before/after one PI, without the change in load.
E. Have earlier though about differentiating the screen grids between one pair to the next. Which i dont think is a bad idea to try, suspect atleast some extra complexity might arise, but unsure how well it will sound. But even though interesting concept, now we've left the PI completely.
Also, method C+D, which i think will have more effect than A or B, i guess can be had without two PIs altogether, and instead just paralleling coupling caps and grid leaks to separate EL84 pairs, from the same PI. But then again, inabling DC theft from neighboring EL84s, making me uncertain if im really adding or actually decimating harmonic compexity. Method E ofcourse could be had without even touching the PI.
Can you think of any other methods that might be viable?
I think what im getting at is having both PIs perfectly well balanced, but not identically balanced, if that makes sense. But in my minds eye any discrepancies always end up being corrected by the invisible hand of NFB. In other words, i dont mind the dish, in fact i love the dish, im just wondering if theres any way of sneaking some spice into the soup when the chefs not looking.
/kw
Topic: Parallel split load PIs for quad EL84 output stage (one PI/pair, shared OT). (Read 2386 times)