Charlie Wilson wrote: ↑Sat Oct 17, 2020 6:06 pm
Delmar, so the question I have is, were you able to get a balanced output before with the 1 meg grid leak and now with the 10 meg are not? The more I think about what the LNFB is doing the more the 10 meg makes sense. Martin, I have been reading up on the split load invertor, so basically the plate load(100kish) is split in half. Half goes on the plate and half goes on the cathode(Fender design 56k) so the same signal current that flows through the plate flows through the cathode also. Dumble doubled the Fender value, I would assume, to get closer to unity gain. If you add up the resistance of the cathode you get 103.7K and the plate is 110k. So it is a bit unequal and I guess you suggest the 100k to get the plate and cathode resistance closer in resistance. My source material(Tales from the Tone Lounge) also states that a split load invertor is a terrible way to try and balance the signal since the actual impedance characteristics of the plate and cathode are different. So is a balanced output even possible with Fender's or Dumble's circuit or is the trimmer(and LNFB) possibly a way of smoothing things out so that is is more pleasing sonically? I don't have a Tweedle Dee right now to play around with but I do remember that I could hear a difference in the way the amp sounded by adjusting the trimmer in the first replica I built. Finally, has anyone who has built this circuit been able to get a dead on output balance by adjusting the trimmer?
CW
Ive not been around much lately, but i have some experience with cathodynes i can offer, probably been covered but anyway. The one amp ive sold the most of has a cathodyne PI, fixed bias 6V6 or 5881. I simply love cathodynes. Theyre beautifuly balanced if you dont drive them and i really like the way they clip provided a big grid is used on them. Where they fall down in balance is when overdriven hard. Both anode and cathode loads have an AC reference to ground. cathode directly and anode thru the PI filter cap. But the cathode also has a DC reference. And its that DC ref that holds, or clamps the cathode fed signal out. Whilst the anode sees no such reference, and is free to draw more bias current to the place it actually starts robbing the cathodes. I find on my amps the balance is perfect til clipping on the outputs starts. As you really push them hard beyond say 50-60ma each, the cathode side will hold there while the anode will push past to 70-80ma as measured at its cathode/s.
That doesnt sound terrible, tho wears that side tube out faster.
WHen i first saw the tweedle dee mods, i overlooked what HAD had done as simply his fascination with nfb and balance. But in trying the nfb out, i found the wonderful answer. WHere a given amp would hold 60ma at the cathode side but push to 80ma at the anode, they both pull strong to 70ma. The nfb only applies to the anode side of the PI, which should tell you something, and the way it clamps the anode from pushing harder, rectifys the way it robs bias current from the cathode side.
Ive seen another builder try and overcome this using a higher grid resistor, like 10k on the anode vs 1k5 on the cathode. But the 2nf/3m3 is about darn perfect.
My guess therefor for the loads being greater is they simply sound better off the leash. I used to run 56k but moved to 82k loads on 6V6;s, 100k on 5881's. But HAD also raised the cathode value to 4k7 so its kinda moot really, would be essentially equivalent to 56k and 1k cathode. The pot is seen on the SVT's cathode purely so a little different but again recognising the need to compensate.
I didnt notice any loss of harmonic content using the nfb nor did i bother with the loads or trimmer, the nfb simply cured the extended swing of the anode by itself. I run individual fixed bias on my amp so a shared cathode bias arrangement may vary some. As far as the 10M grid leak goes, I would hazard a guess that the nfb may take some of the tweeds rawness away, some of that glory tone that we all love, perhaps it cures the cure.
I would love to have met HAD.