Radford pentode-triode LTP splitter

[smoking-amp]

"attach the screen's AC bypass to the cathode instead of ground?"

Sure o'nuff. Takes screen current out of consideration then. Or return it to the plate with the LND150 screen driver idea.

Don

[smoking-amp]

I should'a said it takes AC screen current out of consideration then. The CCS still has to supply the DC screen current with the bypass cap to cathode.

[ray_moth]

With a CCS, wouldn't you want to attach the screen's AC bypass to the cathode instead of ground?

Yes, SY, that's what I did in my model. I found 0.082uF did the job, one of the variables I twiddled with in trying to smooth out the LF response. (I didn't bother to try bypassing the screen to ground because it didn't seem to make sense.)

The other variables that affect LF stability are the cap in the step-network on input to the splitter and, of course, the caps from the splitter to the cf driver stage.

You wanting to improve it by putting a CCS in the fully intentional very short "long tail" (actually, this is not an LTP whatsoever), clearly shows you didn´t understand how it really works

Ya got me, Tom, I admit I don't understand how it really works. More to the point, I dom't understand why it works, but it certainly seems to. I went for a high-mu triode but I could have gone for one with high Gm like a 6DJ8 - although with a 6DJ8 the plate-cathode voltage would have to be treated with caution.

[kenpeter]

If you are using a low Gm triode, it works for the very reason
I have already explained. The triode's Mu is the most current
limiting factor, therefore the entire U-bend cascode is linear.
The Pentode isn't in control of anything, except both cathode
voltages. As a cathode follower, and top half of a cascode.

But Smoking has asserted that the Triode in Tom's schematic
of the Radford has the higher Gm. If that is the true case, my
theory falls apart (or at least isn't linear anymore). I do not
yet fully understand how the original design actually worked
with a Pentode in control of current, and remained linear???

I'm assuming Tom will offer a translated explanation sooner
or later.

[smoking-amp]

I suppose the ground referenced screen could be seen acting like a hidden UL feedback, and so effectively linearizing the pentode side some. Putting the bypass to cathode would null that effect out.

So, I would go with the LND150 trick to put the screen current back into the plate circuit (so CCS'able then), and leave the screen referenced to ground (Vref for the LND150 gate ground referenced. This only works for cases with the min. plate voltage staying above the screen voltage.).

Don

[kenpeter]

What was I thinking???? As long as there is a functional
conspiracy to install Mu as the most current limiting factor.
And the anti-current is within the capability of the Pentode
to conduct. It doesn't matter at all the Triode's Gm may be
"too high".

[smoking-amp]

"What was I thinking???...."

Translation (I think):
The triode's plate voltage reflects back to it's cathode via inverse Mu effect (4th circuit topology thingie, grid grounded). That connects to the pentode's cathode, affecting the pentode grid to cathode voltage. So the triode is still controlling. Also, the pent. screen UL effect is coming in as 1/Mu(p) there too, to linearize a little more.

Don

[kenpeter]

No, it works even if the Pentode holds the cathode voltage
constant. A change in Triode plate voltage will be reflected
in the Pentode's plate voltage. Assuming equal loads, and
a pentode that is willing to conduct enough to keep up.

I'm just saying that Mu and the Triode's plate resistor may
be restricting Triode current far more than the Triode's Gm.
If that brings the current change within the grasp of the
Pentode to conduct the opposite, the problem is solved.

[smoking-amp]

Well, in any case, I think the hidden screen UL, and the triode's inverse Mu, keep this thing humming linearly. But the whole scheme seems to be just for the purpose of getting the Miller effect down on the input (pent.) grid.

So why not just use two identical triodes with a cross neutralization cap from the 2nd plate back to the input grid, so much simpler. Maybe put both neut. caps in for complete symmetry.

Another useful mod. would be putting in a center tapped inductor for the two plate loads (for either scheme). Guaranteed 180 degree splitting and twice the output voltage swing. Good for screen drive amps.

[Miles Prower]

Looks like a Nachbaur cathode coupled paraphase. A good way, perhaps, to get rid of some of those oddball 6X8s (pentode/triode with cathodes internally connected). This would have the same problem as any other paraphase: unbalanced harmonics between phases.

Why these folks persist in doing stuff like this escapes me completely: Van Scyoc's, Isodynes, and now this -- all more complex than either a cathodyne or an LTP, and they don't work as well. The paraphase is a quick 'n' dirty solution, but still, more complex than a good old cathodyne. Somebody's got too much time on their hands.