ESLDD with "anti-triode"

[revintage]

Hoktuna is now playing the proto he built of my SETOR with great success. He reports an "amazing soundstage"!



I have for a long time been thinking of building a DD for my ML Script panels. I have even let Jack at Electra-Print wind me a pair of PP-chokes.

Why not unwind the secondaries of a 30VA to use as PP-choke in this configuration? A 6CA7 or 6550 could easily be arranged as opposing element to a 300B. A Lundahl LL1660/18mA as IT.

[kenpeter]

One area that still concerns me, is PSRR. The power
supply here might need to be awfully clean. 120Hz
ripple is within the passband for differential mode.

Noise currents will be carried entirely by the Triode
side, and double trouble when the Anti-triode tries
to anti-copy them?

I am still thinking along the lines of letting the brute
bridge float on the current sources, and track B+ by
means of ultrapaths for the Anti-Triode's reference(s)
screen and grid bypass to b+ rather than ground.

The input stage might then float the signal up to a
common noise by means of Transformer, Loftin White,
or high impedance voltage-variable current source
(Sand, Pentode, or Cascode).

High PSRR might be achieved without the need for
excessive filtering of B+.

---------------------------------------------------------------

Also keep in mind the input can be made at either
Triode or Anti-Triode, inverted result has the same
tri-ody single-endedness. The Pentode might have
less Plate-Grid capacitive losses to drive than the
Triode.

[Michael Koster]

Originally posted by kenpeter
One area that still concerns me, is PSRR. The power
supply here might need to be awfully clean. 120Hz
ripple is within the passband for differential mode.

Noise currents will be carried entirely by the Triode
side, and double trouble when the Anti-triode tries
to anti-copy them?

The power supply noise is coupled through the Rp/Rk divider
and transmitted to the anti-triode in an additive way. I'm working
on some ideas to deal with it, but double the hum has not been
a huge issue for me with 94db/W speakers. My B+ ripple is
about 100mV if I recall correctly and I get about 3mV at the
output, but that's with a mu of 6 and a 10K:8ohm OPT.

I thought it would be easiest to use a variation of the LW noise
balancing techniques and feed an anti-noise signal into the
signal mirror side directly using a cap (similar to cap-coupling
PS noise to the driver cathode which would also work, right?)
Of course, I have the opamp circuit which makes it simple to
inject a controlled amount of noise into the anti-side. There
will be a way...

Michael

PS by my calculations the ripple at the output should be 5mV for
100mV on the B+

[kenpeter]

Thats a great idea!

With a Mu of 6, inject 1/6 PSRipple into the antitriode's grid.
Or the other way round if anti-driving from the dark side of
the listening glass...

The Triode's plate will then be common mode noise with B+,
anti-triode's differential misbehavior automagically corrected.

[revintage]

Hi guys,

A PSU with 100mV isn´t good enough. Even though you have no hum problems.

With data from a 100VA mainstoroid at 250VAC, brigde ,CLC 220u/3H/220u, 120mA DC-load, ripple should be below 20mVptp. So why not adjust the problems at the source?

[kenpeter]

I dunno, cause a pair of really good 220uF at that voltage
is a bit large and possibly expensive? A circuit with inherent
noise cancellation/immunity usually involves smaller caps.
Possibly caps made for audio rather than motor run?

Cause 440uF and 3 Henries can store dangerous amounts
of energy, far more than needed to carry the amp through?
Once a fuse blows, I want the amp to stop immediately.

Cause chokes are resistive and also have hysteresis not
accounted for in sims? Need another lump of mystery iron
like I need a hole in the head...

A really smooth power supply isn't a bad thing to have,
its just maybe not as necessary as some people think.
Its entirely possible to abuse BOTH approaches for a
really really quiet amp.

[Wavebourn]

Even when a conventional pentode tube P-P does not hum when idle, intermodulation of a signal and 120 Hz is visible and audible on a power. For triodes it will be even worse.

[kenpeter]

In circumstance where the sum of both branch currents (class A)
is held to a constant, what is to be seen modulated onto the
ripple? Only those losses due hysteresis, or delivered into the
actual load... What does that have to do with a circuit immune
to care of such variances as they might appear on the B+?

[Wavebourn]

Mu, Gm, Rp depend on a plate voltage.
Why don't spend one MOSFET ?
R-C with high time constant on a gate, one gate stopper, one Zener to protect the gate, and one more optional resistor in drain to protect against an overcurrent. A huge ripple reduction for expense of only 3V of a loss.

[revintage]

How come you guys are talking about modulation of a 120Hz signal in a thread I started about an amp for mid that is crossed over 2 octaves higher?

But I understand your interest in the "assymetrical self-split" with toroids(or 6B4G SE on steroids). Hoktunas reports are encouraging. For the proto he used an old PSU with 6D22, LC 15H 20uF filtering.

Also in a lowbudget project like this it is OK with electrolytics, no need for motorruns here. Or we could maybe get away with smaller caps and CLCLC. Hammonds are cheap......

kenpeter: You are a true audio-poet ! For a Swede it isn´t always so easy to follow your turns .