I have waay too many 6DJ8's

[kenpeter]

HP and Tektronix pulls. Amperex and Telefunken.
Looking for the best way to burn these usefully.
Twenty or thirty at least... And no, not all at once!

I understand the 6dj8 has two issues of concern.
Nonlinearity, and Microphonics.... Or is this all a
big urban legend to worry about nothing???

Anyways, I come up with this silly stage... I look
inside and see the identical triodes back to back.
And I wonder how I might trick microphonics to
buck by adding Mu? Does physical symmetry in
the 6dj8 triode permit or not any such bucking?

The other thing I was shooting for was linearity.
I figure any crummy triode can be linear if you only
feed it constant current. So I strapped the top pair
up as an augmented Cascode and try to Mu follow
the sum of the lower pair. Reasonable?

I hope the cascode might also help improve PSRR.

With B+ as specified (+350V) would this atrocity have
sufficient linear headroom for normal line input levels?
Any significant screwups to be fixed here?

Thanks,
Ken

I see already the 300R isn't burning off 2*Bias,
cause the Q current on that chain is different.
Where did my calculator go?...

[Eli Duttman]

I hope the cascode might also help improve PSRR.

PSRR in cascodes is nearly non-existent. Very tightly regulated B+ is essential.

IMO, many of the sonic ills attributed to the 6DJ8 family are due to parasitic oscillation, not lack of linearity. Silicone rubber "O" rings deal with microphonic specimens.

Instead of "oddball" circuitry, think about culling some nice tubes out and selling them, preferably to other forum members. You get cash to fund your building efforts and scarce, W. European, tubes get into circulation. It's a win/win situation.

[Bandersnatch]

hey-Hey!!!,
The cascode plate load is a fine thing. I have done it with a pentode stacked on a triode in very similar fashion to a Gary Pimm MOSFET circuit. I did the voltage ref's differently.

On the use of a cascode below, what you've got is a pentode loaded by a CCS. Huge output Z and gain. Not generally a good idea I think. The cascode loading a single triode is quite another thing... Think of it as a tube CCS.

The cascode has terrible PSRR only if resistively loaded. A TX loaded cascode has excellent PSRR( just like a pentode ). For a triode it is te reverse; the resistive loaded one has the better PSRR.
cheers,
Douglas

I also got lots of 6DJ8's pulled from instrumentation stuff like Tek 'scopes. Lots were noisey, but a run through a phono pre found all of those. I am down to about 40 good ones, and perhaps 80 which are still useful where extreme quiet is not needed.

[kenpeter]

Originally posted by Eli Duttman
PSRR in cascodes is nearly non-existent.
Very tightly regulated B+ is essential.

The cascode (at the very top) is the regulator...
Maybe not a true regulator, but a gyrator certainly.
No other outputs are tapped from it but to hold
the plate of it's brother (the Mu+Mu Follower) at
constant voltage above that Follower's cathode.

It woulda been more akin to an SRPP if I hadn't
done anything special to stiffen the Mu Follower.
I chose to stiffen it by regulating the plate. But
increasing the resistance at cathode would have
been the more "normal" thing to do.

I may still do that (stiffen the follower's cathode),
and eliminate the cascode at the top. Bringing the
total to 3 twin triodes for the full stereo pair...

But again, I am concerned how microphonics are
intercepted by the triodes? What mode of shaking
is most sensitive? And is the pickup of microphonics
(in 6dj8) a common mode, differential, or neither?...
Do I make the problem better or worse by stacking?
Should I be differentiating instead?

Does adding (stacking) Mu in triode pairs that appear
to be oriented in physical opposition reinforce or cancel
the pickup of microphonics? I can't see inside the actual
triodes to make a determination if both might internally
be symetrical. And therefore outward appearance may
make no difference in voltage orientation of pickup?

No.. Wait.. Maybe I CAN see inside! I got access to
pretty good XRay machine at work. See right through
gold plated lid on a heavy duty BGAs... I should snap
some screenshots (with 6dj8) on Monday for shiggles.

[kenpeter]

Originally posted by Bandersnatch
On the use of a cascode below, what you've got is a pentode loaded by a CCS. Huge output Z and gain. Not generally a good idea.

There is no cascode below, its a Stackode. Mu adder, doubler...
The ref point (if it were a cascode) is swinging along merrily at
1/2 the sum of Mu1 + Mu2. Its not a fixed voltage reference.

If we compare to your Pentode model, as-if G2 were triode or
UL strapped to exactly 1/2 the plate voltage... And that pentode
also just happens to physically position G2 exactly at the halfway
point to the plate. Same thing, only different .)

[Bandersnatch]

Originally posted by kenpeter


There is no cascode below, its a Stackode. Mu adder, doubler...
The ref point (if it were a cascode) is swinging along merrily at
1/2 the sum of Mu1 + Mu2. Its not a fixed voltage reference.

If we compare to your Pentode model, as-if G2 were triode or
UL strapped to exactly 1/2 the plate voltage... And that pentode
also just happens to physically position G2 exactly at the halfway
point to the plate. Same thing, only different .)

That is what I'd call an Ultra-Linear cascode. It is running along at 50% tapping. John Broskie ran something like this in his Tube-Cad. Not exactly a bad thing either. It will allow running the high-Z plate load.
cheers,
Douglas

[smoking-amp]

"reinforce or cancel
the pickup of microphonics?"

How about a servo driven electromagnet to damp out the vibration of the steel plates. (just kidding) Needs corner reflectors on the plates for a laser interferometer to sense the vibration.

Don

[kenpeter]

Why not just put noise cancelling headphones over them? (also kidding)

[kenpeter]

Originally posted by Eli Duttman
Instead of "oddball" circuitry, think about culling some nice tubes out and selling them, preferably to other forum members.

Nope not selling. Trade with the local natives for other juju, I'm OK with that...

[pmillett]

Looks to me like it will work - it would be interesting to see how well.

So rather than theorize, why don't you build and measure it?

Pete