Cathodyne Mu Follower Line Out

[merlin]

6922 different than a 12AX7 in that regard?

Oscillation is uncommon in single stages, especially when the signal source is physically close to the tube. A couple hundred ohms should suffice, I would think.

Yes, that and 4:1. Pretty much everyone makes one.

Do you have a preferred part?

 

[atavacron]

Oscillation is uncommon in single stages, especially when the signal source is physically close to the tube. A couple hundred ohms should suffice, I would think.

Great, normal stopper.

Do you have a preferred part?

An API 2623 backwards. 30dB max primary like that. But Jensen, Lundahl, Cinemag, Crimson all do this, with varying flourishes. Edcor too. @ruffrecords did a deep dive into the distortion of a few mfr options a few years ago, I’m not sure what the link is but I recall an excellent PDF.

 

[merlin]

Just playing around in simulation, this puts 20dBu into 600 ohms (Ra and Rk trimmed for unity gain). Not stellar but certainly economical.

 

[atavacron]

Just playing around in simulation,

Thanks for looking into it. Why fixed bias vs self?

 

[atavacron]

View attachment 124851

I see your circuit and raise you a lower output impedance and Av closer to 1. Back to you.


edit: slightly higher gain coming off the plate instead of the follower, also obv then the mosfet is just a ccs and not in the signal path.

and: i don’t get why cathodyne phase splitters are often set up around a lower-than-mid-supply voltage. does it have to do with cathode stress?

 

[merlin]

also obv then the mosfet is just a ccs and not in the signal path.

A follower is not in the signal path? That's a bold claim!

and: i don’t get why cathodyne phase splitters are often set up around a lower-than-mid-supply voltage. does it have to do with cathode stress?

It's because triodes cannot swing rail to rail (they need more voltage across themselves), unlike transistors or pentodes.

Why fixed bias vs self?

Generally preferred because it makes the input impedance independent of the load impedance.

 

[ruffrecords]

Just playing around in simulation, this puts 20dBu into 600 ohms (Ra and Rk trimmed for unity gain). Not stellar but certainly economical.
View attachment 124851

Definitely struggles at +20dBu output into 600 ohms but then we don't need that. PSRR at -10dB is pretty poor. Almost as bad as an SRPP but definitely more economical Could be a serious contender with adequate HT smoothing.

Cheers

Ian

 

[atavacron]

PSRR at -10dB is pretty poor.

Maybe there’s a way to inject the power supply noise in reverse phase, a la many Broskie examples over the years. Series cap and resistor from B+ to the cathode? Something in that direction, null it out…

Oh yeah here’s a 2010 TCJ entry on this

Question is, what’s the simplest way with a single supply. Dual supply is out of scope in most scenarios.

 

[ruffrecords]

Maybe there’s a way to inject the power supply noise in reverse phase, a la many Broskie examples over the years. Series cap and resistor from B+ to the cathode? Something in that direction, null it out…

Definitely possible but it would need another tube. Since we only have the one and the current draw is modest then perhaps the simplest solution is an RCRCRC filter in the HT. It is not too hard to achieve 90dB ripple reduction with this method

Cheers

Ian

 

[atavacron]

Definitely possible but it would need another tube.

I edited the post, curious to know thoughts on that TCJ post.

It is not too hard to achieve 90dB ripple reduction with this method

Or possibly better with a Maida.

 

[ruffrecords]

I edited the post, curious to know thoughts on that TCJ post.

TCJ? Can't seem to find that

Or possibly better with a Maida.

But a lot more expensive and regulation is not really needed.

Cheers

Ian

 

[atavacron]

TCJ? Can't seem to find that

https://tubecad.com/2010/12/blog0196.htm

 

[merlin]

PSRR at -10dB is pretty poor. Could be a serious contender with adequate HT smoothing.

Some boost converters offer bipolar outputs. If the ripple is symmetrical it would sidestep the PSRR issue. Just a thought.

 

[Cqwet Dbdfte]

I see your circuit and raise you a lower output impedance and Av closer to 1. Back to you.

View attachment 124871
edit: slightly higher gain coming off the plate instead of the follower, also obv then the mosfet is just a ccs and not in the signal path.

and: i don’t get why cathodyne phase splitters are often set up around a lower-than-mid-supply voltage. does it have to do with cathode stress?

Did you try that circuit?

 

[atavacron]

Did you try that circuit?

I think the current set by the self bias would “fight” the CCS, or vice versa. Fixed bias should work better. But I’ll stand by that topology otherwise. Haven’t tried it yet, no. Also noting here that in “source-o-dyne” splitters, people split up B+ into four rather than three, so it’s possible that the CCS would allow for a higher Va and lower Vk, resulting in greater swing. YMMV.

On the CMR topic, a CCS would help isolate the valve from ripple.

 

[Cqwet Dbdfte]

The circuit now is an oscillator.

 

[atavacron]

The circuit now is an oscillator.
View attachment 127944

View attachment 127946

Let me guess…changing to fixed bias does not kill the oscillation, but taking the inverting output directly from the plate does?

 

[Cqwet Dbdfte]

Overall the circuit does not make sense to me. What is the objective? A buffer, no gain or a buffer with some gain?

 

[atavacron]

Overall the circuit does not make sense to me. What is the objective? A buffer, no gain or a buffer with some gain?

No gain. Same as the circuit Merlin posted. The point of the method is almost no noise gain. The point of a CCS on the plate is to offer the triode a high impedance load. There are other ways…non-critical though.

 

[Cqwet Dbdfte]

Maybe the best very low noise gain is a transformer ($$), a CCS loaded triode will have gain, which can be stepped down of course, but a no-gain buffer should have a large enough signal from a low noise source.
A CCS is "infinite" impedance, and if not carefully executed can be an added noise source, worse than a resistor.
I have seen examples of CCS cathode loads, but never (yet) tried it.
My circuit diagram above works OK with the bottom end of transformer grounded.
The DN2540 works better with another in series. All cathode/source/emitter followers use the power supply impedance as its "load", and should be isolated from other stages or have a very low (regulated) output Z power supply.