Experiences with White CF/CCS CF output stage

[mashaffer]

I have been thinking about a tube project (for when I have some money to try it). I am in the very early research stage and I was intrigued by the Lizzy 10Y-6L6GC power amp published in Valve magazine. It uses a transformer coupled driver stage and a parafeed autoformer CF output stage.

My idea was to try a MOSFET loaded Mu-Stage (Alan Kimmel) voltage amplifier stage and an active load CF output stage in an attempt to get high linearity and low Zout. The output transformer would of course be capacitively coupled to the output stage ala parafeed.

Has anyone built a white cathode follower output stage or any type of tube or hybrid CCS CF output stage? I would be interested in your results.

P.S. Is the Kimmel Mu-Stage a patented circuit?

[badman]

I'm currently using a white cathode follower with the 6922 tube. Got some goofy results at first, but I think I'm getting it to where it needs to be, getting quite the impressive sound out of it, with the exception of some heater hum (Schottkys are on the way to alleviate *that*) It's an easy circuit, 6922 white CF, 100 ohms on plate and cathode, 100 ohm gridstops. I used 4uF auricaps for output, and 10,000 pF silver mica for input. Running at about 195V B+
6X4 rectifier into CRCLC
1uF, 1k, 10uF, 30H (600R) ,28uF. With the exception of the hum, it's dead quiet.

[Sch3mat1c]

Did you try hum balance and bias on the heaters?

[badman]

Heaters are via a 6.3VCT winding on a hammond 369EX. The center tap is tied to the B+ with a 470k resistor.

[Sch3mat1c]

Ack Use a voltage divider with impedance at least 50k (e.g., for 1/2 B+, use 100k up, 100k down) and bypass with a cap.
Voltage should be above or equal to the highest cathode voltage if it's allowed (i.e. within H-K ratings), otherwise somewhere inbetween as best you can get it. In your case, I'm guessing either 160V or 320V supply and the CF is biased in the sweet spot (approx. 160V bottom plate/top cathode given 320V) so there's about 160V between cathodes... +80V would be best to bias at, putting an even 80V H-K on each.

Tim

[badman]

Well, the B+ is 195, so there's no 320 to be found in the circuit, we're more at like 97 from top cathode-bottom cathode. 320 would roast the tubes, 6922 isn't very good about high voltage.

But that aside, let me make sure I get you straight. Use a voltage divider along the lines of 100k from B+ to heater CT, and 100k from heater CT to ground, with a cap bypass on each? Or am I completely misunderstanding you? I see where you're going with this, but want to be sure, I'm still a relatively new t00bie

[tubetvr]

P.S. Is the Kimmel Mu-Stage a patented circuit?

In that case the patent have already expired, the circuit was known already in the 1940-ties and is described in principle, (together with SRPP) in "Vacuum tube amplifiers" by Valley and Wallman written 1943, probably in other books too.

Kimmel only "re-discovered" the circuit, he didn't invent it.

Regards Hans

[Sch3mat1c]

Originally posted by badman
Well, the B+ is 195, so there's no 320 to be found in the circuit, we're more at like 97 from top cathode-bottom cathode. 320 would roast the tubes, 6922 isn't very good about high voltage.

Ouch, what a waste (in the PSU resistor that is). I get 5W SE with that transformer (er, 269AX), 330V into a 6L6. Ratings, shmatings...

97V between y'say? Then half the difference (~48V) added to the bottom cathode voltage (<5V, so who cares) is a good starting point. In this case, since the top cathode has to swing voltage, it would be a good idea to bias the heaters closer to it than the lower one. It's also better to bias above a cathode than below as leakage is higher due to the heater acting as a filament cathode against the inside of the cathode sleeve. Obviously not so high as to put the bottom triode in a worse situation. So, I'd go around 60 or 70V. Off +200V, that's around 1:3, so 47k and 100k would probably work for the divider (voltage divides as R2/(R1+R2) so the resistors are 13-1) in ratio, thus 100k to 1/2(100k) = 47k common value). Now, you calculate the power dissipation in them for proper sizing.

Tim

[mashaffer]

[list=1][*]When one uses a CCS in the cathode of a CF output stage how do you choose the impedence of the output transformer? I am assuming Plate connected to B+, CCS between Cathode and ground. Transformer primary capacitively coupled to the cathode (i.e. para-feed CF with CCS as the cathode load).[*]I am thinking about the Svetlana 6BM8 for a first design. Would the triode in this tube be able to drive the gate capacitance of a MOSFET used as the CCS in the Mu-Stage voltage amplifier or should I use two sides of a double triode (in parallel) for the Mu-Stage and a seperate power tube for the output?[*]I understand that Pentodes and Beam Tetrodes work very well as CFs but I have also heard that pentodes don't like CCS loading. If you were making a CCS loaded CF would you wire it as triode or pentode? Why?[/list=1]

[Fuling]

Good questions on an interesting topic!

1: For max. power and minimal distortion I´d say the same impedance as if the transformer was connected to the plate as in a usual power stage. 5-8 kohms for a 6BM8.
I might be wrong here, but this is what I´ve come up with when researching for a similar project (two PL504 cathode followers loaded by an autoformer)

2: The 6BM8 triode section works at very low currents (1mA or so), so it might be a bit weak. If you want a lot of gain and voltage swing together with a bit of current, look at EC/PC86.
Don´t know what they are called is the U.S though.

3: Pentodes seems to make fine CF´s but the problem is to keep the G2 voltage constant WRT the cathode.
You can do that either with caps and resistors (which loads down the output a bit) or with at Mosfet follower that feeds the screen grid (requires a bit of circuitry including an additional PSU).

I´ll probably run my PL504´s as triodes.
Why? Simplicity and (probably) better sound.

Loading the plate of a pentode with a CCS and nothing more wouldn´t be such a great idea, but now the situation is different.
The gain of the output stage will be less than 1, CCS or not, and the the "real" load is not the CCS but the output transformer.