Cathodyne "not good at driving heavy loads"?

[potatofarmer]

I've read this a number of times but I've never seen a flushed-out explanation.

It seems to me that since the output impedance (to either output) is quite low, the cathodyne should be fantastic at driving heavy loads... until the power tubes start to clip, of course, when the output impedance at the anode skyrockets. That doesn't seem like a huge concern to me though; once the power tubes start to break up all that matters is "does it sound okay" and I've built/played with three amps with cathodynes and come to the conclusion "yeah, that sounds pretty good."

So!

Out of boredom I've been thinking about making a 4xKT88 ultralinear amp with a Hammond 1650TA output transformer (1k9 Ra-a) using either a Antek 320V or 400V PT for ~450V or ~550V. Datasheet conditions right from Genalex.

I'm thinking I want to drive these KT88s with a cathodyne, but the grid leak for a pair of parallel KT88s is going to have to be around 47k (or possibly lower, depending on grid stoppers). This seems like quite a heavy load, especially if I pick, say, 22k with a 47k grid stopper.

While I'm not interested in poking into AB2 territory, I was thinking about adding AC-coupled cathode followers between the cathodyne PI and the grids. These would ensure that both outputs of the cathodyne would be equally loaded and the output impedance from the cathode followers would be low enough that a 22k grid leak would result in very little attenuation.

But would this be a waste of a tube (or a pair of MOSFETs)? I feel like it would be, but I can't quite figure out the caveat of cathodynes not being good at driving heavy loads.

What am I missing?

[tubeswell]

Cathodynes can have really low output impedance, but the lowest output impedance is achieved when they are perfectly balanced.

Having said that, the cathodyne's output signal swing is limited to the headroom achievable within the voltage drop across the plate-cathode load resistors (which is typically 1/4-1/3 of the HT voltage for each load resistor). If the output voltage swing exceeds this, you get frequency-doubling distortion etc.

You can crank the HT voltage up to increase the maximum possible unclipped output swing, but you are limited by the peak voltage rating of the tube.

You can use a high current tube (like a 12AU7 or 12AT7) with an optimum load, to get the highest possible tube current. But then you will want to drive a 12AU7 with a high input signal, to get the most out of it, because its got a lower amplification factor than a 12AX7. But even that still suffers from frequency-doubling etc when its overdriven (It's just that its harder to overdrive than a 12AX7). So there's 'ways and means' of optimising your cathodyne, but it still has limitations.

Another possibility is to use a 12AX7 in a conventional driver/cathodyne arrangement, and then use one triode of a 12AU7 on each side of the output stage of the cathodyne, either as a CF buffer, or as a conventional gain stage (to boost the cathodyne's output), to drive grid-current-heavy output tubes.

[bea]

To my understanding, in a cathodyne the tube must deliver twice, better just twice the current each of the tubes in a LTP needs to deliver.
A 4xKT88 power stage in normal AB operation can still be driven by a 12AX7 in an LTP - that's occasionally done in some Hiwatts (although the original layout used a 12AT7), and it works well. So if the ECC83/12AX7 is able to drive two KT88 the ECC81 should be able to drive 4 of them. And if the 12AT7 is not sufficient any more You'll need either high power double triodes like the ECC99 or buffering of the outputs.

I would consider using a 12AT7 (ECC81) in a cathodyne - but prior to this ensure that the maximum output swing is sufficient for KT88 stage with its relatively low µ. It is sufficient to drive 4 EL34 with 800 V on the plates (search for the schematics of a Dynacord Gigant, an old PA amplifier), and it performs reliably.

The 12AX7 has the same output current as the 12AU7, its output impedance is only marginally higher than that of the 12AU7, and thanks to that and its high µ amplification under load potentially larger than even that of the 12AX7. In the abovementioned Gigant, the other triode is used as a normal amplifier stage where it delivers an amplification of 50 or slightly more.

Alternatively You might consider using the ECC832 in a cathodyne which combines one system of the 12AX7 with one of the 12AU7

[g1]

The 12AX7 has the same output current as the 12AU7

Am I taking this out of context?

[tubeswell]

The 12AX7 has the same output current as the 12AU7, its output impedance is only marginally higher than that of the 12AU7, and thanks to that and its high µ amplification under load potentially larger than even that of the 12AX7. ....

Hi bea. in fact a 12AU 7 has a much lower plate resistance and higher transconductance than a 12AX7, and a much lower amplification factor (17-19 as opposed to 100) See the handy comparison tables attached. The 12AU7 grid curves are steeper - comparable to a 12AT7 (comparison between 12AU7 and 12AT7 examples under a 20k load attached)

[bea]

Am I taking this out of context?

That's simply a typo which occured while i needed re-type my posting.

Of course I meant the 12AT7, not the 12AX7. I think this should be pretty clear from the whole context of my posting.

Please remind that i think in German naming - ECC81 and ECC82 are the tubes with a similarly low output impedance and the same maximum current, but largely different µ while the ECC83 (=12AX7) has a large plate impedance and can deliver just 10% of the plate current the other two can.

[g1]

That's simply a typo which occured while i needed re-type my posting.
Of course I meant the 12AT7, not the 12AX7.

Got it. Danke!

[potatofarmer]

This all makes sense; thanks. The Fender silverface 12AT7 PI has no problem pushing adequate signal across 47k grid leaks, so that is always an option. I have played around with a 6C4 cathodyne (essentially one 12AU7 triode) and increasing the supply helps but yeah, then you have to worry about the maximum quiescent plate-cathode voltage as well as the heater-cathode voltage.

So I guess it really boils down to not an impedance bridging issue per se, but more of a question of maximum voltage swing. The datasheet suggests the KT88s are going to require upwards of 75V for full output and while it seems possible to get a cathodyne to accomplish that, the post-PI driver really seems like the way to go to ensure the power tubes break up first.

That ECC832 looks tempting, but so does a 6GH8.

Also I've since found out that the (Matamp-designed & built) Orange Lead 200 essentially answers this question, with a 12AU7 used for driver + AC coupled cathodyne and another 12AU7 used for a driver for each push-pull side. Like the good old Williamson.



Seems like it'd be a good idea for stability's sake to nix a global NFB loop, but with an UL power amp I don't think it'd be necessary.

Thanks everyone!

[tubeswell]

... the post-PI driver really seems like the way to go to ensure the power tubes break up first. ...

NZ manufacturer Jansen also did this in the 1960s in a number of amps

[bea]

Hi,

there is an important limit of the ECC81: the maximum voltage between cathode and heater is only 90 V. You might consider using an elevated heater (to 45 V as long as the other system is on normal potential). The vintage Dynacords operate directly at that limit.

[potatofarmer]

Hi,

there is an important limit of the ECC81: the maximum voltage between cathode and heater is only 90 V. You might consider using an elevated heater (to 45 V as long as the other system is on normal potential). The vintage Dynacords operate directly at that limit.

Yep, I ran into this with the 6C4 cathodyne too. 6GH8s also have this limit, as do 12AT7s. Just a casual observation, I'm no tube encyclopedia, but the only preamp tubes I've seen that claim a tolerable Vhk greater than +/- 100V are 12AX7s. Actually the Sylvania 12AX7 datasheet from 1955 specifies a maximum of 100V DC, but up to 200V "DC + Peak". A GE datasheet dated February 1951 just specifies +/- 180V with no caveat about quiescent vs. signal conditions.

That would be interesting to investigate further - maybe I'm cynical but I suspect 100V is probably a reasonable limit for 12AX7s too.

Just another problem that post-PI driver stages solve, you can use more sane supply voltages for the cathodyne if you don't have to worry about wringing every possible bit of voltage swing out of it.

[g1]

When checking into various h-k ratings, the 180V number seemed to be the one most commonly mentioned for 12AX7's.
But it is important to remember many common designs are using tubes way beyond their max ratings, often with no issues.
I got worried about a JCM800 I was looking at last week, it had a 12AY7 (90V h-k rating) in the cathode follower position, with 146V on the cathode. The owner said it had been working fine like that for years. It was NOS though, and I wouldn't try a modern production 12AY in that spot.
So I put in a 12AX, and with the shifted operating points, it ended up with 200V on the cathode anyway , so again beyond rated max.

[potatofarmer]

Someone must have studied this back in the golden age. Seems like this would be a more gradual failure mode, or possibly a noise issue. I try to stay close to the datasheet boundaries if only because I don't like paying to replace tubes.

For what it's worth, I always use Merlin's arc protection circuit on direct-coupled stages. The Valve Wizard