Parallel Cathodyne

[hylaphone]

Hi folks, I am working on a 12AU7 cathodyne PI for KT88. Looking at a typical 47K Ra and Rk, if I were to parallel the triodes would it make sense to halve the Ra and Rk values, as one would do in a paralleled gain stage?
I'm thinking it would allow for smaller grid leak values on the KT88s?

[eschertron]

I think yes, you will double the current and halve the source impedance.
You'll still have the peak-peak voltage limits of a cathodyne though. Approx 1/3 HT for the signal reaching the grids of the KT88s, but you already knew that

[Helmholtz]

I'm thinking it would allow for smaller grid leak values on the KT88s?

Grid leak means a DC path between grid and cathode. Lowered AC source /PI impedance doesn't matter.

[hylaphone]

Ok, wanna call them bias feed resistors? I'm referring to the 100K's in this old chestnut:


I see Max Rg1-k listed as 100K for KT88. Why not use a smaller value with a lower source impedance?

FWIW I'm playing with the Dynaco topology, with a pentode driver into 12AU7 cathodyne.

[eschertron]

I'm curious too. Actually, curious TWO:
1) first, why lower the bias resistors? Most folks are trying to UP their values.
2) why won't the source impedance affect the knee frequency of the network?

[Helmholtz]

I see Max Rg1-k listed as 100K for KT88. Why not use a smaller value with a lower source impedance?

Sorry, I misinterpreted. As Rg1-k is specified as a max value, any lower value is fine with the KT88s. And a lower source impedance is able to drive a lower load impedance. So no problem with halving the grid leak/bias feed resistors. For full power the PI must be able to supply at least a peak-to-peak signal of twice the grid bias voltage.

As Eschertron indicated, the coupling cap values should be doubled with halved bias feed resistors for same bass response.

[hylaphone]

I'm curious too. Actually, curious TWO:
1) first, why lower the bias resistors? Most folks are trying to UP their values.
2) why won't the source impedance affect the knee frequency of the network?

1. Just a harebrained idea since low source z is free here, and modern KT88 seem sensitive to that parameter.
2. Yes indeed

I'll likely stick with 100K, but figured I'd ask about this. Thanks --

[Rhodesplyr]

FWIW I'm playing with the Dynaco topology, with a pentode driver into 12AU7 cathodyne.

My take from personal experience having played around with Dynaco amps for years and having rebuilt two pairs of Dynaco MkIIIs with improved driver boards is that the Dynaco driver circuit shows its limitations when trying to drive a pair of KT88s to full output. I personally prefer a circuit like the tubes4hifi board that uses a 12AU7 long-tailed pair driver with a constant-current source to drive KT88s. That configuration raises a MkIII to an entirely different level of HiFi performance.

The Dynaco pentode gain stage/cathodyne driver works much better with EL34s and (in the SCA-35 and ST-35) EL84s.

You have to keep in mind that David Hafler's goal was budget HiFi, and that meant keeping the parts count as low as possible: one driver tube.

I think if you're going to use both sections of a 12AU7 to drive KT88s, you're better off using the LTP than trying to build a better cathodyne.

[Helmholtz]

Approx 1/3 HT for the signal reaching the grids of the KT88s

Peak PI output is actually less than 1/4 HT.

[Dave H]

Peak PI output is actually less than 1/4 HT.

Reading the sentence before I assumed it was peak-peak that was 1/3 HT.

[SoulFetish]

My take from personal experience having played around with Dynaco amps for years and having rebuilt two pairs of Dynaco MkIIIs with improved driver boards is that the Dynaco driver circuit shows its limitations when trying to drive a pair of KT88s to full output. I personally prefer a circuit like the tubes4hifi board that uses a 12AU7 long-tailed pair driver with a constant-current source to drive KT88s. That configuration raises a MkIII to an entirely different level of HiFi performance.

The Dynaco pentode gain stage/cathodyne driver works much better with EL34s and (in the SCA-35 and ST-35) EL84s.

You have to keep in mind that David Hafler's goal was budget HiFi, and that meant keeping the parts count as low as possible: one driver tube.

I think if you're going to use both sections of a 12AU7 to drive KT88s, you're better off using the LTP than trying to build a better cathodyne.

Patrick Turner of Turner Audio has an interesting article on "reforming" a Dynaco MKIV, where he drives the output stage in a similar way using a LTP with a CCS tail.Dynaco MKIV Reformed
.

.
Another option could be to use a Pentode/triode tube (6U8A, 7199, etc.) to provide your gain/phase inversion, and buffer the Cathodyne using the 12AU7 as cathode followers. This gives you a lower impedance drive. You could even DC couple the cathode followers, in which the cathode load resistors would ultimately be the grid leak resistors. (You'd need a - voltage supply though) just some thoughts