DC cathodyne splitter as a gain stage

[Gaz]

I have this two channel amp idea. One channel I want to "simulate" the overdrive of a DC coupled cathodyne splitter (in all its ragged glory), by using at the end of the preamp followed by a master volume. After that it would go into a cleaner running LTPI. Other channel is high gain.

Would it be as simple as putting the DC coupled cathodyne at the end of the preamp, and only tapping off the cathode or anode? Does it matter which I tap off of?

Curious if anyone has tried this or see it elsewhere.

[Chuck H]

Kind of impossible to simulate since the grid reaction of the power tubes is integral to the PI output waveform shape. But at face value it could be done with summing resistors from either cathodyne output to a single reference point.?.

[Gaz]

Thanks for the reply. Yeah, I knew that was part of the recipe, but I have an EL34 amp (based off an Orange) that has a post phase inverter master volume, and even with volume way down it sounds virtually the same as full up. Maybe it's the big bottles? So for the sake of this idea, pretend the cathodyne has a PPIMV after it turned down to almost nothing. Will my idea replicate in the preamp by taking off only one output?

But at face value it could be done with summing resistors from either cathodyne output to a single reference point.?.

What do you mean by summing resistors? Do you mean summing both channels? If so, I'll be doing that bit with relays. I just wanted to explain my idea in context, because if it were single channel I'd just use a cathodyne PI

[Helmholtz]

Kind of impossible to simulate since the grid reaction of the power tubes is integral to the PI output waveform shape. But at face value it could be done with summing resistors from either cathodyne output to a single reference point.?.

Don't forget that PI outputs are anti-phase, so signals will cancel when added/mixed.

Why not just try which output sounds better? Easy to change.

[Chuck H]

Don't forget that PI outputs are anti-phase, so signals will cancel when added/mixed.

Why not just try which output sounds better? Easy to change.

Yes, I had considered that but thought perhaps imperfect cancellation could allow enough small signal for the purpose. Now I'm thinking any imperfect cancellation may cause the waveform to sound weird. I suppose diodes could be used to simulate either, typically amplified swing from each output. Diodes could also be used to simulate the power tube grid clamping effect on the PI. Low value resistors in series with the diodes might soften any hard clipping artifacts. Just thinking out loud.

[Helmholtz]

Now I'm thinking any imperfect cancellation may cause the waveform to sound weird.

I think so because you would always get some difference signal.
You actually don't want a difference signal but rather the positive sum of the PI outputs because the OT reverses the phase of one side wrt to the other.

I suppose diodes could be used to simulate either, typically amplified swing from each output. Diodes could also be used to simulate the power tube grid clamping effect on the PI. Low value resistors in series with the diodes might soften any hard clipping artifacts.

Yes, diodes + series resistors can be used to simulate grid current effects.


I don't quite see why the amp needs 2 PIs. Why not use the DC coupled cathodyne alone with a PPIMV?

[Chuck H]

I don't quite see why the amp needs 2 PIs. Why not use the DC coupled cathodyne alone with a PPIMV?

This^^^^^^^^^^^^^^^^^^^^^^^^^^^

Plus the previously mentioned diode clamping could be valued very slightly above the level where the power tube grids clamp. That way you don't have any diode clipping in the unattenuated signal (master volume full up). The diodes would only clip when the master is in use. Just slightly later than the power tubes would. I've actually experimented with this in SPICE and it seems to work pretty well.

EDIT Negative feedback loops not considered for this circuit and might complicate the matter.

[Gaz]

I don't quite see why the amp needs 2 PIs. Why not use the DC coupled cathodyne alone with a PPIMV?

Yeah, I knew this would be hard to explain It needs two PI’s for several reasons: 1) The amp has second high gain channel that needs a regular pre-phase inverter master. Using two pre masters simplifies channel switching a lot. 2) I want to use an effects loop. A PPIMV with effects loop is useless. 3) I don’t want headroom to be affected by master volume settings. With a PPIMV, nfb is reduced with lower settings. 4) I want to use typical presence and resonance controls, which won’t work with a PPIMV at low settings.

So with that said, it sounds like I can just build the cathodyne and take the signal off one output? I would loose the grid clamp effects, but in my case that’s not something I’m worried about.

Is there any real benefit to doing this rather than adding an additional gain stage or DC coupled cathode follower? I was hoping to get some compression and asymmetrical clipping of it

[Helmholtz]

So with that said, it sounds like I can just build the cathodyne and take the signal off one output?

Sure, with equal loading both PI outputs should be identical but mirror-inverted as the signals are produced by the same plate current across equal resistors.

[Gaz]

That's helpful. This may be a stupid question, but do I need to load the unused output?

[Chuck H]

That's helpful. This may be a stupid question, but do I need to load the unused output?

Wow that's an excellent question!!! (that's kudos in case it wasn't obvious). I'll Spice it and post opinions in a day or two.

[Gaz]

Wow that's an excellent question!!! (that's kudos in case it wasn't obvious). I'll Spice it and post opinions in a day or two.

Thanks, Chuck!

[Helmholtz]

I'd say it is not necessary to load the unused output (loading includes coupling cap to not change DC operating conditions). At least not with typical load resistance of 220k or higher.

But there is some mutual loading effect, especially loading the cathode increases the output at the plate. Reason is reduced local NFB at the cathode.

In principle the cathode is able to drive lower impedance loads than the plate because of different source impedances.

If you use the cathode output, the circuit operates as a DC coupled cathode follower.
If you use the plate output it works as common cathode stage having a gain of 1.

[Chuck H]

Here are some spice simulations:

The circuit includes the proposed master volume and a large grid stop resistor. Simulations were done with and without the resistor and with and without loading the cathode. I can do more sims taking signal from the cathode side if you'd like.



Green trace is the signal from U5 plate and blue trace is the top of R8. I am intentionally clipping the cathodyne. This is with the grid stop and the cathode side loaded.



This is with the grid stop and the cathode side unloaded



This is without the grid stop and cathode side loaded. Note that without the grid stop the grid behavior of the cathodyne is now affecting the input signal.


And this is with no grid stop and the cathode side unloaded.

[Gaz]

I'd say it is not necessary to load the unused output (loading includes coupling cap to not change DC operating conditions). At least not with typical load resistance of 220k or higher.

But there is some mutual loading effect, especially loading the cathode increases the output at the plate. Reason is reduced local NFB at the cathode.

In principle the cathode is able to drive lower impedance loads than the plate because of different source impedances.

If you use the cathode output, the circuit operates as a DC coupled cathode follower.
If you use the plate output it works as common cathode stage having a gain of 1.

Thanks, really helpful along with Chuck's sim. I was planning on taking output off the cathode so it would be less sensitive to loading (and why not, it the more exotic output But based on your comment, is it really any different than just using a DC coupled cathode follower?