DC-coupled cathodyne and feedback

[passaloutre]

I'm thinking about making a Kay 720 for my next project. In my journey to study and experiment with all the different phase inverter configurations, I've found this one to be quite unique. I've provided a schematic that appears to come from a SAMS file, as it provides voltages, which the Kay schematics don't (this will be relevant below).

I've come across one of these pentode-triode cathodyne inverters before in my Ampeg Gemini, but this Kay has the confounding features of 1) DC coupling between the pentode plate and the triode grid, 2) DC feedback from the triode cathode to the pentode screen, and 3) smaller load resistors on the cathodyne (15k) than I'm used to.

I think I can wrap my head around (1), the DC coupling is a cheap way to bias the cathodyne, though it comes with some caveats described by Merlin, et al.

(2) What in tarnation? Is this negative feedback? applied to the screen? to linearize? the pentode? Why is there so much voltage dropped across the 180k resistor: 92V at the cathode and 34V at the screen? That seems to imply a lot more screen current than I would expect, are these voltages erroneous? Modeling the circuit in LTSpice does not show the same voltage drop, but I've learned to not always trust LTSpice in complex vacuum tube simulations.

(3) the 15k load resistors seem to allow a lot more current through the tube than usual (most cathodynes I've seen with 56k or 68k), which seems odd to me, as we would want to maximize the voltage swing on the 6L6 grids.

Can anyone help me understand what's going on here? Why the designer might have made these choices, and what the results are? I find myself drawn to weird phase inverters, as they seem to provide a lot of character to simple amps. I love the ratty sound of a poorly tuned concertina. I'm not trying to fix it, just to understand what makes it tick.

Two more questions while I'm here:

a) Where/how would you add a master volume here? I have a feeling the 7199 pentode get overdriven by the preamp, so I want it after that, but my usual cathodyne MV spot isn't there due to DC coupling. Maybe that weird screen feedback could be used for MV? For the record I like all the weird nipple distortion and frequency doubling you get from the cathodyne, so I'm inclined to PPIMV.

b) How would you change this circuit if using a 6AN8 instead of 7199? They seem similar enough on the datasheets, but just curious.

[tubeswell]

(1), the DC coupling is a cheap way to bias the cathodyne, though it comes with some caveats described by Merlin, et al.

Yes

(2) What in tarnation? Is this negative feedback? applied to the screen? to linearize? the pentode? Why is there so much voltage dropped across the 180k resistor: 92V at the cathode and 34V at the screen? That seems to imply a lot more screen current than I would expect, are these voltages erroneous? Modeling the circuit in LTSpice does not show the same voltage drop, but I've learned to not always trust LTSpice in complex vacuum tube simulations.

The g2 voltage is being set by the cathodyne's cathode voltage - and it acts to provide feedback to the screen

(3) the 15k load resistors seem to allow a lot more current through the tube than usual (most cathodynes I've seen with 56k or 68k), which seems odd to me, as we would want to maximize the voltage swing on the 6L6 grids.

It's to do with the tube characteristics for the triode. If the triode resembles a 12AU7 characteristics, then 15k on each resistor (for a 30k load in total) is not unusual.

a) Where/how would you add a master volume here? I have a feeling the 7199 pentode get overdriven by the preamp, so I want it after that, but my usual cathodyne MV spot isn't there due to DC coupling. Maybe that weird screen feedback could be used for MV? For the record I like all the weird nipple distortion and frequency doubling you get from the cathodyne, so I'm inclined to PPIMV.

b) How would you change this circuit if using a 6AN8 instead of 7199? They seem similar enough on the datasheets, but just curious.

PPMV (if you really need it)

Can you link the datasheets?

[passaloutre]

Here are the datasheets. Most examples I can find with 7199 or  6AN8 phase inverters use larger loads on the triode.

eg

https://el34world.com/charts/Schematics/files/Ampeg/Ampeg_G-12%20(Gemini%20I))_1.pdf

https://el34world.com/charts/Schematics/files/Univox/Univox1221.pdf

https://el34world.com/charts/Schematics/files/Sunn/Sunn_1200.pdf

Also the 6CG7 (similar characteristics to those above and 12AU7) phase inverter I’m most familiar with has 68k loads:

https://el34world.com/charts/Schematics/files/Silvertone/Silvertone1484.pdf

There has to be a reason for the 15k in the Kay.

[DummyLoad]

15K will yield a somewhat lower Vout to grids of 6L6, draws more current, as you already pointed out, g2 of the driving pentode is biased from Vk of that cathodyne, I suspect that's the reason for the 15K Ra&Rk and the higher current draw, also that NFB loop probably adds a lot to the tonal character of the circuits. Cathodyne output with 15K Va&Vk should be around 50Vpk with that Vg bias; more than enough to drive cathode biased 6L6 to clip.

6AN8 and 7199 are so close just sub with socket connection changes. IME the pentode of the 6AN8 sounds a little better if biased a bit colder.

--Pete

[tubeswell]

The triodes of both those tubes are similar to a 12AU7 - with similar plate resistance, amplification factor and plate dissipation. So any load between 10k and 30k will work (provided the bias is appropriate).

[Merlin]

(2) What in tarnation? Is this negative feedback? applied to the screen? to linearize? the pentode?

That was a semi-popular way to design a hi-fi pentode stage back in the day (see another example below). The screen grid gets negative DC feedback from the next stage (cathodyne), which ought to stabilise the screen voltage, but there is no AC feedback since it is shunted by C4.

Why is there so much voltage dropped across the 180k resistor: 92V at the cathode and 34V at the screen? That seems to imply a lot more screen current than I would expect, are these voltages erroneous?

That's only 0.32mA screen current which doesn't sound crazy to me. Also the voltmeter will pull the screen voltage down a bit, making the difference seem larger.

[passaloutre]

Thank you all for the help and the additional examples.

Is it possible that the additional current through the phase inverter afforded by the 15k loads improves its ability to drive large low-frequency signals (it is after all a "bass amp") on the grids of the 6L6 power tubes? Or am I mixing up my mental models?

I'm having a hard time finding any guitar amps with such small loads on the cathodyne inverter. The only example I can find with a 12AU7 has 100k loads on it.

https://el34world.com/charts/Schematics/files/Magnatone/Magnatone_m10a.pdf

[SEL49]

You're concerned about nothing. No one knows why Kay decided to use 15K. It's not wrong but use larger resistors if you want to. Doing so will not change the signal level to the output tubes. But if I were building a Kay 720, I'd stick to the schematic.

[passaloutre]

Thanks, Steve. "Concerned" is probably the wrong word, but I get it: just build the thing.

Unfortunately my day job affords me with ample internet screen time, and precious little tinker-on-my-own-projects time, a dangerous combination. So I am left with lots of opportunity to ponder the thing and sparing moments to just-build-the-thing.

Which leads me to wonder: why does this circuit of all circuits have 15k load? When I build it, which may be this weekend or four weekends away, I will know. In the meantime, someone smarter than me probably already knows. That's the cause of my "concern".

[DummyLoad]

Not guitar; Hi-fi.

http://www.tubebooks.org/file_downloads/RCA_HiFi.pdf

--Pete

[passaloutre]

Well there you go. It's straight out of an RCA whitepaper. That's good enough for me.

Maybe I need to be smoking a pipe while tinkering...

[SEL49]

Well there you go. It's straight out of an RCA whitepaper. That's good enough for me.

But wait. Don't you now wonder why RCA chose 15K?   

Several other Kay amps use that same 7199 PI circuit. They all have 15K resistors.

[passaloutre]

Coming from RCA, the extra current is used to burn out your 7199s faster so you buy more tubes.

[DummyLoad]

Thinking that Kay 720 basically copied parts of those RCA ckts.

Meh! ~800mW Pdiss on a tube rated for 2.5W - Lazy work load.

--Pete

[passaloutre]

Definitely joking about the burning up tubes.

And yes, I think you're correct about the RCA manual. The preamp 5879 circuit is lifted directly from that booklet as well.

Now I’m left to wonder why they didn’t use the 7027 tubes or that fancy diode-less fixed bias arrangement…it never ends!

[Merlin]

Is it possible that the additional current through the phase inverter afforded by the 15k loads improves its ability to drive large low-frequency signals (it is after all a "bass amp") on the grids of the 6L6 power tubes?

It will slightly improve its ability to drive grid current yes, but not by much.

[tubeswell]

why does this circuit of all circuits have 15k load?

Firstly, a cathodyne with a 15k plate resistor and a 15k cathode resistor = a 30k load


The resistive load determines the gradient of the  loadline. A smaller resistive load = a steeper gradient. This means all other things being equal, that there will be more tube current change per amount of input (grid) voltage change. In a cathodyne, the plate and the cathode are the current ‘sources’ for the signal outputted from the stage. More current at the source means the signal will better withstand the attenuating effect of whatever comes after the stage (all other things being equal).



(Edit: a balanced cathodyne has unity gain and low output impedance, but there is an advantage in maximising tube current by reducing the load, in that better impedance bridging is achievable for driving the following stage)