Possible NFB scheme for simple PI?

[imaradiostar]

So I have a design that's an attempt to use as few tubes as possible- I have only one triode available for a PI and it's driving el84's so very little grid drive is needed.

Picture a single 12au7 section configured as a fixed bias (3m3/1m) cathodyne phase splitter into a pair of cathode-biased EL84's. B+ is around 300-350 volts. There is a 1 meg volume pot with the wiper connected to the input of the PI as a master volume with a .01 uf cap.

Could I connect a 100 or 47 ohm resistor between the ground lug of the pot and ground and use it as a point to inject negative feedback? I assume the amount of feedback would change as I adjusted the master volume pot.

What about injecting NFB using a simple resistive network between the wiper of the MV and the grid of the phase splitter? How do I prevent volume changes from having a major effect on the amount of feedback? Should I add a mosfet source follower to prevent the loading of the previous stage from being an issue? Could I inject the NFB at the bottom of the source resistor of the source follower (like you would at the bottom of the cathode of a blackface champ).

Failing the above, can you think of a way to use negative feedback in with a cathodyne PI and a master volume without using an extra gain stage? I know that I have minimal gain available but I really want the option of tightening up the output section a bit with NFB.

thanks again for sharing your knowledge!

jamie

[Wakculloch]

As the PI is unity gain and it'll take 10 or so volts to get to full power, you'll need 10 volts into your PI. To then make that requirement higher by introducing negative feedback seems to be adding to difficulties, which is why the feedback is conventionally introduced into the driver stage of the PI. Having said that i can't see why the NFB can't be fed into the grid of the split load stage through a suitable dropping resistor, unless i've missed something, just never seen it done.

[imaradiostar]

I had another thought- I might try driving an interstage transformer with the au7 and making up some gain there. This would also simplify the nfb arrangement. I have a Fender musicmaster tf that I could try.

back to the other method- If I were to use a resistive divider at the input to the cathodyne phase splitter I'm concerned for the "bootstapped" nature of the circuit- input impedance calculation is more complex than just adding up the resistors. Does this only apply to a cathode biased inverter or would the fixed bias cathodyne use a similar formula for input z?

How large a cap and resistor should I use to prevent interaction with previous stages? This is why I was thinking about the mosfet- the low driving impedance of the mosfet would allow me to apply NFB without loosing too much signal going into the PI.

jamie

[Wakculloch]

Don't know about that, but just thought about the biasing of the grid in the split load stage, so the grid will be at 100v or so, and then we're talking caps in the NFB path. If you could move the volume to before the driver stage then everything could be pretty much as a conventional split load PI, (as a tight arsed Brit, Cathodyne has a different meaning which is why I avoid using it), using one ECC82 as driver with cathode injected feedback, then second half as split load stage. Gain has to come from somewhere, but it's just a case of whether the volume can be moved.

[imaradiostar]

maybe I'm using the wrong term but split load is what I mean... a pair of 27k resistors plate and cathode, fixed bias at the input or smaller value grid bias resistor and cap with a 1M grid leak resistor. I don't know if there is a difference between the two as it applies to the formula at http://www.freewebs.com/valvewizard/cathodyne.html.

This morning I was debating the purchase of some sort of simulation software...but by the time I do that and learn it I could have just built something and tried it. I have an old Eico signal gen that'll make a pretty stout signal. I can wire up a pair of el84's and a 12dw7 PI pretty quickly and start inserting different NFB methods to see what happens. I have some high voltage mosfets on order so I'll be able to at least try it!

Transformer PI seems like a neat way to go and it'll solve the problem. If I can get away without it I will but if not...no big deal, it'll do the trick!

jamie

[Bruce / Mission Amps]

You won't get quite as much power but use 6BM8s instead of EL84s.... they have a built in low gain triode that can be used as part of the phase inverter for push pull.

[imaradiostar]

Thanks for the reply Bruce. I should have just told you guys what I'm planning. I wanna build a single channel non-mv marshall preamp into a simple phase inverter. I plan to use V1a&b as the gain stages and V2 as a cathode follower and phase inverter. I've got some 12dw7's so I intend to use the au half as the phase inverter. If the tone stack creates too much signal loss I'll just ditch it.

It's an experiment in seeing how much I can do with a really simple circuit- I'm not against V1a being a Jfet (J201) or Mosfet (2n7000) if that's the only way to skin this cat but I'm trying to see what EL84's and a cathode follower sound like without a LTPI. It seems most LTPI's can abuse an el84's grids before they run out of headroom so why not try something a little bit different?

One of my amps is a plexi pre (bright and normal chans linked) into a single EL34. I figure if there is enough signal for the EL34 to distort then there should be enough to destroy a pair of EL84's using a split load PI.

Is a bassman/plexi circuit a waste of time without the LTPI adding to the mix?

jamie

[MWJB]

So why not tap the NFB at V1b's cathode? You might need a bypass cap here to bump up gain, so use the SF Champ style loop, 47ohms between cathode resistor/cap and ground.

Early 4x10" bassmans had a split load PI, albeit with a different tone stack (but still a CF).

If it doesn't pan out & you need more gain, I'd simply ditch the CF, feed the tonestack from V1b's plate and have a regular split load inverter with preceding gain stage. Vol pot will need some tweaks to give a useful sweep.

[imaradiostar]

So why not tap the NFB at V1b's cathode? You might need a bypass cap here to bump up gain, so use the SF Champ style loop, 47ohms between cathode resistor/cap and ground.

Early 4x10" bassmans had a split load PI, albeit with a different tone stack (but still a CF).

If it doesn't pan out & you need more gain, I'd simply ditch the CF, feed the tonestack from V1b's plate and have a regular split load inverter with preceding gain stage. Vol pot will need some tweaks to give a useful sweep.

Good questions, I hope I have good answers!

Initially this project started exactly as you describe it- then I started to think about other things I've wanted to try recently.

I don't wanna insert NFB at V1B because I'd like to be able to use a pre-pi master volume if needed, to vary which stages clip first more than anything. Why? because I wanna be able to hit the cathode really hard. Why? Because the other experiment on this amp is swapping Rk on the cathode follower to see how it effects the tone of the cathode follower. The other thing I intend to do is have variable bypass caps on both cathodes to dial in different amounts of gain but retain some bass rolloff.

I guess I'm falling into the normal trap- too many switches and options just makes a mess. I should stick with a normal marshall/bassman pre and put a phase inverter after it and just see how it sounds- one experiement at a time. If I want to try different Rk or feedback schemes or master volumes I can always try them later, one at a time so I can really hear if I like the difference or not.

Have any of you built anything like this before?

jamie

[Gainzilla]

WAY late to the party, but I'm building an amp with a cathodyne PI and wondered the same thing. Turns out Orange do this a lot (as per Merlin's article above). Not sure what the sonic differences are, but I'll give it a go and report back.

http://el34world.com/charts/Schemati...c_mkii_120.pdf

Cheers,