NFB applied to power tube grid, possible?

[Catmaigne]

So I built a very small bedroom amp utilizing only 3 tubes, but capable of putting out decent high gain tones. The front end is essentially a modded JCM800 (minus the CF) followed by a cathodyne PI feeding a 6SN7 wired in push-pull. In certain configurations, it puts out a max of 2.5-3w from what I've crudely measured so far with a plate voltage of ~430v.

I'd like to add some NFB to the circuit to tame the woofy lows and add a proper presence control. Being that the 6SN7 is cathode biased, I've experimented with shrinking the cathode bypass cap to accentuate the highs -- although that only seems to go so far. Dumping the lows with smaller coupling caps and shrinking the cathode bypass caps in the preamp changes the character of the gain too much. V1 is a 12AX7 and V2 is a 12DW7, so there are only 3x AX7 triode halves that comprise the preamp with the tonestack/MV coming off of the 3rd plate. The final triode is the AU7 half of the 12DW7 for the cathodyne. Although I have nothing to drive the cathodyne with except the signal off the MV, the 6SN7 doesn't need a lot of voltage to drive it.

I looked at some of the schematics of the 1w Marshall heads (JCM1, JTM1) and the Peavey Classic 20 which inject NFB at the cathodyne grid, none of them utilize a presence control in that network. The Classic 20 does dump some NFB high end content, but I think adding a presence control here might be too interactive with the tonestack/MV and be too lossy. I also noticed those amps have voltage dividers going into the cathodyne grid which are probably to isolate the NFB slightly, but I'm trying to avoid dumping any voltage there. This got me thinking...

Could I inject NFB from both sides of the OT secondary right onto the 6SN7 grids with a presence control ala Vox? Or am I thinking about it all wrong? I can't think of an amp where I've seen an arrangement like this.
Something like this:



Anyway, here's a pic of the amp and a clip of me noodling. This was with the cathode bias set pretty cold at 75% and a very small bypass cap. I also had the Vox "cut" control rigged up to the cathodyne plates.

https://drive.google.com/file/d/1Lum7_rbUnCNub2gz7dBFnd3K1oQ2W49O/view

[PRR]

Welcome. Very pretty build.

> I can't think of an amp where I've seen an arrangement like this.

That's a hint.

Without thinking, I'd say the 6SN7 has loaded gain of 12 and the OT has step-down like 30:1. So there is "no" voltage gain, no NFB, even strapped to unity gain.

Also that bit of shoving NFB into the prior plates works very much better with pentode driver; a triode absorbs most of the NFB.

You know how most amps do it. Add a gain stage and run the NFB back to there. If that does not fit with your preamp philosophy, you are becoming an amp designer. Some overall plans suit some schemes better than others.

[_Steve]

I'm interested in this too. Could he split the cathodes and connect the NFB to an additional split in the cathode R?

Also is there any risk of charge build-up by not grounding the common of the OT secondary? (My guess is no, there's a route to ground through the nfb injection point?)

[Catmaigne]

Welcome. Very pretty build.

> I can't think of an amp where I've seen an arrangement like this.

That's a hint.

Without thinking, I'd say the 6SN7 has loaded gain of 12 and the OT has step-down like 30:1. So there is "no" voltage gain, no NFB, even strapped to unity gain.

Also that bit of shoving NFB into the prior plates works very much better with pentode driver; a triode absorbs most of the NFB.

You know how most amps do it. Add a gain stage and run the NFB back to there. If that does not fit with your preamp philosophy, you are becoming an amp designer. Some overall plans suit some schemes better than others.

The OT I'm using has a tap on it that would effectively be 32R which would be useful if the NFB voltage is too small. It's a P-T291 I got from Amplified Parts that I have wired up for a 16k-18k load.

I'll try some other things first to see if I can get closer to what I'm looking for tonally. My biggest gripe with the amp is that it sounds too mellow and compressed, almost like there's a blanket over it. I chalked it up to the lack of NFB combined with the preamp topology making it flabby like a Mesa Rectifier. I could reconfigure the preamp to be more like the Krank amps where they're loaded with treble peaking circuits to combat the lack of NFB, but it makes the upper mids and highs too grainy and pronounced for my tastes in addition to a woofy low end.

I think the first thing I'll try is adding a capacitor coupled bias supply and converting it to fixed bias. I was trying to avoid this from the start because the chassis is very cramped, but I'll find a way. Keeping it on the colder side should shave some bass and add some high freq crossover distortion to make the amp sound a little more "present." If that doesn't do it for me, I'll experiment with my goofy NFB arrangement.

This all might be the nature of the beast being that it's still a triode amp. Feels like I'm painting tiger stripes on a housecat and expecting it to roar.

[shooter]

time for an O-scope
your description sounds like you might be hammering the PA with to much signal, or you have to much cathode cap on the gain stages.
I'm running PP 6V6 ~8W for my audio, driving 4 speakers, while it is loud enough for the space and clean audio, it just misses ROAR.

[tubeswell]

A small amount of local feedback can be applied at the output stage by including the secondary of the output transformer in the cathode circuit (but you need a special tap on your OT secondary for this).

[66Strat]

You might try implementing local negative feedback from plate to grid of the 6SN7s a la AB165 Bassman and implementing the Vox Cut Control as you mentioned in your original post.

[66Strat]

The Cut Control along with the local NFB may make the amp too dull. If you try this NFB topology, you may want to experiment with the value of the plate to grid resistors to get the amount of NFB that sound best. A 1 meg variable resistor in a single end 6SN7 Champ style circuit would be a good platform to experiment with.

[Catmaigne]

So I played around with a few different arrangements.

Converting the amp to fixed bias seemed to tame a little bit of the flabbyness in the lows. Biasing cold or into class B didn't add much high end grit if any. I was hoping for a healthy dose of 5150-like crossover distortion, but the triode doesn't respond the same way.

I reinstalled the cut control, but the amp was still too dull sounding even with it disconnected. I then tried a few NFB/presence arrangements at the input of the cathodyne, but it killed too much gain to be useful in my circuit. As a last resort, I tried out some positive feedback as a presence control and really like what I'm hearing. Seems that the triode has so much internal NFB that you can get away with a decent amount of PFB without any oscillation.

[1blueheron]

I was told if I converted a pentode output tube amp to triode mode, it would sound about like you describe.  I think your triode output amp sounds pretty good and gnarly. 

I would love to see a schematic of your first tube as well.  My ear says if you stripped just a tiny bit of the lowest frequency out of the signal in the first stage, it might tame the flabby bass syndrome.  What do you have in the way of power supply filtering?  Could it be power sag is also contributing to the flabby bass? 

[Catmaigne]

I was told if I converted a pentode output tube amp to triode mode, it would sound about like you describe.  I think your triode output amp sounds pretty good and gnarly. 

I would love to see a schematic of your first tube as well.  My ear says if you stripped just a tiny bit of the lowest frequency out of the signal in the first stage, it might tame the flabby bass syndrome.  What do you have in the way of power supply filtering?  Could it be power sag is also contributing to the flabby bass?

The first stage is pretty much an exact copy of the Soldano Hot Rod and Jet City amps minus the big grid stopper after the gain pot. Basically a JCM800 first stage with the 1n/470p treble caps flipped around, 220k/1.8k + 1uF for biasing, and a 22nF coupling cap. I did try shrinking the 22nF coupling cap down, but I've always found cutting the bass too early just makes the lows muddy when you bring them back later. Instead, I shrunk the coupling cap down on the 2nd stage down because it seems way more effective downstream from the gain pot. I might try shrinking some other coupling caps farther down the line, maybe I'm slamming this little tube with too much bass in general. Also, the three power supply nodes are 47u/22u/22u. I will try boosting the preamp filters a little and see what happens.

To be honest, the actual character of the bass probably isn't as flabby as I think, the triode is messing with how I'm perceiving it. I'm theorizing that the higher output section damping of the 6SN7 flattens out the impedance spike of the speakers too much down by the low E, so the sub-bass frequencies are more prominent than they would be with a PP pentode. The impedance curve flattening is also why the presence isn't very prominent either. I originally had the primary impedance of the OT set very high and conservative, but the amp was more dark and flat sounding than the 16-18k I have it set to now.

I think the secret sauce that this amp is missing is a healthy dose of high order odd harmonics to make it more edgy and aggressive sounding. I will play around on LTspice and see if I can induce that with bypassed plate resistors, more PFB, etc? I'm also curious as to how a 12AU7/12BH7/ECC99 would sound compared to the 6SN7. I've read some reports of them being less linear, but how much less? This makes me wonder if all of those 1w Marshalls have the bite I'm looking for, or if it's a fantasy to expect that from a triode.

[pdf64]

…
I tried out some positive feedback as a presence control
…

That’s a bit cryptic, it may be useful to explain how you went about doing it.

[Catmaigne]

…
I tried out some positive feedback as a presence control
…

That’s a bit cryptic, it may be useful to explain how you went about doing it.

I'm still in the tinkering stage, but here it is anyway... (edit:pres pot is supposed to be 100kb, still tweaking)
Basically feeding some in-phase signal off of the OT right back into the PI grid. I fiddled with some values around the PI grid to make it less interactive with the MV. The shelf it creates starts around 1-2kHz. I originally had a half assed feedback setup on the OT secondary, but it was too leaky when compared to the high input impedance of the cathodyne and didn't supply nearly as much PFB.

I'll upload a clip later.

[Catmaigne]

I've been experimenting with some values and managed to bring out more sizzle. Definitely getting closer. Here's how it sounds atm:
https://drive.google.com/file/d/1OBAsF8ULDf_rBt9FJSHaQj9rKtrsAthy/view?usp=sharing

Forgive me for the compressed cellphone audio. I'll mic it when the circuit is nailed down. On the left is how I have it set currently, except the 100k and 250k resistors are actually pots I had hanging on there so I could fine tune the response... Spoiler alert, I like them turned all the way up. On the right is what I'm going to switch it to. I changed the pot connection (pretend it's a lamar PPIMV) so rotating it shouldn't affect the cutoff frequency much. I may experiment with dropping the 5M tail resistors, although that comes with the risk of oscillation.