Volume leak with bootstrapped master volume

[Gaz]

hi everyone,

i decided to try a bootstrapped master volume exactly like the one here, but am experience some volume leakage with the control all the way down.

i've scoured the web for some info on this and came up with an archived ampage thread that quote a kevin o'connor book:

"We do find a quirk in the bootsrap MV... There can be signal feed through at the 'zero' setting. This happens because the impedances are so high and the signal feeding the pot is quite high, too. The signal can electrostatically couple across the wiring and the pot itself. One solution is to use a lower-value pot so that the bootstrapped impedance will be closer to the stock value (TUT4, p5-7)."

i tried replacing the 1M pot with a 100k pot, but the problem remains. has anyone experienced this? is there a fix, or a better way to do the bootstrapped MV? i know o'connor has an 'improved bootstrapped MV' section in a later book, but i'm... strappe for ca$h at the moment.

[soundmasterg]

Email KOC and ask him would be my suggestion.

Merlin tried out some of this in sim I think and the bootstrapped EQ definitly has some issues. The bootstrapped MV I haven't tried myself so I cna't comment directly on it.

greg

[mooreamps]

Honestly, I tried it once, and I didn't like it. It made a bad squeel at half volume. So, on that amp I changed it to just a volume pot just past the cathode follower. Now, I do the electronic volume control.


-g

[Gaz]

mine definitely doesn't squeal, but i'm not sure if it even sounds different enough from a typical pre-PI master volume to warrant all the trouble...

i really prefer no master volume, but there are always situations where it'd be nice to be a little quieter on stage without your basic amount of distortion. i might even try an PPIMV, but the whole thing gives me a headache - i just want to rip it out, give my self an earache instead.

and dude, KOC is like totally famous! i can write him!! his name had been acronymized!!!

[Merlinb]

Why use a bootstrapped MV?
The whole point of a bootstrapped MV is to reduce the loading on the tone stack, so increasing the available gain and range of the tone stack a bit.
If you lower the value of the pot then you basically lose this benefit, in which case there's not much point bootstrapping it! (Except to reduce a little bit of resistor hiss I guess, but it's a pretty pathetic contribution to overall noise).

The problem:
KOC says:
"There can be signal feed through at the 'zero' setting. This happens because the impedances are so high and the signal feeding the pot is quite high, too. The signal can electrostatically couple across the wiring and the pot itself."
This explanation doesn't really make any sense when you think about it. The wiper is at minimum, so there should be no signal between the bottom of the pot and the wiper- that's how a pot works t right? But if the signal were leaking electrostatically across the pot to the wiper/bottom, then... there should still be no signal between wiper and bottom (excuse the pun), they are connected together remember!

The leakage has nothing to do with electrostatic coupling, it is because signal current flows through the MV to the bottom of the bias resistor (which the wiper/grid is now connected to). This signal also appears at the top of the bias resistor (the cathode), but the cathode has a low input impedance, so the voltage signal here is attenuated somewhat. Result? You have a small differential signal between cathode and grid, and that is your "leakage" signal which gets amplified.
The larger the bias resistor, the more 'leakage' you will get.

How to cure a leaky bootstrapped MV:
Bypass the bias resistor on the LTP with a big cap (10uF or more I guess), or replace the bias resistor with an LED.

For those it interests:
A bootstrapped tone stack does not work at all - it just reduces the range of control of the tone stack, like a kind of partial 'defeat' switch. I don't know why KOC ever published it, it's a silly idea.

[Gaz]

merlin,

thanks so much for the detailed answer- just all the info i was hoping for and more! i think i will just revert to the average pre-PI.

there is a lot of missing info in KOC's tomes, precisely why i ordered your book last week, merlin! look forward to that

alex

[mooreamps]

Kevin is one of the most knowledgeable and talented engineers ; of whose books I've read. Maybe he likes the circuit, maybe he uses the circuit, maybe he does not. His books are published primarily for the educational value, and perhaps less focus on what works the "best". I would presume it would be up to the individual designer to determine which circuit is best to choose from. I do not always build "everything" I read from a text book, or what I see on the internet. .

-g

[benjofisch]

I am having the exact same problem in an amp I am designing except with a cathodyne PI. I am using a 1M audio taper pot instead of a grid leak resistor, wiper connected to grid, and at minimum setting there is still signal coming through. I am hoping it is not just leakage capacitance–I don't have much experience with how much an effect this can have. I tried bypassing the bias resistor with a 100uF cap to no avail. Looking around on other forums, it seems like this type of volume arrangement is always going to have this effect.

[Dave H]

I tried bypassing the bias resistor with a 100uF cap to no avail. Looking around on other forums, it seems like this type of volume arrangement is always going to have this effect.

The 100u should prevent the cathodyne passing the signal but I think the signal could be bypassing the cathodyne by going through the 1M vol pot directly to the grid of the power tube. Anyway, would you ever want to run the master volume at such a low setting?

[d95err]

Anyway, would you ever want to run the master volume at such a low setting?

My question too. With the master volume at zero, how loud is the amp? Louder than "bedroom" volume? If not, is the volume leak really a problem in practice, or it is mostly an academic issue?

[benjofisch]

Ah I hadn't thought of that! That makes sense. I'm using a 6v6 push-pull output and it is WAY too loud for my bedroom with the master volume between off and 0.5 The volume control doesn't do much, honestly... it gets extremely loud very quickly, and then continues to get louder throughout the range of control, but it doesn't feel logarithmic.

Perhaps I need more attenuation before my cathodyne–it could be that anything above "off" is essentially driving the power valves (and maybe even PI) into clipping, and I just can't play loud enough to really tell, being considerate of my ears and neighbors. This is only my second amp design (on a wooden board with euro terminals and Pete Millet's PCB's) and I admit I probably didn't do as many calculations as I should have.

Very interesting point that a signal can "leak" through a bootstrapped stage! Out of curiosity, does the input impedance of the stage depend on the potentiometer's position? I would assume so, since bootstrapping depends on the signal appearing across the grid leak. Thus, effectively no bootstrapping would occur at the minimum setting, and you would get a ~1/20 attenuation across the 1M volume pot and the cathodyne's 47k cathode load to be passed on to the power tube. At higher volume settings, then, the stage would have some bootstrapping and probably have minimal leakage, though the desired signal would probably be much louder anyway. Am I thinking about this correctly?

Thanks!

[pdf64]

...I think the signal could be bypassing the cathodyne by going through the 1M vol pot directly to the grid of the power tube...

I'm a bit fuzzy how it might work, but shouldn't the low (internal) impedance at the cathode effectively short the signal out?

[benjofisch]

I'm glad I've snagged a few interested and seasoned circuit designers! I disconnected the PI from the power stage and signal is still getting through! Maybe it's bad power supply isolation? I'm using a knight KG-664 (400V,200mA) at 285V pulling ~54mA. I'll try using bigger dropping resistors for the preamp stages, as I'm already using 22 to 47uF bypass caps. If it's stray capacitance.... then so help me god.


EDIT:
Tried this and it didn't work... the master volume control actually affects how loud the leakage volume is–and the grids of the power tubes are still shunted to ground...

EDIT2:
Definitely a capacitance issue. If i move my probe around the setup, near tubes, capacitors, signal carrying wires, etc., the waveforms appear on the scope. They also definitely appear on the grid of the power tubes even though nothing is connected there.

FINAL EDIT:

Fixed... by abandoning the idea altogether. Once I reigned in the gain (which was much too high), the capacitance was not an issue any longer.

But the master volume pot still wasn't right–it would increase the volume rapidly and then taper off. The reason I think, is that only the lower portion of the potentiometer is bootstrapped, and this factors into the potential divider it creates. So as you turn the pot from 0%, the bottom resistance is multiplied by the bootstrapping and quickly begins to dominate the voltage divider. So it kind of ruins the logarithmic taper, and would be worse with linear or reverse log.

shouldn't the low (internal) impedance at the cathode effectively short the signal out?

Still curious to know if this is true. When I tested with a scope and signal, the cathode DID have a signal even when the wiper was all the way down. It also had the near exact voltage that would be created by a divider of a 1M pot and 47k cathode load...

[Merlinb]

Still curious to know if this is true. When I tested with a scope and signal, the cathode DID have a signal even when the wiper was all the way down. It also had the near exact voltage that would be created by a divider of a 1M pot and 47k cathode load...

Yep. Remember, the cathode impedance is only low when you try to wiggle the cathode but the grid remains independent or grounded for AC.
If you short the grid and cathode together then you lose that independence; the valve is converted into an ordinary diode, so it no longer has a low cathode impedance. It just has the ordinary run-of-the-mill internal resistance (maybe 60k). So the bottom of the pot is now basically connected to 47k to ground, and 60k+47k to HT, i.e. a 1M-33k divider.

The whole bootstrapped MV thing is a big fat waste of time if you ask me.

[Dave H]

I'm a bit fuzzy how it might work, but shouldn't the low (internal) impedance at the cathode effectively short the signal out?

That's what I thought too but when I simulated it the signal still got through. It must be as Merlin says, the bootstrapping makes the cathode impedance look higher than it normally would be.