High gainer, clipping, post PI MV, tone shaping, etc!

[jimboyogi]

Hi All,

There is a very interesting thread in the 'Guitar Amps' section about get better distortion tone from a high gain pre-amp. I read it with interest, because I am having the same kinds of problems in trying to design one myself.

One of the messages that I got from this other thread was that low frequencies should be cut, and high frequencies boosted, before the pre-amp distortion stages. Then the LF should be boosted, and the HF cut, before the power amp.

But…

What if you are using a post PI master volume, when much of the distortion is generated in the PI, and hence after the tone controls?

I'm wondering the best way to then boost LF and cut HF after the PI.

I can think of a couple of options.
One is to use the VOX style 'cut' to lower HF.
Another idea is more of a question than anything else. Can the tone stack be moved after the PI, and instead of shunting signal to ground, shunt signal to the opposite phase? A crosscut style tone stack? Is this possible, or advisable? Will animals be harmed?

[jazbo8]

Isn't that like the presence control by varying the amount of NFB to control LF/HF gain?

[uneumann]

Hi All,

Can the tone stack be moved after the PI, and instead of shunting signal to ground, shunt signal to the opposite phase? A crosscut style tone stack? Is this possible, or advisable? Will animals be harmed?

Yes - I think animal harm is a certainty... :-)

Sure - I'll go on a limb and say you "could" do it. I can see a few problems though... First, a passive tone stack attenuates the signal, so you'd have the problem of drive levels to the output tubes. Second, a classic passive tone stack has a single-ended output. You'd need two to handle both phases, or you'd have to redesign it to provide dual inputs and outputs.

Jazbo8's comment about the presence control is likely a better direction. Merlin's book has a good section on feedback tone controls in the power stage.

[guitician]

The PI is part of the power amp, so having the tone stack before the PI is what you want.

[jimboyogi]

Thanks for your responses.

jazbo8, after reading your post I went and read up on 'resonance' controls, which are the opposite of 'presence' controls, and can therefore increase LF output, albeit with decreased damping on the LF. It's a good idea, but as this amp has a PPIMV inside the NFB loop, the LF gain would reduce drastically as the MV was turned down.

uneumann, I fear for the animals also. Your point about a tone stack decreasing the amplitude of the signal voltage to the output tube grids is a very valid one. Especially as this is using a cathodyne, so it's not that easy to get a big signal out of the PI anyway.
WRT needing 2 outputs, I was envisaging the TS working 'cross line' between the 2 out of phase signals. 1 phase would go to the TS input and output, and the other phase would connect where the TS was grounded in conventional usage.


I have finally got to some experimenting tonight. I have been playing around with a Vox style treble cut, cross line between the 2 PI output signals. I have attached a pot and cap network between the phases, post the PI output coupling caps, effectively in parallel with the output tube grid leaks (250K per tube, so 500K total grid-grid). The cathodyne is the triode half of a 6U8, medium mu, Rp of ~ 5.5K on data sheet, Ra and Rk of 47K, output coupling caps = 22nF.

Does anyone know how I can calculate the required capacitor value/freq response in the cut network, if I use a 250KA pot?

[Chuck H]

I can see attenuating highs after the PI, but that's about it. I don't personally like the affect that NFB loop tone controls have on the distortion character of the power tubes. NFB does different things than passive attenuation does. Not that "different" = bad, but high levels of NFB, as might be used in such a circuit, squelches harmonics and overtones and basically does dirt to your efforts.

This is a sort of "in the box" approach, but an amp either clips the power amp, or it doesn't. If you want distortion followed by EQ control you need to generate your distortion before the EQ control. WRT power amp distortion there isn't a good way I know of to do that. To me it's sort of like compensating for a car with bad brakes by only choosing routes that go uphill. It's not practical. Your EQ controls for power amp distortion are basically OT, speaker and cabinet choice. If you want power amp distortion AND EQ control the best solution I know is to design with a small power amp, pad and then re amplify with a bigger, clean power amp (been done), play an existing amp into a load and then process and re amplify with another amp or use a fancy attenuator designed for tone shaping. For power amp clipping it's just too much trouble trying to manipulate the operating conditions, control the frequency and deal with typically component value unfriendly impedances simultaneously without sacrifices. The power amps job is to make power. That it clips favorably is a bonus. Now asking it to also be the EQ is too much. If you try to make it serve triple duty the most common end is mediocrity on all fronts.

JM2C

[guitician]

Here's an online calc for RC filters... RC pad corner frequency upper and lower cutoff frequency calculation filter calculate time constant tau RC voltage power calculator capacitance resistance - sengpielaudio Sengpiel Berlin

[Tooboob]

Hey! No fair! You answered his question!

I don't think anybody proposed cutting into an amp that's working pretty good to cut bass before the lead preamp and boost it afterward as the killer mod that lets you stop soldering and go play. A better claim would be that distortion can be and is frequency dependent and you can intentionally manipulate it without changing the general frequency response at the output of the amp. It's not unusual to see a low-pass function after a distortion stage to take some of the edge off the clipping. Cathode caps provide a bit more gain, but they are frequently chosen with a bandwidth in mind. The design may parallel the supply resistance or plate resistor with an RC circuit or capacitor. The supply cap could be too small. The designer already tuned the original circuit, so any "improvement" is a matter of personal preference, and your preference on pre- and post-filtering could run in either direction.

The result is different distortion, and its quality is subjective. Before you start cutting, try an EQ between the guitar and amp, and one in the effects loop. Try different complementary settings.

Obviously there are lots of frequency-dependent things going on in the preamp already. Any cathode cap, the coupling caps, the plate resistors, the supply filter cap, caps parallel to the plate resistor, etc. are shaping the response, and if they are before a distortion source, the distortion will change. It's generally true that you can't change any component much without changing one or more things downstream to compensate. It's easy to get lost in the experiment, or change only one component and judge that change based on the "uncorrected" result. I like to try changes that result in an understandable change, and pre- and post- filtering let you directly varies things that you normally listen for - frequency response and the character of the distortion. One guy tried it with a preamp he was unhappy with, and found he preferred to cut, distort, then boost the treble, not the bass. Once you find something better for you than what you've got, you may be able to get close by changing existing components, and two or more distortions can be made switchable.


The advantage of the Post-PI MV is its biggest problem. Since it lies inside the NFB, as you turn it down, the PI gets less feedback, and drives the power tubes harder. Output volume isn't reduced effectively until the PI distorts and breaks the loop. You still need a pre-PI MV or the channel volumes to adjust the level of the undistorted portion of the signal. The two knobs are interactive. Now, if you were to simultaneously boost the PI output by decreasing the feedback, and simultaneously reduce the PI output to the power tubes, you could add an adjustable amount of PI distortion at most reduced volume levels, and you'd be left with the original pre-PI volume, working as a single volume control, and a new distortion threshold knob, reducing the need to iterate to find a desired setting.

PI distortion is pretty good, and there are many Post-PI master volumes out there. You still need a pre-PI MV or the channel volumes to adjust the level of the undistorted portion of the signal. Lots of folks run their Carvin X100Bs at the 50% setting, where PI distortion takes over for power tube distortion. Carvin does their magic by dropping the supply voltage to the PI circuit:

http://www.carvinservice.com/crg/sch...01C29apr08.pdf. Thirty years and still shipping, now with a pre-PI MV.

The Post-PI MV isn't the only circuit with the "inside the loop" problem. If you power scale by changing the power tube plate and/or screen supplies, again, the PI will try to compensate. O'Connor doesn't make a big deal of it, but he pretty much always breaks the loop with an ultra-linear transformer (he sells one), avoiding the problem. He also proposes various solutions he finds adequate that drop the supply to other portions of the amp, including the PI.

Ideally, for a distorting power stage, the PI/NFB would simply linearize the output, especially during crossover, and let the power tubes follow their own characteristics when overdriven. In reality, the PI will correct for the onset of power tube saturation by driving harder, making the clipping much more abrupt as the PI output pegs positive. Breaking the loop with single-ended cathode resistor feedback or an ultra-linear transformer can help the circuit clip like the power tubes do naturally, though both make the onset of power tube clipping a bit more abrupt. Your ears can decide if this is better, but at least it gives you a real choice between, say, 6L6s and EL34s, instead of substituting PI/NFB induced mayhem.

Many adventures are possible.

[jimboyogi]

Thanks for your help!

Chuck H - yes i should have been clearer in my previous post, but I have shelved (……..) the post PI tone stack idea, because of the reduction of signal voltage to the output tubes, as well as the NFB issues that it would create.
I am concentrating instead ATM on a post PI treble cut, and trying to find the appropriate values for a cross line RC one.

guitician - thanks for the link to the RC calculator. From my limited understanding of the cross line treble cut, I think the filter has the R and C in parallel, a situation not covered in the online calculators I have found so far.

I am trying to think through the maths for this calculation (please be kind to me, I'm a winemaker and not an EE).
When the 'cut' pot is short circuited (max cut), then the treble cut cap C is in parallel with all of the impedances that occur between the 2 PI outputs.
So far the impedances I can think of are -
the triode's plate-cathode impedance (6U8 triode, plate resistance on data sheet of ~ 5.5K)
the triode's Rp and Rk, which are 47K each. My understanding is that the Rp and Rk are in series here?
the output tubes grid leaks, 250K per tube, in series, so 500K phase-phase

If I have the above correct, then all these impedances are paralleled, giving an R of 5.1K

Using the formula f = 1/(2*pi*R*C)

I get a -3dB frequency of 290 Hz for a C of 100nF.

Seems like a very big cap!

[uneumann]

OK - well I won't comment in detail about all that, Tooboob, but post-PI MVs do not really require a pre-PI MV or work that differently from other MVs. For example, the THD Flexi-50 has a post-PI MV and it operates very much like a pre-PI MV. The MV is in the FB loop but it's behavior is not all the different from my other amps with pre-PI MV controls.

IMO - where you put the MV is mostly based on what stage you most want to overdrive. The last stage before the MV will usually clip first, so you better like the tone of that clipping because the MV will get you lots of it.

[jimboyogi]

Hi Tooboob, thanks for your reply! I was just typing out a response when you posted. This fooling around is on a breadboarded amp that I am experimenting with, so no cutting, and lots of options as far as varying many things.

What you wrote about the PPIMV and NFB is interesting. My understanding of how it works is that when the MV is wide open, the max NFB is operating, lowering the gain of the PI, and linearizing the entire loop (PI, output tubes, OT). As the MV control is reduced, less NFB is 'shared' through the loop, increasing PI gain, with less linearizing of the signal everywhere in the loop, allowing for more harmonic and frequency distortions. I thought that this was a good thing, with MV dimed all aspects of the loop are as clean as possible until the output tubes begin to clip, therefore max clean output this way. With the MV reduced, more distortion is coming from the PI, which is less affected by NFB, at the same time the actual output volume is reduced as the output tube grids don't get as much signal. So quieter, and more PI distortion.

[uneumann]

Thanks for your help!

So far the impedances I can think of are -
the triode's plate-cathode impedance (6U8 triode, plate resistance on data sheet of ~ 5.5K)
the triode's Rp and Rk, which are 47K each. My understanding is that the Rp and Rk are in series here?
the output tubes grid leaks, 250K per tube, in series, so 500K phase-phase

If I have the above correct, then all these impedances are paralleled, giving an R of 5.1K

Using the formula f = 1/(2*pi*R*C)

I get a -3dB frequency of 290 Hz for a C of 100nF.

Seems like a very big cap!

5.5K is a low plate resistance for a triode - are you sure about that? It's dominant in your calculations.

As a sanity check, have a look at this Vox circuit. It's using 12ax7 (~40-50k for Ra) as the PI, and uses 4.7nf.

[jimboyogi]

Right you are uneumann! 5.5K is very low. I got that value from the data sheet, but that was for 'normal operation'. I have just had a look at the plate characteristic curves for the 6U8 triode. The bias that the cathodyne is running is ~10V, and if I use the -10V plate curve and calculate ra, then i get a figure of ~ 40K. My new total parallel impedance is then 27K
Substituting this into the f formula, a 22nF cap gives a -3dB point of 270Hz. 4.7nF gives 1.25 KHz.

Not sure if I am doing this wrong, but I think that it's time to try a few caps and measure freq responses.

[uneumann]

Not sure if I am doing this wrong, but I think that it's time to try a few caps and measure freq responses.

IMO - That's the best way to proceed at this point. The theory tells you why you're doing it. The rest is up to your ears.

good luck

[jazbo8]

Still don't see how you got 5.5k for 6U8's plate resistance, can you post the schematic or provide the plate current for the cathodyne? In any case, the formula for calculating the hi-cut is C = 1 / (6.28 * F * Zo), where F is the corner frequency and Zo is the plate resistance in parallel with the plate resistor.