Overcoming interaction between control settings in FMV tone stack

[Mike Sulzer]

The primary problem with the FMV stack (with or without the mod described in the previous post) is that when you turn up the mid, the high frequencies go up , too. There is a fix for this; it does not turn it into a graphic equalizer but it is an improvement. It is only necessary to an a capacitor (about 6800pf with other normal values) from the slider of the mid pot to ground. The effect in one particular case is shown in the attachment.



Note the change in scale between plots. The top plot is with the treble most of the way down and the mid on zero. The next plot has the mid on 10 without the modification (no added C). In this particular case, you could not call the result a mid boost. The bottom plot is with the C from slider to ground on the mid pot installed. There is some mid boost, although you do not have nearly the flexibility as with a GE. However, IMO it is worth doing. If you start with normal settings and then raise the mid a bit, it gives a thicker sound. Kind of like adding flour to a stew that is a bit watery.

I assume that this mod has been used before (it is sort of obvious), but I have not seen it anywhere.

[Pedro Vecino]

I assume that this mod has been used before (it is sort of obvious), but I have not seen it anywhere.

Marshall has used this formula (0.0047 cap) in some combos. My interpretation (guitar in hand) is that it was to get an emphasis on perceived mids raising the control and limiting in a parallel form the harmonics in the upper side (a matter of pure contrast). At five and down it has no striking effect, but it rises gradually to the maximum.
I have always associated it with a solution for the hollow sound of the Marshall combos with G1275 in relation to the same amp (head and a closed cabinet).

[Chuck H]

Dumble did something similar on a couple of incarnations of the ODS preamp, but you have to dig past all the "rock", "jazz", etc. switching to see it. I worked on and ran sims for adding a cap a few weeks ago to solve for a shortcoming in one of my own designs. This one needs fairly low bass control resistance to sound "right" so I use a 10% log for the bass control and as a result, using the standard Fender topology, the mid control just sort of moves everything up and down. So I worked out a dual ganged mid control where the second gang adds resistance to the bass circuit as I turn the mids up. Now I can use the Marshall type (mid cap to wiper) arrangement for more independent mid function without much added LF through the circuit. Strangely, I tried Mikes proposed coupling cap to the stack and no cap to the bass control and it also makes for more independent control function.?. This amp is voiced to set up for power amp overdrive, so the tone stack is more mid voiced. I made a rough image of the circuit for this thread and some sims of mid control function (mid@ 0-5-10) treble stays at 6 and bass stays at 5. It's rough looking because I cut and pasted them over one another and stretched to scale (red alignment markers are @-8dB and -18dB). Magenta is a standard tone stack and green is the circuit shown.

[Mike Sulzer]

Marshall has used this formula (0.0047 cap) in some combos. My interpretation (guitar in hand) is that it was to get an emphasis on perceived mids raising the control and limiting in a parallel form the harmonics in the upper side (a matter of pure contrast). At five and down it has no striking effect, but it rises gradually to the maximum.
I have always associated it with a solution for the hollow sound of the Marshall combos with G1275 in relation to the same amp (head and a closed cabinet).

Interesting; thanks for pointing that out. I have been thinking that a possible application would be to some Fender circuits, ones that have a lot of attenuation, and thus a higher potential "boost" range, but with a deeper scooped middle that cannot be raised independently with the stock mid control. The capacitor would allow that to come up more without raising the highs as well, extending the adjustment range more towards the Marshall sound.

[Mike Sulzer]

Dumble did something similar on a couple of incarnations of the ODS preamp, but you have to dig past all the "rock", "jazz", etc. switching to see it. I worked on and ran sims for adding a cap a few weeks ago to solve for a shortcoming in one of my own designs. This one needs fairly low bass control resistance to sound "right" so I use a 10% log for the bass control and as a result, using the standard Fender topology, the mid control just sort of moves everything up and down. So I worked out a dual ganged mid control where the second gang adds resistance to the bass circuit as I turn the mids up. Now I can use the Marshall type (mid cap to wiper) arrangement for more independent mid function without much added LF through the circuit. Strangely, I tried Mikes proposed coupling cap to the stack and no cap to the bass control and it also makes for more independent control function.?. This amp is voiced to set up for power amp overdrive, so the tone stack is more mid voiced. I made a rough image of the circuit for this thread and some sims of mid control function (mid@ 0-5-10) treble stays at 6 and bass stays at 5. It's rough looking because I cut and pasted them over one another and stretched to scale (red alignment markers are @-8dB and -18dB). Magenta is a standard tone stack and green is the circuit shown.

So in going to a smaller bass control, (which is just a variable resistor), you restricted the bass boost range and spread out the available boost over the whole range of the pot, choosing the right taper for the best "feel". But this puts you in place where the mid performs even worse than usual, and so you stop the highs from rising too much with the cap. This leaves the problem with the bass rising with the mid, which you fix with a very clever use of a ganged pot using the path that couples the bass to the output. Nice.

[Chuck H]

Yeah, well, your idea for feeding the tone stack from a coupling cap and then using a plain lead to the bass control helped a bit too, though I'm not sure why. I did simulations both ways, with the bass cap in front of the tone stack and with the bass cap off the slope resistor to the pot. Definitely more independent function with the bass cap in front of the tone stack.

[Gnobuddy]

I applaud both your efforts, but I wonder if attempts to improve on Leo's "cheapest tone control in the world" don't fall under the umbrella of trying to polish the proverbial turd?

The two-pot version (as used in the Blackface Princeton Reverb reissue, etc) isn't too awful. The three-pot version of Leo's monstrosity though, is beyond help through minor tweaks, IMO.

Part of the problem is certainly the lack of buffering between controls, but another part is the rather narrow frequency bandwidth of the guitar. There are only five octaves fom 80 Hz to 2.56 kHz, and it is a tall order to pack three first or second-order filters with acceptable control ranges (say +/- 10 dB minimum) into that narrow a frequency band, without having them step all over each other's feet.

A first order filter has at best only +/- 6 dB of control range per octave, and when the poles and zeros get closer together, that drops to even less. If you only have ~4 dB of control range per octave, you need three octaves to get 12 dB of control range, and now you need nine octaves of frequency bandwidth to keep bass, mid, and treble controls from interacting...and you don't have that much frequency range to work with.

At one time I tinkered a fair bit with LTSpice simulations for a three-band active Baxandall tone control, and even that topology has a lot of interaction once you try to get adequate control range into the limited frequency range of the guitar.

I think one may have to either settle for two-band controls (which have enough frequency space to be non-interactive), or use something like an actual graphic EQ, with fairly high-Q bandpass / bandstop filters. Or come up with very flexibile parametric EQ.

As usual, bass gear is technologically more advanced than electric guitar gear (apparently because the typical bass player is much more open-minded than the typical electric guitar player), and those are the directions bass guitar EQ seems to have taken in recent decades.

-Gnobuddy

[Mike Sulzer]

To think about this correctly, start mentally with the M on zero and the B and T on 5. The low pass and high pass sections are separated enough so that you have a hole in the middle, which can be 8 db or a bit more with some useful values. The M control with the cap on the M pot allows the hole to be filled in smoothly as the M pot is raised so that it is mostly gone on 10. You cannot achieve this flexibility with just two controls. As I pointed out, it is not a GE, but being passive and all in one unit has advantages in a guitar amp.

The guitar bandwidth is limited by the speaker, typically about 5KHz. Even the most overwound humbucker, when played with distortion, can make use of the full bandwidth.

I applaud both your efforts, but I wonder if attempts to improve on Leo's "cheapest tone control in the world" don't fall under the umbrella of trying to polish the proverbial turd?

The two-pot version (as used in the Blackface Princeton Reverb reissue, etc) isn't too awful. The three-pot version of Leo's monstrosity though, is beyond help through minor tweaks, IMO.

Part of the problem is certainly the lack of buffering between controls, but another part is the rather narrow frequency bandwidth of the guitar. There are only five octaves fom 80 Hz to 2.56 kHz, and it is a tall order to pack three first or second-order filters with acceptable control ranges (say +/- 10 dB minimum) into that narrow a frequency band, without having them step all over each other's feet.

A first order filter has at best only +/- 6 dB of control range per octave, and when the poles and zeros get closer together, that drops to even less. If you only have ~4 dB of control range per octave, you need three octaves to get 12 dB of control range, and now you need nine octaves of frequency bandwidth to keep bass, mid, and treble controls from interacting...and you don't have that much frequency range to work with.

At one time I tinkered a fair bit with LTSpice simulations for a three-band active Baxandall tone control, and even that topology has a lot of interaction once you try to get adequate control range into the limited frequency range of the guitar.

I think one may have to either settle for two-band controls (which have enough frequency space to be non-interactive), or use something like an actual graphic EQ, with fairly high-Q bandpass / bandstop filters. Or come up with very flexibile parametric EQ.

As usual, bass gear is technologically more advanced than electric guitar gear (apparently because the typical bass player is much more open-minded than the typical electric guitar player), and those are the directions bass guitar EQ seems to have taken in recent decades.

-Gnobuddy

[Gnobuddy]

The guitar bandwidth is limited by the speaker, typically about 5KHz.

I would say the guitar bandwidth is limited by the nature of the instrument (there's nothing musically attractive above a few kHz). But we're essentially in agreement...

Even the most overwound humbucker, when played with distortion, can make use of the full bandwidth.

I may have failed to make myself clear in my earlier post. Given the full audio bandwidth (20 Hz to 20 kHz), you can make a three-band tone control with first-order (6 dB/octave) slopes, and still get enough control range, and still have enough frequency range for the three controls to not interfere/interact with each other. You can find schematics for active 3-band Baxandall controls like this on the 'Net.

Given the full guitar bandwidth (< 5 kHz), I don't think this is true any longer. With two octaves lopped off the bass, and another two or three octaves lopped off the treble, there is no longer enough frequency span for three non-interacting controls with 6 dB/octave slopes and adequate control range.

So, given the limited guitar bandwidth, and a desire to reduce control interaction, I think one either has to (a) go to two tone controls, giving each one more frequency span to work with, or (b) go to slopes steeper than 6 dB/octave.

Incidentally, 5 kHz may be too generous by an octave. 5 kHz is certainly in the ballpark for the on-axis frequency response of many typical guitar speakers, but nobody listens to a guitar amp on-axis to the speaker - it's usually harsh and nasty-sounding in that position. So we listen considerably off-axis, and these big, floppy speakers beam treble like searchlights. 5 kHz is down in the weeds when you are 30 degrees off axis.

For my current project amp, I tinkered with a graphic EQ pedal and found a frequency response peaking at 2 kHz and falling (second order) above that sounded best. In this amp, the speakers do not limit the treble within the guitar's range, so all the high-frequency rolloff has to come from the electronics.

Does a typical floppy-coned 10" or 12" guitar speaker go significantly higher than 2 kHz, when measured substantially off-axis? I wouldn't be surprised if the answer is "Nope"!

-Gnobuddy

[Gnobuddy]

In case this is of any interest, here are two pics, one of the graphic EQ settings that sounded best (clean guitar, 6.5" speakers with extended treble response compared to guitar speakers), and the other of the actual measured frequency response of the same graphic EQ pedal with the knobs set as shown in the photo.

The part of the curve relevant to the present discussion is the dramatic treble roll-off above a surprisingly low 2 kHz. Consider this to be "speaker emulation", because it is what was needed to make those 6.5" speakers produce the right amount of treble to sound like a guitar amp.

I suspect that when we set up an electric guitar amp the usual way - aim its 12" speaker at our calves, and stand nearby with our ears about 4' higher than the speaker - the frequency response of the guitar speaker may start rolling off treble as low as 2 kHz, too. A 2 kHz sound wave has a wavelength of less than 7 inches - far less than the diameter of a 12" or even 10" speaker - which means 2 kHz will most likely be severely attenuated that far off-axis.

-Gnobuddy

[Mike Sulzer]

I cannot get enough resolution on your plots to see all the details, but I can point out that there are various factors that enhance the response above 2 KHz, thus counteracting the roll off of the speaker.

First, the very thing we are discussing: the tone stack tends to "boost" some in that range with the controls on 5.

Then there is the pickup-cable resonance, varying from from pickup to pickup, with a maximun in the 4KHz range.

Also the magnetic law of induction, which contains a time derivative and thus rolls pickup frequency response up at 6db per octave.

Furthermore, the on axis response of a 12" inch guitar speaker usually rises above 2KHz, as much as 10 db before falling very fast at 5KHz.

And then, although your head is not directly on axis, you get some of that response from reflections.

Also distortion emphasizes harmonics; many are in that range.

Finally, the human hearing response peaks above 2KHz and below about 4 KHz.

In case this is of any interest, here are two pics, one of the graphic EQ settings that sounded best (clean guitar, 6.5" speakers with extended treble response compared to guitar speakers), and the other of the actual measured frequency response of the same graphic EQ pedal with the knobs set as shown in the photo.

The part of the curve relevant to the present discussion is the dramatic treble roll-off above a surprisingly low 2 kHz. Consider this to be "speaker emulation", because it is what was needed to make those 6.5" speakers produce the right amount of treble to sound like a guitar amp.

I suspect that when we set up an electric guitar amp the usual way - aim its 12" speaker at our calves, and stand nearby with our ears about 4' higher than the speaker - the frequency response of the guitar speaker may start rolling off treble as low as 2 kHz, too. A 2 kHz sound wave has a wavelength of less than 7 inches - far less than the diameter of a 12" or even 10" speaker - which means 2 kHz will most likely be severely attenuated that far off-axis.

-Gnobuddy

[Gnobuddy]

I cannot get enough resolution on your plots to see all the details

Sorry about that, I seem to have accidentally posted thumbnails rather than the full-size images I intended to.

but I can point out that there are various factors that enhance the response above 2 KHz, thus counteracting the roll off of the speaker.

There is certainly room for some discussion, but the bottom line is still that guitar tone controls have to work in a frequency range narrower by four to five octaves, than Hi-Fi tone controls do.

Also the magnetic law of induction, which contains a time derivative and thus rolls pickup frequency response up at 6db per octave.

This is certainly what happens when you excite a guitar pickup with a sinusoidal AC magnetic field of constant peak value and varying frequency.

However, this doesn't apply to the signal actually coming out of a guitar (just look at it on a 'scope!)

The reason is that the higher frequency notes come from strings that are vibrating through smaller amplitude, and also contain less magnetic material. Both those effects reduce guitar pickup signal, preventing the 6 dB/octave rise you mentioned from actually occuring.

The first of these effects (amplitude of vibration) can be modelled with simple math. A simple harmonic oscillator of mass "m", frequency w, amplitude x, has an energy of 1/2m w^2 x^2.

If we make the assumption that we are picking notes with uniform force, the guitar pick provides the same impulse (total energy) for every note. So the left-hand side of the equation above becomes a constant.

This means we have arrived at the result: wx = constant, where w is the (angular) frequency, and x is the amplitude of vibration.

Moving w over, we find that: x ~ 1/w, or the amplitude of vibration falls linearly as frequency rises.

This exactly cancels the effect of the time-derivative in the guitar pickup signal: the derivative causes the signal voltage to rise at +6 dB/octave, while the strings amplitude of vibration inherently falls at the same rate of 6 dB/octave.

Net result, if all the guitar strings were equally thick, the output from the guitar pickup wouldn't rise with frequency at all!

In practice, we have thinner strings for higher notes, we have non-uniform spacing between pickup pole and guitar string, we have unevenly varying string core thickness (the infamous too-loud unwound G!), etc. All those things will modify the inherently flat-frequency response of the (guitar pickup + guitar string) system.

-Gnobuddy

[Mike Sulzer]

Sorry about that, I seem to have accidentally posted thumbnails rather than the full-size images I intended to.


There is certainly room for some discussion, but the bottom line is still that guitar tone controls have to work in a frequency range narrower by four to five octaves, than Hi-Fi tone controls do.

-Gnobuddy

Yes there are various factors to consider. But the range above 2Kz and below 5KHz is very important for an electric guitar. This we know from trying guitar pickups with various resonant frequencies and Qs. In this range it is harmonics, requiring not so much power in order to have a big impact on the tone.

There is also no question about this: In order to make his guitars sound bright enough for him Leo Fender used a tone stack with a hole in the middle so that the response rises up to about 2.5KHz before flattening out. This is achieved with a high pass filter and the amount of these highs in the output is adjustable with the treble pot. The low side of the hole is set by a low pass filter. The low pass capacitor is connected to a pot that allows the roll off to be stopped at a variable frequency, adding mid and highs back in. The addition of the highs can be reduced by use of an additional capacitor as I have described. This is audible and useful IMO, whether or not the turds are enabled for specular reflections.

These things are true no matter how restricted you claim the frequency response is compared to HiFi. But really, HiFi contains just two more octaves, a region of rapidly falling sensitivity in the ear-brain, and not so important in the interior mental IT, especially as you get older and the sensitivity in the upper range falls to zero.

[Gnobuddy]

But really, HiFi contains just two more octaves

I disagree, so lets count them!

First, let's do a quick approximate calculation:

1) One octave from 20 Hz to 40 Hz.

2) One octave from 40 Hz to 80 Hz.

3) Taking your 5 KHz upper limit, one octave from 5 kHz to 10 kHz.

4) And one more octave from 10 kHz to 20 kHz.

I count four more octaves, not two, in the traditionally accepted Hi-Fi frequency band, even with the generous assumption that electric guitars need a 5 kHz bandwidth.


And now, let's do it with a bit more mathematical precision. The number of octaves (n) in a frequency interval from f1 to f2 is given by : n = log (to the base 2) of (f2/f1)

Converting to base 10 logs for convenience, this becomes: n = 3.3219 log (f2/f1) (log to the base 10 this time)

With f2 = 20,000 and f1 = 20, this calculates to 9.966 octaves.

With f2 = 5000 Hz and f1 = 82.4 Hz (low E on guitar), this calculates to 5.921 octaves.

Once again, the difference between the two is 4.044 octaves.

Four lost octaves is a lot - even a first order filter attenuates by up to 24 dB in that big an interval.

-Gnobuddy

[Mike Sulzer]

I agree; I meant two more octaves in the high frequency direction.