Some background

Please be aware that I'm not an electrical engineer. I don't really have any qualifications or formal education for electronics.

I have been interested in wah pedals probably since I got my first electric guitar. I have also been interested in building effect pedals for almost as long.

When R.G. Keen published his The Technology of Wah Pedals article back in the late 90's, I eagerly read it, but I couldn't quite understand everything about that Vox/Dunlop -style circuit. I'm not even sure how much I understood at first. I have reread the article countless times. At some point I understood how the capacitors apparent value changes (maybe, or at least had/have some idea of it), but I couldn't really see where the filtering actually happens in the circuit when comparing to simple passive RC or RLC circuits, or even opamp circuits that use capacitors in the feedback path. The information is actually there in the article, but somehow I always missed it.

Because of the mechanics, it's usually not viable to build a wah from scratch, but a few years ago I had an idea. I have a feeling that the standard Cry Baby and others like it are not quite optimal for how I want the wah to feel. The travel of the treadle is quite short and somehow they are always kind of stiff and if you loosen the bolt it usually gets too loose. It always feels somehow wrong to me no matter how I tweak it.

The idea I had was to build a wah to the shell of an Ernie Ball volume pedal. Those are quite expensive. Most wah pedals are cheaper. So, it's a bit dumb to buy one just to use it as a case for a DIY-project, but screw it. I wan't to try it.

This article relies heavily on R.G. Keen's The Technology of Wah Pedals article. If you haven't read it yet, go read it first. On schematics I'm trying to use same part numbering where applicable.

Other sources of information I used:

• Active Filters Using Gyrators - Rod Elliott (ESP)

• Follow The Leader - Voltage Followers & Buffers - Rod Elliott (ESP)

• Simple, Easy Parametric and Graphic EQ's, Plus Peaks and Notches - R.G. Keen 

• Wah annoying RF oscillation on breadboard - thread on diystompboxes.com forum

• Frequency response of different inductors in Crybaby wah - thread on diystompboxes.com forum

• Vox/Crybaby Wah Transfer Function - thread on diystompboxes.com forum

• Experimenting With a Wah Circuit - yet another thread on diystompboxes.com forum, but this one was started by me while experimenting with this stuff.

• Probably something else I forgot...

I have already bought the volume pedal to use as the case, but I've mostly too busy and sometimes too lazy to finish it. I have also wanted to do some experiments with the circuit. While experimenting, I have found some things. Let's start from where the filtering actually happens. Some time last year I finally got it. (I think.) I'm no electronics expert and I'm only tinkering with this stuff as a hobby. I'm writing this down so other hobbyists like me can benefit from my findings. Sometimes it is useful to have another angle to the things to gain better understanding.

Table of Contents

Part 1

Where does the filtering happen then?

It's at the input!

I had always thought it was somehow after the first transistor stage and it had more to do with the feedback than it actually does. The key point that I missed for decades is that for audio signals the top end (on the shematics) of the inductor Lf is grounded by the capacitor Cbp.

The information is actually there, but somehow I always missed it.

The red arrow is the path of low frequencies that are filtered out. The blue arrow is the path of the path of the high frequencies filtered out. The frequencies between those will pass to the first transistor.

This is ofcource a crude simplification. The reality is that some frequencies get attenuated more and some less. The filtering components (inductors and capacitors) will affect phase too and I quess the combined response is a bit diferent than it would seem based on the amplitudes of single filtering components only. I quess that's one of the reasons the amplitude of the peak doesn't really change that much when the apparent capacitance of the parallel LC circuit changes... This is only me quessing though.

After understanding this, the signal from Q2 changing the apparent Capacitance of the Cf capacitor and moving the resonance frequency of the Parallel LC-circuit starts making sense to me.

Word about testing and schematics

I actually tested the original above and most of the variations under on a breadboard. I used BC549 transistors mainly because I have bunch of them. Other components I layed on the breadboard are also as in the  schematics, but I don't really know the inductance of my wah inductor. I don't have a meter that can measure inductance, and I'm currently way too lazy to rig up anything I could use to figure it out. It's a yellow Fasel I bought from a local suplier a while ago and the resonant peak of the effect seems to fall aproximately in the range it is supposed to. I haven't really measured all the other parts, but they should be within regular tolerances.

The pot I was using is 250 kohm, because that's what's in the volume pedal I'm planning to use for a case, and its more fun to experiment when you have a rocker pedal to play with.

I also had a buffer in front and after the actual wah-circuit, because I'm planning to use those in my build anyway, and experimenting is easier when the input and output is somewhat isolated this way. I have left the buffers out from most shematics for clarity. I'll make the my final schematic with them when I get to it.

But lets get back to the experiments...

What can I do with this newly gained understanding?

Well... Better understanding is always usefull when tweaking circuits to your liking. I had a few ideas I wanted to try to the circuit and understanding this actually made it clear that at least some of them would not work. However, I figured out something else that is not necessarily related to the performance of the circuit, but still interesting (at least to me).

Let's start by rearranging things a bit.

I added Cbp2 cap and connected one end of the inductor Lf to that instead of the original Cbp. The difference is mainly in the feedback biasing. The DC resistance of Lf is low. The one I'm experimenting with is 15 ohms.  This leaves only the 33 kohm Rq resistor to provide the the feedback biasing to the base of Q1 through Rfb. Hovewer, Rb1 is still quite large compared to Rq and even Rb2 is larger, so it might not be too big of a deal.

For AC I think this is practically same. It does not matter if the other end of Lf is connected to ground through Cbp or Cbp2. Rq is still practically in parallel with Lf. The positive side of both, Cbp and Cbp2 is practically ground for audio signals.

I didn't measure the biasing in this stage, but the next one is almost identical. I'll measure it there.

This may seem quite pointless... All I have achieved is using up one more capacitor, but let's keep rearranging just a bit more.

The order of passive components in series doesnt really matter, so I switched Cbp and Lf Around.

Now the bias. Voltage I measured from the collector of Q1 was about for the original version was about 4,9 volts and for the rearranged version it was just tiny bit under 5 volts. So the change in biasing is insignificant, at least for diy projects. I'm going to tweak it anyway.

This might still seem a little pointless. It still doesn't do anything the original circuit couldn't. However, now one end of the inductor is actually grounded.

Replacing the inductor with a gyrator

Inductors can be replaced with a gyrator circuit, made from an active part acting as a buffer, a cap and a few resistors. The requirement is that one end of the inductor to be replaced has to be grounded. I don't really understand gyrators on a deep level, but I can copy bits and pieces from other schematics and use online calculators or calculate part values using formulas on the internet. Good source for basics on gyrators is Active Filters Using Gyrators article on Rod Elliott's site.

The values for the gyrator were calculated with the formula given on that Gyrator article. (Google works pretty well on the calculations if you manage to give units in the form it likes.) It comes out as a bit under 500 mH. That can be easily changed by changing Rg2 or Cg1. The values used were mainly chosen based on what I found from my junk boxes. I aimed to produce something close to the real inductor I have.

Resistor Rg1 defines the series resistance of the inductor the gyrator is simulating. 33 ohms is more than resistance of the the inductor I used on the previous stages of this experiment. However, it is quite low for the opamp to drive so I didn't want to go lower. Most examples of gyrator circuits seem to use 100 ohms at lowest. However, the values shown above seem to work on my breadboard.

The higher resistance of the inductor might not even be too big of a problem. There have been a few threads on on diystompboxes.com forum discussing the series resistance of the inductor. Links to the thereads:

• Frequency response of different inductors in Crybaby wah - thread on diystompboxes.com forum

• Vox/Crybaby Wah Transfer Function - thread on diystompboxes.com forum

Based on those it seems like it mainly affects the resonance peak on the the darkest end of the sweep (typically heel down on the pedal).   For the way I want the wah to work, I might want to tame the resonance on the low end a bit.

The odd thing to me is that I do not remember any wah schematics I have seen, that replace the inductor with a gyrator in the regular Vox/Dunlop style wah circuit. I have to admit that I haven't been looking this stuff for years and I didn't search for it after figuring this out.

Testing and sound sample

I mostly tested this using pink noise from a computer as the signal source and fed the signal back to a computer for monitoring and spectrum analysis. This provided nice and easy way to hear the filtering without having a guitar on me all the time. This arrangement has one major flaw though.

That flaw became apparent, when I started recording sound samples to compare the inductor version to the gyrator. The filtering was fine for guitar too, but the background hiss is about 16 dB louder with the gyrator than with the real inductor!

Soundsample is below. I used a Youtube video for this, because Google sites doesn't really support embedding audio on the pages. The full schematics for what was on the breadboard is on the video for now.

To Be Continued...

For now it seems, that the gyrator version is too noisy to be usefull. Maybe this is why gyrators have not been used in this kind of wah circuits. I have an inductor for my build and I'm probably going to use that.

However, this is still interesting to me. I might try scaling the impedances on the gyrator to see if it affects the noise level. I might also try using a discrete transistor for the gyrator circuit to see if that lowers the noise, as I was going to experiment with that anyway.

I already tried using NE5532 instead of the TL072 for the opamp and the noise was a bit lower, but still not as low as the real inductor. I didn't measure the difference though.

I also have a few other things I want to try with the wah circuit. I will update this page if I find anything usefull or interesting.

Part 2

Gyrator noise measurements

Different Resistor Values and Different Opamps

I decided to test the noise levels with a few different resistor values and on NE5532 in additoin to TL072. I precalculated some part values based on parts I have in my parts bin.

All tests were done on version of the circuit with the input and output buffer in place like the ones on the soundsample video above. It might have been a better test to just leave those out, but I was too lazy to take them out. I'm planning to build a wah with buffers on both ends anyway.

Tests were done with cable running from audio interface to the circuit and from there back to audio interface. There was no signal coming from the output of the interface and I only recorded the residual noise of the combination back in. The input of the circuit was connected to the output of the interface mainly because I also tried some test signals to test the filtering anyway and I did not want the circuit input floating on the noise tests. The interface is Focusrite Scrlett 2i4 gen 2. Recording software I used was Ardour, mainly because I already had that on for the test signals for the other testing. The comparison analysis for the noise levels was done to an audio file exported from Ardour and loaded in to Audacity. That was mainly because Audacity has a quick and handy contrast analysis tool.

Tests were done with the wah pot fully on the treble side. That seemed noisiest and the noise is most audible there anyway. So, that's the most relevant part of the sweep for the noise.

Caps were some sort of plastic caps. They all are red and say WIMA. :) At least they are same brand and most are probably similar type. Resistors were carbon film and from same assorted pack. For NE5532 I used 10pF caramic cap between the power pins to make sure it will not oscillate. It would have been good idea to do that for TL072 too, but I forgot about it when recording the test clips. I quickly tested it just by ear with one set of resistor/cap values and it did not really make an audible difference.

It should be noted that the measurement is a bit unfair for the inductor version, as the parallel resistance for the inductor is lower for the gyrator versions. Inductor version only has Rq there, but the gyrator versions have Rq and Rg2 in parallel. (It doesn't look like that immediately, but in practice it is.) That means the filter Q is lower and the gain on the resonance is lower.

Metal film resistors are a bit quieter in theory. I picked out a set of values I have as metal film and tested them against carbon film. Here are the numbers:

You may notice that even the carbon film numbers are lower than on the previous measurements with same values. These measurements were done on different day. I did do the reference measurement for the inductor again too, so the comparison should be valid. I quess something changed. I could have bumped the gain on the interface and the interference from external sources might vary from day to day. The circuit was on breadboard so it's not really ideal when it comes to noise levels. However, I'm pretty confident that each measurement round was consistent enough within itself to see what makes a meaningfull difference.

I'd say the resistor type is insignificant here.

Based on these measurements the opamp type is very significant, and the value of the Rg1 resistor is very significant.

The value of Rg2 seems to have a small effect, but it also is in parallel to the simulated inductance. That means it is effectively in parallel with Rq and it lowers the Q of the filter. By listening to the noise I hear that the noise is affected by the filtering. So, higher the resonance, more gain the noise gets on the resonant frequency. So lower resonance probably just means that the noise was dampened a bit more on the loudest frequency and that caused the lower measured noise level. Same damping will also cause the wah effect itself to be milder. I actually tried it with a guitar too and it works that way.

With NE5532 and 220 ohm Rg1 it might be possible to get usable results based on how it sounds to me when testing. I haven't really played much of actual guitar through it though. It seems that going even higher value Rg1 might lower the noise, but it also defines the series resistance of the simulated inductor. As mentioned earlier, that seems to affect the resonance peak height at the low end of the sweep.

Discrete Transistor Based Gyrators

One more thing I wanted to try is using a gyrator based on a discrete transistor instead of an opamp. I had two main reasons for this.

Firstly, it might use less current. At band practice I use only three pedals, and the current ones don't really use much current. The batteries last almost forever in them. If I replace the wah with something more current hungry, it would make sense to go with a powersupply. That means the for quicker setup I'd need to build a pedalboard for those. That would be a lot of changes just to switch the wah for a new one.

Secondly, now that I have run into noise issues, I want to see if the transistor version is less noisy.

I quess I would also be slightly annoyed to put an opamp in the classic wah circuit, even if replacing the inductor with gyrator takes it quite far from the actual classic circuit. I'd count that only for a half of a reason so maybe that's kind of two and half reasons for twidling around with the transistor version. :)

For reference I used Rod Ellliot's Active Filters Using Gyrators article again. It states that there is a large difference in perfomance to the opamp version, but I think for a guitar effect it might not even matter. If the noise is lower, it might be worth it. Any distortions or other deviations from ideal inductor performance it causes, might even sound good.

When testing the one above, it seemed to be slightly less noisy than the version with NE5532 opamp, but the sharpness of the resonance (or well... the Q) seemed to suffer a bit. I only tested with a pink noise signal, but it seems to show how the filtering works quite nicely. I quessed that the either the buffer (Q5) is too weak to drive the 220 ohm Rg1 or the not so perfect gain of the transistor buffer causes theseries resistance of the simulated inductor to be more than that 220 ohms. I decided to try replaicing the buffer with another one on another Rod Elliots article (about buffers). There was no great science behind which one I chose. I just picked one that seemed simple enough, had lower output impedance and gain closer to one. Then I just smashed it in there with part values and biasing that looked like they would work for a 9 volt single supply.

It did work and the resonance seemed to be similar as with the opamp.

I measured the NE5532 version, single transistor gyrator version and the two transistor gyrator version against the inductor and had the following results.

The one transistor version is the quitest, but the resonance is audibly lower than the other versions. I quess that the apparent lower noise is not actually real. There is less gain through the circuit at the resonance so that's why there is less noise. The actual signal fed through the circuit will probably be lower in volume too.

The version with the two transistor gyrator is 0.7 dB quieter and the resonance sounded identical to me. That said, I will not try to claim I will hear the difference of 0.7 dB especially when I didn't actually arrange the circuits so I could switch from one to another instantly.

One more comparison with a sound sample

I tweaked the gyrator values a bit to make Rg2 larger (100 kohm). That way it should affect the Q less. Cg1 was lowered to compensate (0.022µF) to keep the simulated inductance in a similar value.

There is still more noise with the gyrator version, but for clean sounds it doesn't seem too bad. The difference will probably be a significant if you place an overdrive or a distortion after the wah (or use your amp for overdrive).

The measured difference in noise levels on these test recordings is about 10 dB. The difference compared to the previous tests is probably because of lower input gain on the audio interface. I had to lower it quite a bit because playing an actual guitar through it would have clipped. The level was probably set somewhat similar to the first audio sample. I quess the noise floor of the recording/measuring setup limits the results for the real inductor version at least when the gain is set low enough for real world use.

The tone is a bit different on this gyrator version. The darker end of the sweep is thinner. The reason for that is the series resistance (defined by Rg1) of the simulated inductor is higher. I could add a resistor in series with the real inductor to match the gyrator, but I don't really see the point. I might add a bit of series resistance for the inductor in the wah I'm building if the low end feels like too much, but it will probably be more like 20 ohms than 200 ohms.

If I wanted to continue refining the gyrator version, I might try to scale up Rin resistor and see if the series resistance of the simulated inductor will affect the dark end of the sweep less. That would ofcourse lead to quieter wah, but that could be solved by tweaking a bit more gain from Q1. That has the added benefit (to me at least) of making the sweep range wider. I might continue experimenting with gyrators some time in the future, but for now I'll move on. I have a few other wah experiments in mind.

Again, I will update this page if I find anything usefull or interesting.

Part 3

Inductor Magic

The heading above is pulled straight out of R.G. Keens The Technology of Wah Pedals article. If you still didn't read it go do it now. If it feels too long, at least read the Inductor magic part. (And if it really is too long, how are you reading this? :)

So, some vintage wahs have an inductor that saturates asymmetrically and that is supposed to sound good. Asymmetrical distortion is something that is often considered to sound good in effects that are specifically meant for producing distortion, like fuzzes and overdrives and well... distortions.

Using a Diode to clip

If I've understood it correctly, when the inductor saturates, the inductance decreases and it essentially starts to look much more like just the winding resistance. I had this idea that maybe a diode across a regular inductor could simulate this behavior at least partially. When there is not much voltage over the inductor, the diode does nothing. When the voltage rises to the diodes forvard voltage threshold, it starts conducting and the inductor gets bypassed.

However, an actual inductor saturates more easily on low frequencies. Diode doesn't care about the frequency. If the voltage goes over it's threshold, it starts conducting. When in parallel of the inductor, the diode might actually conduct more easily on higher frequencies. It is easier to get a voltage difference on the ends of an inductor on at higher frequencies than on low frequencies, because the impedance is low in the low end and rises when the frequency rises.

I thought that it would not be same as real inductor satuaration, but I was not too concerned about that. I was more interested if it makes a nice sound and if it would maybe do a similar effect but extending to higher frequencies.

What I was thinking looks like the alternatives A and B in the next picture.

When doing some preliminary testing it seemed to make some kind of difference when playing through it. I used some random germanium diode for the test.

However when I tested it more thoroughly, it seemed to just make the Q of the filter lower. I took the diode out and measured the AC voltage across the inductor. I tested this with a sine wave input. With 0.1 volts AC on the input of the wah, maximum AC voltage across the inductor was about 0.03 volts. That was on the heel down position when the frequency of my test signal was tuned to the resonance of the filter (407 Hz). I also tried several other wah positions and the maximum voltage across the inductor seemed to be 0.025 when the test signal was tuned to the resonance. Only on the lowest setting it was a bit more. When setting the test tone frequency further away from the resonance, the voltage across the inductor fell quite quickly.

I tried to increase the input signal to 1 volt. That makes the output of the full wah circuit already distorted. The voltage across the inductor only increased to 0.113 volts. This means that even a germanium diode wouldn't really start conducting with any remotely sensible signal voltage.

The other thing to observe here is that the voltage over the inductor did not increase linearly. Maybe I have to test the inductor separately at some point. Could it be saturating already at that voltage.

Back to the germanium diode I was using while testing this. It seems leaky. I used my multimeter's resistance measurement on it. Even in the reverse direction it gives some readings, and not even that high readings. Silicon ones don't do that. I'm quite sure that the difference I originally heard was that the leaky diode parallel to the inductor and Rq was lowering the Q.

So, this does not work. While discussing this on diystompboxes.com forum I started to realize that the inductor saturation is function of the current anyway, and I was trying to replicate it with clipping the voltage.

Another attempt at the saturation

I still did another experiment on clipping the signal at the inductor. Instead of the real inguctor I used the gyrator and tried to bias it too close to the power rails  so it would clip. I tried both... Biasing it close to the ground and close to the supply voltage. Neither did anything special. It worked quite regularly for most part. When I got really close to the power rails it just kind of started sounding more and more like the inductor was not there. There was some filtering, but it was just what Cf does without the inductor.

So... Only way to get the sound of inductor saturating in a wah seems to be getting an actual inductor to actually satuarate. I felt that it is time to stop chasing unicorns and move on to other stuff.

It worked quite regularly for most part. When I got really close to the power rails it just kind of started sounding more and more like the inductor was not there.

Pot Taper

The control curve of wah pot is usually somewhat similar to audio taper (logarithmic) pots. There is good info on pot tapers on this Rod Elliot's pages in articles called Beginners' Guide to Potentiometers and Better Volume (and Balance) Controls. For wah specific info on the pots, we can go back to R.G. Keen's The Technology of Wah Pedals again.

One thing to note about the wah circuit is, that the pot really works as a volume control in the circuit. It just does not affect the signal amplitude you hear on the output as a volume control. It controls the signal to the part of the circuit that changes the resonance frequency of the filtering. There is one small detail that many may have missed here. (I did for a long time.) The typical toe down (treble, bright) position of the wah is volume all the way down and the heel down (bass, dark) position is full volume.

If you want to do like me and convert a volume pedal in to a wah, it might work somewhat smoothly. The Ernie Ball VPJR I'm experimenting with is actually pretty good like this... There is however one caveat. It works backwards from what you normally expect from a wah. Heel down will be bright sound and toe down will be dark. You can reverse the connection on the pot but then all the change in the sound will be bunched up to the top end of the sweep, worse than with a linear pot. I believe this would be the case with any volume pedal if it is any good as a volume pedal to start with. There's a reason why most volume controls are logarithimic. I think I can solve this mechanically by changing the direction the pedal moves the pot, but I want to experiment with something else first.

I want to try modifying the taper of a linear pot with a resistor, like in the Better Volume (and Balance) Controls article I mentioned above.

The problem here is, when the pot is at "full volume" the combination of Rw and Rl looks like about 13 kohm pot. This loads down the the gain stage formed by Q1 and there will be less signal available for Q2. This in turn makes the signal smaller at the "ground end" of the Cf. The effect for the whole circuit is that it the sweep is limited in the dark end.

On my breadboarded version with just the 100 kohm pot the the resonance frequency is for the dark end is about 410 Hz. If I add the 15 kohm resistor from ground to the wiper of the pot, the lowest resonance frequency of the sweep is about 570 Hz. The bright end of the sweep stays the same, about 2 kHz. The top end of the sweep doesn't seem to jump quite as abruptly, but I need to get the range to stay the same to really know.

I also need to get a linear 100 kohm pot that fits my enclosure to see how it feels under foot control, so I can experiment how well this really works. currently I'm experimenting with an old pot that I hapened to have and I'm just twisting it by hand.

Another thing it affects is the output volume on that end of the sweep. The output comes through Cout from of Q1, so there is less signal there too as well, if it is loaded down.

I could just go up in pot value, and scale the tapering resistor accordingly. However, I'm interested in another kind of solution. Let's try buffering the Q1 gainstage with an emitter follower.

Originally I tried this with 0.22 µF Cout cap. It almost worked, but when I turned the pot to the low end of the range it started oscillating at about 420 Hz. If I pulled out the 15 kohm "tapering resistor", it didn't oscillate and worked completely normally. The resonance frequency at the dark end of the sweep actually went to about 380 Hz. That's a bit lower than without the buffer.

I was really puzzled about the oscillation at first. It took me a while to realize that the Rl is also affecting the high pass filter formed by Cout and Rw. With only the 100 kohm pot and 0.22 µF the low frequency cutoff is 7.2 Hz. I'd say that's low enough. As I pointed out earlier, with the 15 kohm Rl in the circuit the combined resistance in the heel down postiion becomes about 13 kohm. With the same 0.22 µf Cout the low frequency cutoff rises to about 56 Hz. That itself shouldn't be a problem with guitar and even with bass it would probably be barely noticeable.

However, any simple filter, like the coupling cap with the pot, will affect the phase of the signal in the passband too. The phase shift increases as you get closer to the cutoff. My theory was that the phase shift got too big. What would normally be negative feedback through Cf and Rfb stopped being negative feedback at the lowest frequencies. That combined with moving the resonance of the filter to the lowest frequencies and maybe there will be enough positive feedback to make it oscillate. I quess this didn't happen without the Q5 buffer, because the loading created by Rl attenuated the signal and also because the Resonance didn't get as low.

When I got to testing this theory, the first bigger cap that I found was 10 µF electrolytic. I used that as the Cout, and the oscillation was gone. I also tested with 1 µF plastic cap similar to the 0.22 µF I was using earlier and that also worked just fine without any oscillation. So... I quess that confirms my theory.

The  extra benefit of this arrangement is that it that the added buffer also buffers the output. Regular two transistor wah tends to loose some of it's range, if you put it in front of something that has a low impedance input, like some  fuzzes do. I was going to add an output buffer to the wah I'm building after these experiments, but it might not be necessary.

When turning the pot by hand, the sweep seems better than linear, but I think there is still a bit too much jump in the high end. The garphs on the Rod Elliot's articles kind of show what's going on there, but I made my own crude calculator for the tapering arrangement in google sheets.

There is three different curves to represent 10 kohm, 15 kohm amd 22 kohm load resistors on the graph. Just the linear pot would be a straight line from the bottom left corner to the top right.

At first it might not make much sense, but what we have to remember, that the graphs are for signal amplitude (volume) change. Higher signal amplitude going to Q2 pushes the resonance frequency down, so the low end of the graph is the treble end of the wah sweep and the high end of the graph is the bassy end of the wah sweep. I could turn the graph over, but I think this represents better how the pot works in the circuit. (Kind of... Maybe.) The important thing here is to see how the different load resistors affect the signal amplitude in various parts of the pot rotation.

The jump I feel in the high end of the pot rotation is probably because the slight "wrong way" curve we can see in the lower left corner of the graph. At "zero" both ends of the load resistor are connected together and it doesn't have any effect. When the pot is turned it starts affecting the divider gradually.

I have an idea how to make it work better though. If I don't use the full rotation and make the other part's of the circuit produce wider sweep range, I should be able to get the range of the original circuit without the bump (or at least as with less bump). Many wah pedals don't use the full range of the pot anyway. The case I'm going to use uses practically full range of the pot, but I can solve that with a resistor in series with one of the pot pins. I just need to find a way to make the wah sweep range wider. I was going to experiment with that anyway and I already have done some experiments for that. I just need to continue with those experiments and see if I can find a good balance between enough range so I can give up a bit of it and end up with enough for my liking.

So next I will experiment with getting the sweep range wider. That will mostly be adding more gain to the circuit and finding out how much can I add before it oscillates. I also need to get some linear pots that fit my volume pedal enclosure so I can try how the tapering mods actually work in use.