Experimenting With a Wah Circuit

[Elektrojänis]

I did a little experiment on a Vox/Dunlop style wah circuit and documented it on my page: https://sites.google.com/view/elektronijaenis-explains/home/wah-pedal-experiments

I was actually going to do some more experiments and then do a PCB layout that can accomodate several variations based on how you populate it, but with my pace it would take years.

Another thing is that my first experiment was somewhat promising, but it proved to a bit too noisy. Maybe some of you who know a lot more about electronics can shed some light on these things... And even if you don't, maybe you still have ideas.

The basic idea was to rearrange the basic circuit a bit (also added one cap):



And after that rearrangement one end of the inductor is connected to ground and can be replaced with a gyrator circuit:



However, with the gyrator it is considerably noisier. There is a sound sample comparing the two on my page and some more explanations how I think the circuit works etc.

Anyway... I thought that better throw this idea out there or here than keep it to myself. Maybe someone else can get something out of it. This also might be something that has been already done so feel free to point that out too.

[Elektrojänis]

This didn't really spark any interest, but I quess I wont do much harm if I let you all know I have experimented a bit more with replacing the wah inductor with a gyrator.

I mainly experimented with lowering the noise level and I think I managed to get it to reasonably quiet. The changes to make it quieter affect the tone a bit though. There are some soundsamples on the page (as embedded youtube videos), but don't expect any fancy playing.

I added it as a second part to the same google sites page: https://sites.google.com/view/elektronijaenis-explains/home/wah-pedal-experiments (Scroll down to find the new part.)

I still have a few other things in my mind I want to experiment, beside the gyrator thing. With my current speed of doing these things, I might add Part 3 to the article before christmas.

[jorg777]

Interesting work!

I've heard that they use special "mojo" inductors which saturate in just the right way as DC current flows through them; the current (and saturation) vary with pedal position to produce pleasing distortion.  I've also heard that's nonsense.  :-)

[Elektrojänis]

I've heard that they use special "mojo" inductors which saturate in just the right way as DC current flows through them; the current (and saturation) vary with pedal position to produce pleasing distortion.  I've also heard that's nonsense.  :-)

The starting point for these experiments was The Technology of Wah Pedals by R.G. Keen. That article has a part about measuring one onf the "mojo" inductors.

One of the things I have in mind is about those too, but I have to test some stuff first and see how it goes.

[aron]

I have a disco gig coming up LOL. All I did was replace my inductor with a yellow Fasel, adjust the gearing and called it a day! I haven't played a wah in years. I should have made it true bypass but it was a rush job!
Good luck with your work on the wah circuit.

[Rob Strand]

This didn't really spark any interest, but I quess I wont do much harm if I let you all know I have experimented a bit more with replacing the wah inductor with a gyrator.

It's not that it lacks interest but some people have already tried it, and yes, you do get more noise.   FWIW, the way you did it looks OK.

The way to removed some noise is to make you Rg2 as low as you can.   That means pushing up the value of Rq.   However, if you go too high then Rq will affect the biasing so you need to tweak Rb2.   If you consider a variable Re1, then increasing Rq could produce different biasing as Re1 is varied.   In the original the inductors shorts Rq altogether.

Interesting work!

I've heard that they use special "mojo" inductors which saturate in just the right way as DC current flows through them; the current (and saturation) vary with pedal position to produce pleasing distortion.  I've also heard that's nonsense.  :-)

Some of the magic comes from choosing a good DC resistance, and inductance for some people.   The inductors have AC losses, which depend on frequency, which makes the actual circuit see a lower RQ than the circuit values.    The older ferrites are likely to be more lossy than later ones and that might explain why RQ was 100k on early units and 33k on later units, sort of getting a similar ball-park of the final effective RQ.    The losses in ferrites are complicated to model, some losses bundle with RDC and some with RQ.   A better (approximate) model could end-up with the circuit seeing an RDC higher than the measured DC resistance at low AC frequencies.    There is also some dependency of the losses on the drive level. (Reference: If you want to read-up more see Snelling's book on Ferrites and look up the Rayleigh hysteresis model and the Hysteresis constant.   These are approximate models.)

For those toroid cores even the inductance value is up for grabs because it can vary quite a bit with drive level (Reference: AC permeability).

[Elektrojänis]

It's not that it lacks interest but some people have already tried it, and yes, you do get more noise.   

Somehow I've missed all of that, but maybe I haven't really done proper searches in the last 10 years.  And I know I have large gaps when I haven't been following this forum.

The way to removed some noise is to make you Rg2 as low as you can.

What I found with my crude measurements, the Rg2 didn't really do much to the noise level. The differences were so small that it could easily be explained by lower Q of the filter. I only measured the noise through the full wah circuit and didn't change anything else on the circuit.

What I found is that the value of Rg1 (feedback resistor on the gyrator) affects the noise level much more. Additionally, making it higher in value lowers the noise. That seemed backwards to me, because usually lower resistances mean lower noise.

That means pushing up the value of Rq.   However, if you go too high then Rq will affect the biasing so you need to tweak Rb2.   If you consider a variable Re1, then increasing Rq could produce different biasing as Re1 is varied.   In the original the inductors shorts Rq altogether.

For comparing with the inductor version I kept everything I could in the circuit the same. For optimal biasing something would have to change. Removing the inductor and leaving the bias feedback to Rq made asurprisingly small change on the voltage on Q1 collector.

If I would have pursued the gyrator Idea further I might have made a bit different bias arrangement. I'm not sure if it would have worked because I havent really learned all the biasing calculations and that sort of stuff properly.

Some of the magic comes from choosing a good DC resistance, and inductance for some people.

Thats a good one. I've made some preliminary testing about it after I found out about it on some earlier wah threads here (this ans this). I will probably put a series resistor for the inductor in my actual wah build, that I can change to get the low end of the sweep to sound how I want.

The inductors have AC losses, which depend on frequency, which makes the actual circuit see a lower RQ than the circuit values.    The older ferrites are likely to be more lossy than later ones and that might explain why RQ was 100k on early units and 33k on later units, sort of getting a similar ball-park of the final effective RQ.    The losses in ferrites are complicated to model, some losses bundle with RDC and some with RQ.

Seems it can get really complex, but how much will the finer details of the losses affect the sound in ways that couldn't be reproduced by tuning the Rq and resistor added in series to the inductor?

For those toroid cores even the inductance value is up for grabs because it can vary quite a bit with drive level (Reference: AC permeability).

Is this related to the saturation effects described in The Technology of Wah Pedals by R.G Keen?

[Rob Strand]

Somehow I've missed all of that, but maybe I haven't really done proper searches in the last 10 years.  And I know I have large gaps when I haven't been following this forum.

It's probably 20 years ago now since looked at in detail and posted something  .

What I found with my crude measurements, the Rg2 didn't really do much to the noise level. The differences were so small that it could easily be explained by lower Q of the filter. I only measured the noise through the full wah circuit and didn't change anything else on the circuit.

What I found is that the value of Rg1 (feedback resistor on the gyrator) affects the noise level much more. Additionally, making it higher in value lowers the noise. That seemed backwards to me, because usually lower resistances mean lower noise.

I think you are right about Rg2.   It adds noise but only when it's a high value.    The opamp noise is amplified by the circuit by a crazy amount.   I have some recollection of trying a NE5534/NE5532 for low noise voltage.   Those have a high input current noise which might be why I was thinking Rg2 has an effect (as high Rg2 causes the noise current to have an effect).   The noise was much much higher than the inductor circuit.

For comparing with the inductor version I kept everything I could in the circuit the same. For optimal biasing something would have to change. Removing the inductor and leaving the bias feedback to Rq made asurprisingly small change on the voltage on Q1 collector.

For a fixed circuit it's OK but I remember it having some effect when RE was varied - I could have been pedantic at the time.

If I would have pursued the gyrator Idea further I might have made a bit different bias arrangement. I'm not sure if it would have worked because I havent really learned all the biasing calculations and that sort of stuff properly.

You can bend things around but it's not bad like it is.     What you have is pretty good.

Thats a good one. I've made some preliminary testing about it after I found out about it on some earlier wah threads here (this ans this). I will probably put a series resistor for the inductor in my actual wah build, that I can change to get the low end of the sweep to sound how I want.

Yes, those old threads summaries things quite well.

Seems it can get really complex, but how much will the finer details of the losses affect the sound in ways that couldn't be reproduced by tuning the Rq and resistor added in series to the inductor?

It definitely gets complicated.  I'm across all the theory and calculations and I've even come up with ways of matching inductors losses.   Despite all that I haven't come up with anything more than tweaking Rq and the series resistance!   It's does a good job anyway.

Is this related to the saturation effects described in The Technology of Wah Pedals by R.G Keen?

It is related.  I remember analyzing the smaller P14/8 pot cores.   With strong signals you could push them.   I can't remember how close to saturation it was but they were pushed more than you would normally do that's for sure.   Interestingly a lot of the boutique inductors use larger P18/11 pot cores which have much less chance of saturation.

FWIW, you can use a buffered gyrator to make a more ideal inductor.   They problem with those is two opamps and it makes the noise worse.  The buffered gyrator lets you use smaller gyrator resistances.

[PRR]

...Rg2 didn't really do much to the noise level. .....the value of Rg1 (feedback resistor on the gyrator)
That seemed backwards to me, because usually lower resistances mean lower noise.

https://sites.google.com/view/elektronijaenis-explains/home/wah-pedal-experiments

Rg1 is in series with the signal. Smaller is quieter.

Rg2 is in shunt with the signal. Larger is quieter.

[Elektrojänis]

You can bend things around but it's not bad like it is.     What you have is pretty good.

At some point I might try separating the feedback for DC-bias from the filtering altogether. To me it seems like it would only be a few resistors and more... Or actually with the gyrator the Rg2 would work as Rq and I could also get rid of Rb1, Rb2 and Cbp if the DC feedback would be its own network flom Q1 collector to base. Maybe... I'm mainly just copying stuff I've seen elsewhere.

Not that there would be a lot to gain from the experiment, but I might just learn something again.

It is related.  I remember analyzing the smaller P14/8 pot cores.   With strong signals you could push them.   I can't remember how close to saturation it was but they were pushed more than you would normally do that's for sure.   Interestingly a lot of the boutique inductors use larger P18/11 pot cores which have much less chance of saturation.

I did some other preliminary experints and measurements with a real inductor. It's a current production yellow Fasel with DC resistance of 15 ohms. I measured how much AC voltage across the inductor I could get with a test signal and it was not much. I didn't really write it up anywhere, but I think it was only about one tenth or maybe one fifth of the input signal level. Makes me wonder how how small would the core have to be to get any significant saturation at guitar signal levels?

FWIW, you can use a buffered gyrator to make a more ideal inductor.   They problem with those is two opamps and it makes the noise worse.  The buffered gyrator lets you use smaller gyrator resistances.

Something like this? https://sound-au.com/articles/gyrator-filters.htm#s13

Rg1 is in series with the signal. Smaller is quieter.

Rg2 is in shunt with the signal. Larger is quieter.

Odd... That's exactly opposite what I measured. Larger Rg1 was noticeably quieter. Smaller Rg2 was quieter, but just a little. So little that it could have been just the effect it has on the total Q of the filter.

[PRR]

opposite what I measured.

Maybe I'm looking at a different schematic.

[Rob Strand]

At some point I might try separating the feedback for DC-bias from the filtering altogether. To me it seems like it would only be a few resistors and more... Or actually with the gyrator the Rg2 would work as Rq and I could also get rid of Rb1, Rb2 and Cbp if the DC feedback would be its own network flom Q1 collector to base. Maybe... I'm mainly just copying stuff I've seen elsewhere

Yes, there's plenty of options.  I guess the main constraint is not messing with the sound of the original.  It's tempting to bias the gyrator with the same bias supply but with an opamp you really want to bias it at half supply to avoid *any* form of overload.

I did some other preliminary experints and measurements with a real inductor. It's a current production yellow Fasel with DC resistance of 15 ohms. I measured how much AC voltage across the inductor I could get with a test signal and it was not much. I didn't really write it up anywhere, but I think it was only about one tenth or maybe one fifth of the input signal level. Makes me wonder how how small would the core have to be to get any significant saturation at guitar signal levels?

The thing that really cause saturation is the current.   The inductor current is V/XL so as frequency drops the current increases for the same applied voltage.    IIRC those small wah toroids saturate with only a few mA of current.   When you design inductors for audio typically the flux density (B) is kept *much* lower than the saturation limit in order to avoid distortion.   For an ungapped core the distortion starts at maybe 10mT but saturation is more like 300mT so there's a grey area of a factor of 30 as to whether it affects the sound.   To be honest, it wasn't enough to make the worried about it since there's plenty of good sounding inductors which are well away from saturation (or distortion).

Something like this? https://sound-au.com/articles/gyrator-filters.htm#s13

Yes that one.   You can make Rg2 small and Cg1 large to keep noise down.  The trouble is the opamp noise dominates and putting two opamps in there makes things worse!

There's another variation where the opamp has a small amount of gain instead of being a buffer and that allows the gyrator to model the inductor more ideally.   It's not widely known at all.

Odd... That's exactly opposite what I measured. Larger Rg1 was noticeably quieter. Smaller Rg2 was quieter, but just a little. So little that it could have been just the effect it has on the total Q of the filter.

That's what I would have expected.

opposite what I measured.

Maybe I'm looking at a different schematic.

There's two aspects to the noise:   There's a big peak of narrow band noise then there's the general noise level on each side of the peak.   The peak depends on the wah pedal position.   When Rg1 is small the peak is higher.

We can just change Rg1.   There's a good value for Rg1 and we should stick to it.   It's not something we are free to play with if we want the wah to behave like the wah's we know.

What we can do as an experiment is set Rg1 to a low value, keeping L the same, then put another resistor in series with Cbp2 so Rg1 + Radded is the desired wah resistance.  I'm pretty sure it will make noise worse and also affect the inductor Q at high frequencies.   I remember playing with Sziklai buffers, like that site PRR linked, with the aim of producing a lower noise buffer.    When Rg1 is small it becomes difficult to get a high Q inductor (Offhand I can't remember why) but what happens is the wah peaks are reduced.

[Elektrojänis]

I did a few forum searches and now I'm finding threads about replacing the wah inductor with a gyrator. And some of them are quite old...

Oh well... At least I think I have learned quite a bit while experimenting.

Yes, there's plenty of options.  I guess the main constraint is not messing with the sound of the original.

At some point I will also be interested to changing the sound in some way, but first I wanted to see how close i could get.

It's tempting to bias the gyrator with the same bias supply but with an opamp you really want to bias it at half supply to avoid *any* form of overload.

In my experiments I actually had half supply bias already for the input and output buffers I had in there and I used that for the gyrator too.

I'm a bit tempted to find out what overloading the gyrator would sound like... I actually have an experiment in mind for that. It might sound disgusting, but then again some people like Boss HM-2 with all the knobs on full... I wouldn't be surprised if overloading the gyrators in it's eq circuit would be part of that sound.

The thing that really cause saturation is the current.   The inductor current is V/XL so as frequency drops the current increases for the same applied voltage.

Ah... I didnt really think of that enough in in terms of current, but it makes sense now. The voltage doesn't need to be large when the impedance is low.

To be honest, it wasn't enough to make the worried about it since there's plenty of good sounding inductors which are well away from saturation (or distortion).

I still think it might be interesting to hear teh effects of that inductor saturation if it could be isolated from everything else and maybe even exagerated a bit. Making that happen is just probably something far too advanced for me.

Something like this? https://sound-au.com/articles/gyrator-filters.htm#s13

Yes that one.   You can make Rg2 small and Cg1 large to keep noise down.  The trouble is the opamp noise dominates and putting two opamps in there makes things worse!

I might try that at some point. I quess that one could be done with transistor buffers or Sziklai buffers. The single buffer gyrator was not much quiter with a Sziklai buffer when compared to NE5532. I think I might have one of more hifi Texas Instruments opamps lying somewhere, so I also might try that. I like the idea of using relatively common, easy te get and cheap parts though, but it might still be worth the experiment to see if it makes a difference.

I remember playing with Sziklai buffers, like that site PRR linked, with the aim of producing a lower noise buffer.

You mean this one: https://sites.google.com/view/elektronijaenis-explains/home/wah-pedal-experiments ?

That's the my page i linked on the original post. I use it to document these experiments in more complete form. On the posts I have just written a smaller exerpts of the interesting bits.

When Rg1 is small it becomes difficult to get a high Q inductor (Offhand I can't remember why) but what happens is the wah peaks are reduced.

That's very interesting. I didn't really hear much reduction in the peaks whith the smallest Rg1's I tried. I didn't go lower than 33 ohms though. Could it be that if you get too low the opamp (or whatever you use for the buffer) just can't drive that low impedance properly?

[Rob Strand]

Oh well... At least I think I have learned quite a bit while experimenting.

If you don't look you won't find anything.   You might be the one who finds the key step.

To me there is always going to be a problem with the common wah circuit amplifying the noise from the gyrator.

I always liked this one.  It emulates the overall response without focusing on the gyrator,
https://cackleberrypines.net/transmogrifox/TransmogriNotes/skwah/index.html

I still think it might be interesting to hear teh effects of that inductor saturation if it could be isolated from everything else and maybe even exagerated a bit. Making that happen is just probably something far too advanced for me.

It's definitely not a simple thing.   One approach would be to design a 'crappy' inductor which does saturate.   The trick would be to have an inductor with the right inductance *and* DC resistance *and* saturate at low currents.

I might try that at some point. I quess that one could be done with transistor buffers or Sziklai buffers. The single buffer gyrator was not much quiter with a Sziklai buffer when compared to NE5532. I think I might have one of more hifi Texas Instruments opamps lying somewhere, so I also might try that. I like the idea of using relatively common, easy te get and cheap parts though, but it might still be worth the experiment to see if it makes a difference.

The BJT inputs on the Sziklai end up doing the same damage as the BJT inputs on the NE5532.   My ultimate goal was a JFET version of the Sziklai type buffer.

I had a look at why the wah peaks drop with the Sziklai buffer.   It is because the gain of the Sziklai buffer isn't *exactly* unity.  To some degree it is possible to reduce the smaller gyrator resistance to compensate.  However, some care is require to make sure all builds of the buffer worked with that tweak.

You mean this one: https://sites.google.com/view/elektronijaenis-explains/home/wah-pedal-experiments ?

That's the my page i linked on the original post. I use it to document these experiments in more complete form. On the posts I have just written a smaller exerpts of the interesting bits.

Yes that one.  I thought the slightly different spelling of your name was too close

That's very interesting. I didn't really hear much reduction in the peaks whith the smallest Rg1's I tried. I didn't go lower than 33 ohms though. Could it be that if you get too low the opamp (or whatever you use for the buffer) just can't drive that low impedance properly?

So it wasn't impedance but the gain.   Lowering the resistor can help to some degree.

[Elektrojänis]

If you don't look you won't find anything.   You might be the one who finds the key step.

True. However, it's unlikely that I'd discover something totally new. I just try to have fun and keep it interesting to myself and if I find out something interesting, it's a lucky coincidence. More like a byproduct.

To me there is always going to be a problem with the common wah circuit amplifying the noise from the gyrator.

I'm starting to believe that's the case for me too. It might be acceptable when driving a clean amp and no overdrive/distortion/fuzz after the wah, but with overdrive it will just get too noisy.

I always liked this one.  It emulates the overall response without focusing on the gyrator,
https://cackleberrypines.net/transmogrifox/TransmogriNotes/skwah/index.html

That's really cool! It seems really tweakable in ways that the classic circuit isn't. Or maybe it is, but it's harder to separate the parameters with the classic circuit. It even has input buffering built in and it might be less sensitive to loading of the output than the classic.

It's definitely not a simple thing.   One approach would be to design a 'crappy' inductor which does saturate.   The trick would be to have an inductor with the right inductance *and* DC resistance *and* saturate at low currents.

I quess the core would have to be really small, but that would limit the inductance. And that means it needs more turns in the winding. -> The wire needs to be thicker. A lot of work to make it crappy.  

The BJT inputs on the Sziklai end up doing the same damage as the BJT inputs on the NE5532.

Well TL072 was noisier than NE5532. Then again, maybe TL072 is not really low noise part anyway.

I had a look at why the wah peaks drop with the Sziklai buffer.   It is because the gain of the Sziklai buffer isn't *exactly* unity.

That makes a lot of sense since the single transistor buffer was worse in that regard.

I thought the slightly different spelling of your name was too close

Haha... That's my bad. It's an old nick (or handle/login) I've been using for decades. It started as a name for a music project. At some point I decided to change it just a bit, but I only implemented it half way... I have used Elektronijänis new things, but haven't really bothered to change it anywhere. Now I really dont even see the difference anymore.

Anyway... At this point I must thank you very much for good info and pointers!

[Rob Strand]

I quess the core would have to be really small, but that would limit the inductance. And that means it needs more turns in the winding. -> The wire needs to be thicker. A lot of work to make it crappy

I'm not even sure it can be done.   The 30 ohm target limits how small the wire is, that limits the turns, so in order to get enough inductance you need a very high mu core.   Maybe the mu ends up being too high to be realistic.

There's other materials for cores.   Square loop ferrites and core for saturable inductors.  You might not get to saturation but it might misbehave more than common ferrites.   I'm not too enthusiastic about winding a 500mH inductor on a toroid.

Another angle is to make a 30V or more wah pedal, drive higher level signals in, and attenuate the output for the same overall gain.  That way the normal inductor would be driven a lot harder.   Of course things never work out to be 100% apples-to-apples equivalent.

Haha... That's my bad. It's an old nick (or handle/login) I've been using for decades. It started as a name for a music project. At some point I decided to change it just a bit, but I only implemented it half way... I have used Elektronijänis new things, but haven't really bothered to change it anywhere. Now I really dont even see the difference anymore.

Anyway... At this point I must thank you very much for good info and pointers!

FWIW, I thought you did a great job and took it about as far as you can.  I didn't get any further when I looked at it and I don't have anything to add even after all these years!

[Fancy Lime]

If you are hell bent on saturating an inductor with a magnetically soft core like ferrite, you can also manipulate the magnetic side of things instead of the electric side. Core saturation means that the microscopic magnetic dipoles in the core material cannot absorb more flux by orienting themselves along the applied field better, because they are already aligned as well as they can be. If you allpy a static external magnetic field to the core, e.g. by placing a permanent magnet near the inductor, then that creates a kind of pre-saturation. This shifts the saturation threshold but not in a symmetrical way, like increased current would. Depending on the core geometry, core material, and strength and placement of the magnet, it may be possible to get an audible and interesting asymmetric saturation effect out of this. Add a screw or something, to make the magnet distance and thus the core pre-saturation adjustable and you got yourself the next major mojo innovation in wah technology (sorry, I suck at marketing lingo).
Another, less unusual but also less fun way of achieving a similar effect would be to just apply a constant DC current acoss the inductor. Would also shift the saturation threshold closer to the signal level on one side of the swing.

Andy