Variable Flux Inductor Kay Wah

[mac]

The Kay Wah used a movable ferrite core inductor to change the resonant peaks.
Since a typical core has 200x more permeability than air you can get a wide range of frequencies.
And the mechanism is simple and trouble free.
Variable inductors have some issues, like being a nice antenna, microphonic, big and $$$... but a wah pot is also expensive

If you like the idea of a variable inductor but you are too lazy to wire your own, there is another way of getting a variable inductor.
If you connect the coil to a pot and a second power supply, a dc current will flow across the coil changing the flux not linearly when the current is about to saturate the core.
The magnetic amplifier uses the same principle IIRC.
Since you are connecting a resistor and battery, their resistance has to be bigger than the coil dc resistance in all the desired freq range.
And an inductor with a small saturation current is better for the same reason.

I ran a simulation of a modified Kay Wah circuit, using a non linear flux = L*Isat*tanh(x/Isat).
This is the schematic,



And this is the plot compared to a Cry Baby style wah,



As you can see, I used a 50v supply and a very low Isat inductor to make the battery+pot resistance big.
In the case of the Kay Wah the inductor is grounded and its value is about 3H.
Does a commercial inductor of such characteristics exist?
I don't know, but the primary of dichroic lamp 220v - 12v transformers seem to be a starting point to experiment because they normally have low dc resistance, and work near saturation, although at AC and 50-60hz... saturation current could be bigger than needed...

I have this circuit in the breadboard, and while I'm not getting a high Q wha sound, the frequency does change in the desired range when turning the pot up and down. Second DC supply is 9v, and a minimum resistance of 1k is needed to reach 2kHz or so --> higher Isat than needed, most likely.
I have to increase the second battery.

Cons: changing dc current makes noise mainly at very low pot resistance, high voltage supply for the coil, getting a coil with the properties noted above.

I thought of another way of varying the flux and thus the inductance without using a second supply.
If you use the secondary of a 220v:100v to 24v transfomer as the inductor, and put a suitable floating pot acorss the primary, you can change the resonant frequency by varying the pot. Loading the primary forces the secondary.
Unfortunately LTSpice does not allow the use of non linear mutual inductors to show you some freq distributions.
I had some decent result with a 220v:47v+9v and a 220v:12v+9v, secondaries in series to have more inductance. But the pots and capacitors needed were different.

In both cases, the inductor near saturation colorizes the sound a bit in a way I like, others might not.

The Kay Wah does not look like a good wah design, but it is useful for experimenting.

Two alternative wah mechanism. Just for fun

mac

[jatalahd]

So if I understood this correctly you are using a magnetic core inductor, which has a certain hysteresis curve where the flux saturates at the top and at the bottom of the curve. Then the "slope" of the inductance equation L = d(flux)/d(current) will NOT be constant near the saturation and changing the current at this saturation region changes the inductance L ... This is a nice idea.

I was just wondering what happens if you change the polarity of the DC source V3 ? When looking at the simulation, the curves seem to be approximately a mirror image on the frequency axis, when comparing to the Cry Baby curves. Can one get "negative" inductance when changing the polarity of V3 ?

[mac]

So if I understood this correctly you are using a magnetic core inductor, which has a certain hysteresis curve where the flux saturates at the top and at the bottom of the curve. Then the "slope" of the inductance equation L = d(flux)/d(current) will NOT be constant near the saturation and changing the current at this saturation region changes the inductance L ... This is a nice idea.

Exactly.
K=u/uo is not constant,


Note that in my simulation I *supposed* the flux as F=Lo*Isat*tanh(i/Isat) that leads to L=Lo*[1-tanh^2(i/Isat)].
As i increases L gets smaller because tahn(x)-->1 as x-->inf.
In this case changing the polarity does nothing.

I don't remember if I changed polarity in real life... I bet I did it without changes... not sure.

If you try this idea in another circuit the pot and the second battery should float.

mac

[David]

Marcelo, I had a Kay when I was a kid.  It actually did not seem like a half-bad wah.  I'd love to hear a sound clip of this.

[PRR]

Post-1999, you may use a "Constant current source" (Voltage plus a current limiter) to DC-smack the winding without loading the audio.

I remember a LOT of trouble with moving cores. But they were cheap paper-bobbin coils in low-price radios.

Pre-1999, low-price arc-welders were invariably moving-core. Picture an E-I core. With the I slammed tight to the E, the inductance is maximum. When you pull the I away from the E, inductance drops. This would seem to be ideal for a foot-lever contraption. E-I cores are common in power transformers. If wound for AC use, the lams are interleaved and you would have to pull them all out to re-assemble as E and I. However chokes rated for DC current are already E-I construction, held together by the frame.

All these open-core plans are prone to attract all the hum in the room.

[jatalahd]

In this case changing the polarity does nothing.

Ok, good. I was just puzzled   what causes the frequency response to be mirrored. But now I understood it  , had to browse back to the basics again ... In a parallel resonance circuit, the inductance controls the gain on the left side of the resonance and the capacitance controls the right side. Since now the capacitance is kept constant, the shape of the roll-off curve after the resonance stays constant and vice versa.

If your prototype works the same way as the simulation indicates, it should get close to the Kay Wah sound. Indeed it would be nice to get an audio sample.  In this case  at least the inductor's internal resistance affects Q. Smaller resistance gives you higher Q. Of course if the internal resistance of the voltage source and R10 are seen parallel to L, then that affects the Q; lower parallel R --> lower Q.

[mac]

jatalahd,
If you change the Flux=Lo*x*tanh(x/Isat) you get an inductance that is not even, and the polarity does matter.
Although I doubt such a flux exists, some non-even behaviour can occur, so in real life one should change polarity to see what happpens.

About the mirrored response, c10 and the inductance filter the highs and lows. They seem to work ok in the range where the transistor output impedance does not get in the way.
I also had to tweak c6 until I got decent curves.

As I noted before, a proto is hard to get because of the required coil attributes. And since this is grounded, expect a lot of crackle from the pot and battery. BUt I'll post some sound clips.

Paul,
I'm testing wah circuits and mechanisms. DC core saturation is one of them.
I have some moving core designs that faithfully mimic a cry bay.
With proper shielding there is no hum.
And the movement of the core makes no noise.
Old variable air caps wah on the other hand makes for a good seismic sensor

David,
I remember that the Kay Wah has no constant volume, the peaks at the high side drop noticeably unlike a Cry Baby.
The Colorsound Inductorless Wah has a bump half way, but the difference is not as pronounced as in the Kay Wah, and some people like it.
In both cases a simple cap tweak corrects to some extent this issues.