True "variable inductor" wah...

[jasperoosthoek]

That word is actually Miller. All classic wahs are based on sending an inverted signal to the other side of the cap thereby multiplying the caps value.

All I do is use a variable cap instead of a variable gain with a pot. And I use bigger coils and smaller caps.

Right now I don't have the time to try that idea. I've been receiving packages all week as I'm building a new guitar . Also my Firefly has been finished halfway...

[gmoon]

Naw, I'm familiar with the Miller effect, from mucking with tube amps. It's another term.

Maybe I'll recall it. Or maybe I'm mistaken. Anyway, have fun!

[SISKO]

Give it a go, jasperoosthoek.

Seems to me there's a one-word name for that capacitance circuit, but typically, it escapes me...

Gyrator?

[gmoon]

Gyrator?

THAT'S the one... thanks.

(and I was a little off the mark, as it's simulating inductance rather than multiplying capacitance...)

[PRR]

> the wider the bandwidth (lower the Q), or the L/C ratio. So there's probably a realistic range.

(as you know) You adjust this with the R.

In this plan, there's two. The amplifier (transistor) has an output impedance in series with (part of) the cap (and thus the tank). The coil *always* has series resistance. Lower is higher Q. The resistor parallel to the coil also matters,higher is higher Q.

> So there's probably a realistic range.

There's a "practical" range of impedances. The 0.55H 0.33uFd team is nominal 1.3K. For significant Q we want series resistance of a few hundred ohms and shunt resistance of many K. Coils this big are hard to get. A 0.5 milliHenry coil is easier to find. Load can be a few dozen ohms (easy) but source should be under One Ohm (oof!). Even a LM386 will have trouble driving one ohm impedance; 1-transistor stages won't do the job in guitar systems.

> Too high a "Q" just doesn't sound vocal, and too low isn't a "wah."

I don't know wah. In general audio manipulation, Q of 0.7 to 1.3 will give a significant broad boost. Anything higher, in general audio, sounds "ringy" and "peaky". Obviously a wah works in the "ringy zone". But if Q were far over 10, it would peak an overtone then there would be a "gap" until it crosses the next overtone. So the "ideal Q" depends on how dense the overtones are. We could postulate and interpolate. It may be easier to just try. Easier yet to copy Q ratios from existing wahs.

And there's another conflict with reality. Because of coil winding resistance, even with closed iron cores, it is hard to get Q of a small coil over unity at the bottom of the audio band. In the midrange Qs of a few dozen are not too hard... but a variable inductor usually won't be a _closed_ core so Q will be lower. So it may be real difficult to end up at "too high" a Q.

Using Hammond/Wurlie coils is brilliant. They are audio-scale values, and VERY well designed. An organ has a dozen, and the organ-harvesters usually throw away the tone generators.

> I wish the DC resistance for both were lower

The Q was "as high as reasonable" for good oscillator stability. The old guys knew what they were doing. To lower R, you could use larger wire, but the larger build means lower L, so you need more turns to get your L, and that's more R. Q probably scales up with size, but probably very slowly: square-root or even cube-root of linear dimension. If it is ^0.33 then double Q means 8 times longer and 500 times as much costly stuff. We may have better ferrites, though I think most work is for better >5KHz performance (audio is a dead-end).

> How about tunable capacitors from old radios?

Generally 356pFd maximum. Which is about 200K at 2KHz, 2Megs at 200Hz. Your shunt resistance must be much higher, meaning many-Megs. I guess it could be done.

The high impedance does attract electrostatic interference. This same trick, though usually with a bigger (very custom) cap, was used in many sine-wave audio oscillators. The case must be VERY tight, the internal wiring very clean, or the output is contaminated with buzz and fluorescent lamp hash.

Also these tuning caps ARE getting rare, hard to find.

[jasperoosthoek]

Gyrator?

THAT'S the one... thanks.

(and I was a little off the mark, as it's simulating inductance rather than multiplying capacitance...)

Don't take my word for it:

Taken from the wah bible: http://www.geofex.com/article_folders/wahpedl/wahped.htm

No, it's not a Gyrator. I don't simulate an inductance with a feedback circuit. I use a true inductor. Of course you can use a gyrator anyway... As I said, It's a Miller effect capacitor with a parallel resistor. What I did is basically the same as R.G.s circuit, albeit with a single opamp. And I've used the Q resistor in the feedback loop. I miss one resistor though which can be easily implemented.

Generally 356pFd maximum. Which is about 200K at 2KHz, 2Megs at 200Hz. Your shunt resistance must be much higher, meaning many-Megs. I guess it could be done.

If you use the bare capacitance it will be quite difficult. I agree.

The high impedance does attract electrostatic interference. This same trick, though usually with a bigger (very custom) cap, was used in many sine-wave audio oscillators. The case must be VERY tight, the internal wiring very clean, or the output is contaminated with buzz and fluorescent lamp hash.

Maybe the biggest problem. That's one big antenna.

Also these tuning caps ARE getting rare, hard to find.

Search for "variable capacitor" on ebay. I get 339 hits, more than half of them could work. I only get 34 hits for "variable inductor". There is one 33H unit which might be suitable...

Very rare for someone trying make a living out of it. Quite abundant for the lone tinkerer who decides to build a single unit and very abundant compared to variable inductors. I can also buy them right now from a thrift stores. I probably threw away a couple of them some years ago. I know I shouldn't have but I try not to be a hoarder...

(Evil misunderstood villain voice)
You guys just keep pushing and pushin! With this circuit I will rule the world!
(/Evil misunderstood villain voice)

I might even try to breadboard it. Not to prove anyone right or wrong but I'm kind of interested myself if you can use it at audio frequencies. I have a 5H hammond choke that I can use as a passive gyrator simulation .

But I actually feel that if you Millerize the capacitor with a feedback loop you might as well go all the way and replace the passive inductor with a gyrator. Maybe even take a plastic tuning capacitor from an 80s radio.

[gmoon]

Don't take my word for it:

Oh I didn't. I've had a fair understanding of the Vox/thomas circuit for a while (thanks, R.G.), and looked up the online patents for multiplying Miller capacitance. The "Gyrator" comment was an aside (but one that belongs in the conversation).

Not getting enough encouragement? OK, third time   : Work on it. Please. It's an idea worth exploring.

Search for "variable capacitor" on ebay. I get 339 hits, more than half of them could work. I only get 34 hits for "variable inductor". There is one 33H unit which might be suitable...

Very rare for someone trying make a living out of it. Quite abundant for the lone tinkerer who decides to build a single unit and very abundant compared to variable inductors. I can also buy them right now from a thrift stores. I probably threw away a couple of them some years ago. I know I shouldn't have but I try not to be a hoarder...

OK, just playing "devil's advocate" here. Another member of the forum has identified a source for the same Wurlitzer inductors I'm using; and they're NOS, untouched. Apparently they have LOTS of them. Wurlitzer was a fairly large company, and must have had "authorized" regional repair centers, etc. Independent service centers bought up the parts, probably in the 1980's (or the authorized centers were already independents). I'll let the other forum member identify the source, if he wants to.

Plus there are WAY more inductors, chokes, transformers (and ferrite objects) currently being produced the world than those old butterfly caps, so a custom variable inductor, while expensive, could be ordered and delivered fairly quickly (if it weren't radically different in design).

Frankly, it's all beside the point, from my POV. I'm exploring an interesting design, not setting up for production. To paraphrase an old saying:

Want to make a small fortune building boutique guitar effects? Start with a large fortune...

[gmoon]

Thanks, PRR. As always, very pertinent and useful.

And there's another conflict with reality. Because of coil winding resistance, even with closed iron cores, it is hard to get Q of a small coil over unity at the bottom of the audio band. In the midrange Qs of a few dozen are not too hard... but a variable inductor usually won't be a _closed_ core so Q will be lower. So it may be real difficult to end up at "too high" a Q.

Right, as usual. I've disassembled and tested a few chokes. Removing the interleaved steel core reduces the inductance tremendously. "Interleaved" and "closed" being of equal importance--inserting the core leaves from one side doesn't replicate the flux of the closed core.

Sorta drives home concepts about output transformer construction, and why that "air gap" is needed for SE OTs.

The Q was "as high as reasonable" for good oscillator stability. The old guys knew what they were doing. To lower R, you could use larger wire, but the larger build means lower L, so you need more turns to get your L, and that's more R. Q probably scales up with size, but probably very slowly: square-root or even cube-root of linear dimension. If it is ^0.33 then double Q means 8 times longer and 500 times as much costly stuff. We may have better ferrites, though I think most work is for better >5KHz performance (audio is a dead-end).

Size matters  . That becomes apparent when examining old school inductors (like the spider coil). Makes perfect sense: dimension vs. area, etc.

Some funky stuff is at least worth thinking about: combinations of interleaved coils (ala spider), ferrite enclosures, and even movable, selectively interleaved closed cores are possible. Maybe Mu metal leaves as well between coils. Alright, these are a little much, compared to known, proven tech  ...

[jasperoosthoek]

Oh I didn't. I've had a fair understanding of the Vox/thomas circuit for a while (thanks, R.G.), and looked up the online patents for multiplying Miller capacitance. The "Gyrator" comment was an aside (but one that belongs in the conversation).

Not getting enough encouragement? OK, third time   : Work on it. Please. It's an idea worth exploring.

I always read too fast. I thought you meant that I designed a gyrator and you needed more convincing. You did know exactly what we were talking about, nice .

Encouragement is not what I'm after. I thought you guys meant that it would never work. You are right, inductors any plenyful but saying variable caps are hard to obtain is wrong. On ebay you still have a large selection to choose from for a price lower than a good wah inductor.

The interesting thing is that this could both work in a miller and gyrator type circuit. The last one is even better as you don't need an inductor anymore. I have one lying somewhere else. I might even breadboard it after the weekend.

[earthtonesaudio]

You could press the polepieces out of a single coil pickup and replace them with slightly-smaller ferrites, then use that as your variable inductor. 


For that matter, how about a guitar pickup with all its polepieces mounted on a common lever, so you could just drop them all down or raise them up to adjust output?

[gmoon]

You could press the polepieces out of a single coil pickup and replace them with slightly-smaller ferrites, then use that as your variable inductor.  
For that matter, how about a guitar pickup with all its polepieces mounted on a common lever, so you could just drop them all down or raise them up to adjust output?

Good thought. However, the DC resistance of pickups (even single coil) is too high to maintain a decent Q factor. At least in this circuit, at these frequencies.

The Wurli inductor is 190 ohms. Vox wah fixed-inductors range between 15 and 85 ohms. It's possible to find cheapo 1960's MIJ pickups as low as 1.3K DC Res, but the frequency response looks like an ordinary RC filter above 1K. It flattens out considerably @ 500-600 ohms.

But the idea of using a pickup bobbin, one with a slot rather than holes--and a ferrite bar--is inline with what I'm saying about inductor geometry. The shape itself could really have a effect on the linearity of the response, just like a custom-taper POT...

[earthtonesaudio]

I like that idea!

[Nasse]

Goes really off topic, true inductor but this must have been what I was readin

http://www.elektor.com/magazines/2000/december/1-kh-synthetic-inductor.56662.lynkx
http://electronicsprojects.mediadir.in/category/electronics-circuits/page/38/

[gmoon]

Goes really off topic, true inductor but this must have been what I was readin

The gyrator stuff looks pretty useful to me for filters. AFAIK, they may have already been used in an inductor-less wah, somewhere.

Gyrators do fall short of inductors in some ways, though. They can't store energy in the same fashion, and the damping characteristics are different, so they might not resonate quite like a real inductor. For some applications, that's a good thing. For a wah, maybe not. I'm sure they are worth investigating, and I'd guess that the results would be at least as good as a Twin-T wah...

Seems to me (unsubstantiated, at this point), that a variable-inductor filter interacts to dynamic core movement in a fairly complex fashion.

[PRR]

> Gyrators ... can't store energy in the same fashion

No, they get energy from the battery. So they can be "better", or you can throttle to essentially the same action.

> you might as well go all the way and...

At this level of complex abstraction, you may as well go to the next step. Lots of variable-center bandpass filter designs exist. The simple Moog ladder is low-pass which can be fed-back as a ringy low-pass, not a true bandpass. Two multipliers (may be opto-resistor dividers or OTAs) and a quad opamp makes a lovely 2nd-order filter (same as an L and a C). Switched-capacitor filters can be varied with a digital clock. Both of these offer in addition to bandpass, hi, lo, and notch connections (not a Wah, but maybe of some other use).

But stone-age and bronze-age technology IS fun to play with.

> ebay.. There is one 33H unit

I'll bet the seller can't read milli/micro inductance units.

Some of the 339 variable caps will be 30pFd-10pFd jobs which are great for FM/VHF work but will be hard to use in audio.

Many larger caps are "grounded frame" which spoils simple Miller-multiplication. You can float the frame on plastic and use a plastic knob, but it still tends to be a buzz/radio antenna (and Miller will multiply that too).

No doubt you can get something to work, without any post-1960 technology, and have fun in the process.

[jasperoosthoek]

Many larger caps are "grounded frame" which spoils simple Miller-multiplication. You can float the frame on plastic and use a plastic knob, but it still tends to be a buzz/radio antenna (and Miller will multiply that too).

My fear also as expressed in a previous post. Indeed my variable cap has a 'grounded frame', at least if you mean that the frame is connected to one side of the cap (the moving side in my case). Mine is currently connected to a breadboard with a 5H hammond choke. I'm eager to start testing this idea. The choke was not designed for audio either but at least it only has a 135 ohms series resistance. So I expect the Q to be very similar or even slightly larger than the 500mH fixed inductors with 15 to 85 ohms series resistance.

[gmoon]

I haven't been completely asleep. I rejected a different two-transistor circuit, and moved on. Here's a new version; this one uses an LM324 op amp instead of the BJ transistors. Here's a short and rough vid clip:

http://www.youtube.com/watch?v=wQOg6E0zFrk

The LCR filter is setup as a notch filter in a negative feedback configuration, which is different than the Kay original. Thanks to the higher impedance op amp, there is WAY more high end definition than with the BJT. Since I only used two of the quad amps, I'm sure a non-inverting input buffer stage could be added for higher input impedance (and an output buffer, too).

The vid sample omits the R3 resistor, so the Q is at the max for this setup. Frankly, it's more vocal with a lower Q factor, but it still sounds cool.

I apologize (again) for the audio quality, it's a crappy camera.
EDIT: yeah, that really is poor quality audio. It's not that distorted.



Here's the freq response (approx, I had to overlay one graph and resize that by hand, but it's within a couple dB of the sim). Blue is @ 0.07mH, Red is @ 0.57mH.



Mind you, there are a lot of variables I don't know about the inductors, so LTspice sims should be taken with some skepticism. However, I don't think it's too far off.

[jasperoosthoek]

I'll watch your video tomorrow when I have audio.

I haven't slept all the time either. Among other things I breadboarded the variable capacitor wah. Unfortunately it oscillated albeit at audio frequencies from a mid to a high 20k-ish pitch. It least it proves that the miller arrangement worked. I need to simulate it to get some better results.

I have one question about the Kay: the output resistance of the transistor is 330 in parallel to the collector impedance. How can such a small load resistance generate a good Q? I tried to replicate that value in my cap wah but it sounded like a high cutoff with a slight peak maybe. It oscillated at higher resistances loading the cap and inductor. At least the scope traces with a block signal didn't show a decaying sine, which is what you expect. So how does that work in the Kay?

I have to spend some more time on this, maybe I just overlooked something.

[gmoon]

I couldn't stand the audio quality on that vid, so I recorded a couple more clips with decent equipment.

http://www.youtube.com/watch?v=MuY76kIDLGk and
http://www.youtube.com/watch?v=KZEcr5WWQWc

No additional effects. Two mics. Minimal post-processing (just fade in/fade out).

Oh, and R3 is 100K in these clips. The Q was too narrow and harsh in the previous vid...(for my taste, anyway).

There's a few more construction details in the second vid, although the pedal is a little different now.

I have one question about the Kay: the output resistance of the transistor is 330 in parallel to the collector impedance. How can such a small load resistance generate a good Q? I tried to replicate that value in my cap wah but it sounded like a high cutoff with a slight peak maybe. It oscillated at higher resistances loading the cap and inductor. At least the scope traces with a block signal didn't show a decaying sine, which is what you expect. So how does that work in the Kay?

I'll let the EEs answer that, but I view the Kay circuit as a boost, followed by the LCR filter. The "Q factor" is the result of a filter with innate resonance. Even in a passive circuit *I think* it's a second-order filter.

[jasperoosthoek]

I have a good theoretical understanding of these filters, but haven't put it into practice. Indeed by nature LCR or LC filters are second order filters.

I'm very impressed with your variable inductor wah! I must say that it is much easier to get some results compared to the variable capacitor. If the variable inductor is too small you can just use a bigger capacitor. In the variable cap case you would have to go to 100H inductors (!!!) or use a Miller multiplying circuit which works only to a certain extend. Mine started oscillation at high amplification.

Where do find the variable inductor? I have a hard time finding a source for them.