Inductorless wah with Inductored wah response - worth pursuing or not?

[Rob Strand]

These examples show why I messed with the resistors on the Twin-T network.
For the twin-T part of the circuit I've called "R1" the resistor on input side of the Twin-T (which connects to the output of the wah ckt).   "R2" is the output resistor of the Twin-T.

So the first circuit I take my V1.9 circuit and I reduce "R1".  Then I tweak the other parts so the frequency range and height of the peaks is about the same.  These are marked in Red.  I called this V1.9b



In the next case I take the V1.9b circuit then I also reduce "R2".  Then I tweak the other parts so the frequency range and height of the peaks is about the same.  These are marked in Red.  I called this V1.9d.



As you can see the first change we can deal with showing we can cope with a smaller "R1".  However, when "R2" is reduced we see the high-frequency peak reduce and the low frequencies change.  For V1.9d if I tweak other parts (like "C2" at the output of the twin-T) I also get into similar trouble.  This shows a high "R2" value helps.

[Rob Strand]

Hmmm, one thing I noticed in the example is the bass is *lower* in V1.9b.  I interpret that as the highs have been boosted + a gain offset.  Also the Q of the highs looks higher.

OK, so I just investigated trying to take advantage of the apparent increase in HF boost when the Twin-T "R2" is small value.   So it seems the equal Twin-T R1 and Twin R2 can be made to work.   The Q is a little lower and starts to deviate from the inductored wah.  However we can knock back the amp gain and reduce the attenuation.  So maybe if the Twin R2 is a tad larger we are getting close to balancing all the juggling balls with this nasty little ckt.

[bool]

How about a miller cap - would it make it well-behaved?

[Rob Strand]

How about a miller cap - would it make it well-behaved?

I'm not sure what is the best at this point.  Circuits where you have a lot of variables and a lot of interaction are tricky.   If you change something it is not clear what the best parts are to compensate. 

Some problems are best solved with maths but even that has it's limits here:
- The frequency pot significantly changes how the circuit behaves at different settings
- The target response of an inductored wah is already complicated and we need to match
  that over all the pot settings.
- Then there's the fact the T-twin circuit doesn't really match the target response. 
- So we can't actually match the response with maths we can only approximate it.
  And that's why I was setting up the computer tools!

[bool]

Yes ... that's neat. But real-life parts tolerances will screw up the computer model in no time! Unrelated to the wah, I was plowing through a batch of 10% MKT caps and you know, from a well-known manufacturer, most of 4.7nF caps were actually more like 5.1x nF and had to be discarded. Because these were for a guess-what - computer optimized circuit that just had to have a bit tighter matched timing components to work predictably.

Good luck... I still think that forcefully shoehorning a small miller in there could ensure a more consistent build (when you finally "get there" of course). I can also totally understand that you want to absolutely nail the damn bastard ... but for us regular punters I think a work-alike approximation would be more than good enough, ha ha!

[Rob Strand]

Yes ... that's neat. But real-life parts tolerances will screw up the computer model in no time! ..  Because these were for a guess-what - computer optimized circuit that just had to have a bit tighter matched timing components to work predictably.

Sometimes that comes from the spec more than the circuit.  Like in the old days we would have used 4.7nF + 390pF in parallel. 

The computer optimization thing works but the point where you get to the 3.1nF caps it is only the start, not the end point.  If you stop there it's only a half-arsed optimization.  You have to take all the 3.1nF values then swap them out for 2.7nF or 3.3nF then you might lock in those values and re-optimize the others and you keep doing that.  Then there's a third stage where you check the circuit still works when you have 2.7nF +/- 10%.   You might even go back and try 3.3nF and do the whole process again and see if you get a better result.  The reason that works is you might have a 12k resistor and end-up with a 12n cap but then when you go to 15k and a 10nF cap the results might be better (the RC product is close but not exactly the same, and it might trigger better matches elsewhere in the circuit).    You can actually get the software to allow for +/-10% at the start and only choose standard values, then optimize the values so the circuit has the least variation overall when all the parts have tolerances.   This is a tougher job and it's probably wise to start-off with unconstrained solution first and use that as a starting point.    The amount of work skyrockets to do all this.  Like you aren't going to do that to choose a 100nF on the input of an effects pedal.

Anyway the Twin-T wah is a total disaster as far as tolerances go.   If you want consistent results it's going to need a trimpot, probably in series with the grounded cap in the T.

Good luck... I still think that forcefully shoehorning a small miller in there could ensure a more consistent build (when you finally "get there" of course). I can also totally understand that you want to absolutely nail the damn bastard ... but for us regular punters I think a work-alike approximation would be more than good enough, ha ha!

Exactly!   in the last version I tried to reduce the gain by adding RE on Q1 this helps make the circuit less dependent on the BC capacitance.  However a small cap might help in the end.  I'll have to check it out if I ever get there.

[Rob Strand]

How about a miller cap - would it make it well-behaved?

FWIW,  it looks like even on the lower gain circuit the Cbc cap alone is reducing the available high-frequency peaks by about 2.5dB.   So on face value it is detrimental to the cause.  However, it might be helping flatten things out in the bigger picture.

[mac]

I see that there are 2 schools of thought: 1) a mod and 2) a total remodel.

Sounds like a Philosophy class


The B to C Twin-T network makes higher peaks at the middle of the sweep, so I tweaked the output filter to decrease those peaks,





The blue line goes 2db above the gray line. With the notch filter, the blue line is now almost at the same level.

mac

[bool]

Guys, you're working too hard ....

[Rob Strand]

Guys, you're working too hard ....

I think so too.  It's probably more of fighting the wrong battle
Using an opamp would fixed quite a few things.

FWIW, I had a look at what the T-network alone is doing; using the V1.9b circuit.   I plotted the T-network response with the loading of the amp and the input network.  I marked points which correspond to the peaks of the wah response (with EQ present).   The notches of the T-network don't line-up with the response peaks.  The more interesting thing is the peaks line-up with the -41.5dB slice line.   The phase at those points, which is roughly 180deg, must be allowing the peak to form.   It's all too complicate form my little brain to handle this time of the morning.

[mac]

Guys, you're working too hard ....

To be honest, I experimented with the CIW some years ago.
Just copy paste
Except the last sim which has a modified output based on one of Rob schematics 

I managed to tame the peaks, but since getting a wah pot down here is a difficult and expensive task, my main problem was the variable resistor.
A std linear 25kb works fine but it begins to crackle soon.
I tried led/ldr and 4n27, and additional parts to mask the crackling.
A fast or slow sweep and all the good mods are useless.

mac

[bool]

Hey, how about using a darlington? I mentioned before that a mpsa13 is biased happy with 2x 220k in c-b and a 150k b-e at 9v dc supply; and this could potentially be shoehorned to a wah circuit. (??) A darlington may be approximated as a "bad" inverting opamp amp at ac signal. I have done a couple of preamps this way, with some filtering / tone shaping and it could be done.

The other way I can think of is to bootstrap the first bjt (Rob's approach) collector to up the ac gain and make the circuit more "bad opamp"-like at ac.

But that would just mean more hard work for you guys, ha ha!

[Rob Strand]

To be honest, I experimented with the CIW some years ago.

I find it easy to burn-up a lot of hours playing around with it.
With the goal of finding a better way I was thinking about extracting key features like
the 180deg thing and the constants dB attenuation alone.

I tried led/ldr and 4n27, and additional parts to mask the crackling.
A fast or slow sweep and all the goods mods are useless.

You could add a *second order* RC filter between the pot and the LED drive circuit.  It is very effective at removing crackle. The problem with the wah is you still want the pedal to respond to relatively fast changes so that limits how low you can make the filter cut-off.   Once the pot gets dead spots from wear you are stuffed!

One of the Practical Electronics wah projects (Simonton ~1970) didn't use a pot at all.  It had a point light source, an LDR and a beam breaker, which was a blade that went between the light source and the LDR.

Another idea was a de-focussed lens.  As you push the pedal down the light becomes more focussed on the LDR and the resistance drops.   The problem is it doesn't go fully dark like the beam breaker.

I can think of all sorts of inductive and capacitive methods but they end-up being more complicated than the wah itself!       The only simple ones are magnet -> hall-effect -> light drive -> LDR  or Variable capacitance plates -> variable timing -> PWM -> filter -> light drive -> LDR.    I'm pretty sure there's at least one commercial wah out there using hall-effect.  Maybe Roland or Boss from the 80's or early 90's.

EDIT:
It might be the Boss PW-1 Rocker Wah.  I can't find a schematic or service manual.
http://www.effectsdatabase.com/model/boss/pw1

[Rob Strand]

Except the last sim which has a modified output based on one of Rob schematics 

I might need to take another look at what is gained from the output buffer.

[Danich_ivanov]

Sometime ago I was messing around with bandpass, and surely dual gang made everything much more consistent and peaky, i don't remember the exact circuit, but i think it was something along the lines of notch filter where you have 2 equal resistors with cap on top and cap on bottom, in the inverted config.,but i will get back to it at some point, as i want the same thing "inductorless wah' that sounds like a wah. And dual gang is not that hard to simulate anyway. Either way, i would recommend going beyond colorsound arrangement and try other things with filters and their active components, as, at least i found colorsound's filter to be somewhat limited and hard to get right.

[Rob Strand]

and surely dual gang made everything much more consistent and peaky,

Yes, the dual gang circuits tends to make it easier. The thing about the inductored-wah is the Q varies as the frequency changes but the peaks tend to stay relatively constant.   Varying Q + varying frequency is a side effect of a single-gang pot.  When you have a resonant circuit and vary only L or C both Q and frequency vary.  The inductored wah is essentially varying C.

[I forgot to mention, a lot of circuits which have variable Q also have variable gain at the peaks.  So that reduces the number of suitable circuits.]

As far as dumping the colorsound/Twin-T idea goes, the Ibanez circuit Mark posted is pretty good and it only uses a single-gang pot.

And dual gang is not that hard to simulate anyway. Either way, i would recommend going beyond colorsound arrangement and try other things with filters and their active components, as, at least i found colorsound's filter to be somewhat limited and hard to get right.

No doubt about it.  I guess the challenge is to see how far the Twin-T type circuit can be taken.

I've look at the Twin-T circuits in the past and they have all sorts of issues.  I was surprised I could get so close; as the previous posts show.

One example is the Twin-T Wah Project from Practical Electronics by Simonton (~1970). It produces very uneven peaks which rise with frequency.  However it uses a second stage, effectively another pot gang, to dial the gain down as the pedal was pressed down.  (IIRC, RG's site doesn't show the gain compensating stage.)   Others are very weak and/or uneven.

[mac]

One of the Practical Electronics wah projects (Simonton ~1970) didn't use a pot at all.  It had a point light source, an LDR and a beam breaker, which was a blade that went between the light source and the LDR.

Another idea was a de-focussed lens.  As you push the pedal down the light becomes more focussed on the LDR and the resistance drops.   The problem is it doesn't go fully dark like the beam breaker.

I used Photoshop in the past to print a fading pattern on a transparent film.
Much like the blade... but more complicated, doh!

What I like to do is to put a LED/LDR face to face in the down position. If the LDR is fixed to the box, the LED trajectory is an arc. As it moves away from the LDR the resistance increases.
- The arc depends on the LDR to be used.
- The LED wires can be a problem

mac

[Rob Strand]

I used Photoshop in the past to print a fading pattern on a transparent film.

That's a really cool idea!   I thought of using a graduated filter but I couldn't see how
to get one.

What I like to do is to put a LED/LDR face to face in the down position. If the LDR is fixed to the box, the LED trajectory is an arc. As it moves away from the LDR the resistance increases.
- The arc depends on the LDR to be used.
- The LED wires can be a problem

That's a cool idea as well.  You might be able to spec a beam angle for the LED.  However I remember looking at the LED beam angle for my bicycle lights and in practice they vary even if the spec is the same.   They also have funny patterns and dead spots.  I can only think of a reflector to make both parts fix.  Obviously the reflector is going to tarnish over time and the thing won't work after 10 years!

[Transmogrifox]

Here's another angle at it.  This implementation sort of gets at the change in Q with sweep with roughly constant amplitude and is still relatively simple to build:
http://cackleberrypines.net/transmogrifox/TransmogriNotes/skwah/index.html

Then of course you have seen my thread about the transfer function (at least, based on assumptions identified).

I haven't yet synthesized an inductorless circuit that faithfully reproduces the mathematical equivalent to the inductor wah, except for a gyrator used to emulate a grounded inductor.  With this topology you can get close, but there are other problems with the circuit practically speaking.

The gyrator implementation is too noisy to be used in a performance setting and that is where I stopped last time I messed with this...well, the DSP first-pass model is where I stopped last time, but that is not relevant to this discussion...and that model would garner criticism from purists anyway.  It goes wah, and pleases a fumbler like me

[diffeq]

I can think of all sorts of inductive and capacitive methods but they end-up being more complicated than the wah itself!       The only simple ones are magnet -> hall-effect -> light drive -> LDR  or Variable capacitance plates -> variable timing -> PWM -> filter -> light drive -> LDR.    I'm pretty sure there's at least one commercial wah out there using hall-effect.  Maybe Roland or Boss from the 80's or early 90's.

What about a following chain: gyro sensor - > PIC uC -> two matched JFETs (one in control loop, one as a pot)? Stuff like smoothing and min/max position can be tunable, but I don't know if sensor's sensitivity is enough for such an application?