AVOID DISTORTION/INCREASE HEADROOM, in CRYBABY WAH?

[matt239]

How to avoid distortion in a crybaby-style wah?
I'm looking for a wah capable of clean sound with a pretty hot guitar signal.

I have already made the Q1 emitter resistor 1k (usually 470R, or less)
I have the stock 68k input resistor, & stock 33k "Q resistor" stock .01uF sweep cap. - 2N5088 transistors, red Fasel..

It sounds pretty clean, but I still get a little splat on loud notes.
I feel like if I dial it back much more, it may become too mild of a wah sound. - As it is, I WANTED a bit of a mild, mellow wah, and this is pretty good, but if it gets too much less aggressive, it may start to just sound like my guitar's tone control.

I'm afraid that raising the emitter resistor any more will lose to much bass, & not have enough gain to really wah. & I'm afraid if I increase the Q at all, I'll get distortion. - I could make the input cap smaller, but I think it will sound too thin. - I'd actually like to be hearing a little MORE bass. I'd also like to lower the sweep a bit, say increase sweep cap to .022, but I'm afraid THAT will give me more distortion..

I can increase the input resistor, & even divide the signal down with a resistor to ground at the input. (I'm using a true bypass switch) -but padding it down too much and then having a lot of gain in the circuit seems like it's "bad design;" asking for noise.

Am I missing something? Am I going about it all wrong? - Should I be trying to bias Q1 for "soft" non-linearity? Should I try a 300R emitter resistor?
Should I just not sweat wasting signal by dividing it down at the input?
- Do I really need to start running it at 18V?  - I mean I've got big strings, & a hot pickup (14k DCR) & a 1Meg volume pot, but I'm not running a booster or anything before the wah, - guitar straight-in.

At some point, I intend to put op-amp buffers before and after the circuit, & then I wouldn't sweat having a large (150k?) "input" resistor between the buffer & Q1, but right now, I've got the board in the shell "old-school" buffer-less.

P.S.: In the interest of eliminating other things that could be the problem: A SCRATCHY WAH POT should make noise whether or not there is signal right?
- like you engage the wah, and sweep the treadle, but don't play any notes; it should crackle all the time right?
- I only hear artifacts when I actually play.
The pedal shell & wah pot I'm using seem fairly new, & little-used, & it's a sealed Dunlop wah pot. - Seems like it shouldn't be scratchy..
I don't have another one handy to test..

[Gus]

A short cut is to use something like lt spice to test you experiments and find the "cleanest" collection of parts values

One issue is you have a low parts count circuit running at 9VDC and it can only be so "clean"

Have you tried higher value series input resistors?

[PRR]

> even divide the signal down with a resistor to ground at the input.

You want *2* resistors (divider). Just loading-down the source is inconsistent and rude. (And liable to make the *source* distort if it is any active thing.)

> padding it down too much and then having a lot of gain in the circuit seems like it's "bad design;" asking for noise.

True, but-- the total dynamic range is pretty much fixed. And in this case it should not be small. If you can't raise the doorframe, you need smaller people.

Certainly an input pad IS the easy/fast next-step.

If there is no DC on the pot, it *may* not crackle no-signal.

Crackle can also be any poor joint. It may be "enhanced" by signal and filter action.

[Transmogrifox]

Another easy way to increase headroom is to run it on 18V.  You may need to tweak the bias somewhat but you won't need to really *ADD* anything (except an extra battery clip if using batteries).  Get rid of the 9.1V Zener and replace any caps rated for less than 18V.

[matt239]

Thanks everybody.
I think I have tried 100k for the input resistor before & it worked ok. Could go a little larger even, since true bypassing, & don't need high treble to make it through wah filter when on anyway.

Paul, by divider I meant the 68k series resistor, & something in the pull-down-resistor spot on the Tonepad board, normally 1Meg, could use 470k, or 220k.
It comes before the input cap & series resistor..
gaussmarkov raises the question of DC leakage onto the pot here: http://gaussmarkov.net/wordpress/circuits/noomerang/  seems unlikely to me but IDK.
I only hear a very subtle "scratchiness" or subtle, quiet "gak" or "ticking" at a couple of points in the sweep, only if I play the guitar.

I'm beginning to think the main distortion problem may be mis-biased transistors. (Even less headroom if we're sitting too far from 1/2 voltage.)

http://www.diystompboxes.com/smfforum/index.php?topic=107515.msg1035656#msg1035656

http://www.diystompboxes.com/smfforum/index.php?topic=96176.20

What VOLTAGES would we like to see where on the transistors?

[matt239]

OK, I forgot R.G.'s article had a whole section on the bias / voltages on the transistors. And I needed to refresh my memory on emitter-followers..

We'd like to see about:
Q1: C: 4.5V, B: 0.6V, E: 0.1V
Q2: C: 8.5 to 9V, B: 4.5V, E: < 4.5V  ish...

I suspect the bias is pretty far off in a lot of stock crybabys.
The way the circuit is set up, getting the bias right tends to increase the gain of Q1.
I was surprised that changing the emitter resistor, & having more gain, didn't result in much more distortion than having less gain, but the bias offset.

Perhaps it would be helpful to have bias resistors for the base of Q1...
Maybe it would be useful to have a trimmer on the collector of Q1, as well as the often seen Q1 emitter-trimmer.

We have more leeway with Q2 since it's just a voltage follower, not biased to amplify. It's not to hard to bias it better by tweaking it's emitter resistor.
- We could probably change it's base-bias-resistor (the 470k from the collector of Q1 to base of Q2 "around" the wah pot.) without changing the sound of the circuit much, if we needed to.    - It seems worthwhile to check the voltages on Q2 & make sure it's fairly close.

So far I have been able to reduce, but not eliminate distortion, with the hot signal I'm feeding the wah. - There may be some absolute limits;
- The filter has to have fairly high gain to "wah" properly. The gain can be reduced, but only to a certain point.
If we start with a signal that may have nearly 1V peaks, & have gain through the filter of 20X to 60X, & 9V supply.. Distortion may be inevitable, regardless of what circuit/devices we use; op-amps, or transistors..

Even padding the signal before Q1 seems to have limits. - The circuit is sensitive to what comes before it. A large series input resistor reduces distortion, but also changes the effective gain of Q1, & it makes the sound brighter. I had good results from reducing the value of the "pull down resistor" to 470k, which doesn't seem too "rude" to anything that might come before the wah..

At some point, even though the filter still has gain, we go below "apparent" unity gain. It's less than ideal to have a big perceived volume drop when you kick on the wah..
- To be fair, the signal from my guitar is probably "too hot." I have the hot pickup, and only a single 1Meg volume, that's always in circuit. - That was really just an experiment, & not something I'm married to. It sounds good straight into the amp, but it can overdrive delay pedals & such if I'm not careful.
I will probably change to at least have a 500k volume, & 500k tone, & maybe a small series resistor. (4.7k?)
- That would get me less signal voltage into the wah, without a big difference between wah-on/wah-off, & keep me from distorting delay pedals..

I tried a 2N3904 @ Q1, it seemed to make the distortion "softer," but didn't eliminate it, and came with a significant (surprising) noise penalty. Of course 2N5088s are spec'ed to have lower noise. - With 2 2N5088s, the unit was actually EXCEPTIONALLY QUIET for a wah..
- I didn't try a whole bag of 3904s, or 5088s, & I can't at present measure them. - I intend to get a meter with that capability soon.
- I don't usually go in for "transistor-swapping" but hfe at least, will make some difference @ Q1 in this circuit.
- I have some BC109s I could try.. (are they less, or more noise than 2N3904, on average?) I could sort 2N5088s for lower gain. Or maybe any 5088 will work fine with the right resistors..

Input buffer: As I mentioned, the filter seems somewhat sensitive to what comes before. To my mind this seems like an argument FOR an input buffer.. so the wah will filter the same, wherever you put it in your chain.
- When I put a buffer before the wah, it made it significantly brighter, in a way I didn't care for, but I think this could be adjusted out.
- The trick would be to install your buffer, THEN TUNE THE WAH, WITH THE BUFFER IN PLACE, not the other way around..
- Output buffer had little or no impact on the tone, seems good to include one, so wah can be used before a fuzz..

Stock crybaby, & Tonepad layout, has 1k resistor in series with power supply. Perhaps this could be reduced to 100R ? - Some wahs may have series diode polarity protection; maybe we should either switch to parallel diode, or use one with a smaller voltage drop (1N5818, etc.)

Running the wah from 18V would probably work well, & I may do so. - I wanted to see how far I could get with 9V, & learn more about how the circuit works.

I plan to do another round of testing in the coming weeks. - I'll report back.

[R.G.]

There's a world of experimentation to do on the classic wah circuit, OK. The circuit was designed at a time when transistors were the equivalent of several dollars apiece, and when getting an effect at all was earth-shaking, so the best of engineering work didn't go on to find ways around the limits. Another confounding factor is that hot output pickups were unknown at the time.

I suggest you do more of your tinkering in a circuit simulator. It'll save you a lot of time and money over hardware tinkering.

Ultimately, there is a limit to what can be done with only a 9V supply. Really hot pickups may produce a couple of volts of peak output. Since not all the 9V is usable, there's a very limited gain that can be applied to a circuit powered from 9V without distortion.

You may want to consider doing your experiments with a 17V supply. That's pretty easily obtained from a charge-pump doubler.

Another approach is to make the gain stage be a rail-to-rail opamp replacing the first transistor. This will sharpen up what can and can't be done in 9V for you as well as making bias arrangements clearer and separate from the feedback that gives the wah effect.

[R.G.]

I just ran some sims of a wah with the same basic circuit but with the amplifier and follower made from RTR opamps running from +/- 12V. It will happily produce bipolar 10V peaks at the resonance.

I think you're right - it's non-optimal bias points. Both transistors need to be optimally biased for the power supply, and then you only get half the power supply (max!) for voltage outputs.

[matt239]

Thanks R.G.!
I always liked your idea of building the whole thing with op-amps. - I don't really feel up to etching a board for it though..
I'm also interested in inductorless wahs.. It's still fun to see what can be done with a crybaby though.

I should really sim the circuit. Not being an old-hand with the simulators, it might take me as many hours as building the circuit, same with breadboarding,
- It's easier for me to keep track of what's happening with a nice, labeled PCB.. but I must forge ahead and sharpen my skills with those things.
- As it was, I used sockets for several components on the board, transistors, resistors, & I had all the parts on hand. So I didn't have to spend $ trying different parts, I could just swap them in & out.

I well understand your point about the circuit being designed in a different era, & low parts count, etc.
- We seem to cling to these circuits, perhaps longer than we should, out of nostalgia, superstition, laziness, & cost saving..
Even some small tweaks might be really helpful. - What about bias resistors for the bases?

I'm not well versed in the limits of voltage swing we can get from a one transistor amplifier. Is 1/2 voltage really all it can do? peak to peak?
I would've thought it could go, say, 4V each way, but obviously, IDK.

I think I WILL build it with a charge pump, but it's been fun to see what could be done without it.
Also, as I was saying, my guitar output is hotter than I even need/want it to be. - A 14k DCR pickup, & 1Meg volume, is really kinda overkill. I can distort my delays with it. - So I'm going to dial that back a bit.

As you say, lot's of things to play with here. - I've been experimenting with some Boomerang/Noomerang values, & I'm really pleased.
I like the tone, & it works well on middle & neck pickups. - I just can't play on the bridge pickup, full volume, without getting "splat" on loud notes.
- So the tweaking continues.

[R.G.]

I well understand your point about the circuit being designed in a different era, & low parts count, etc.
- We seem to cling to these circuits, perhaps longer than we should, out of nostalgia, superstition, laziness, & cost saving..
Even some small tweaks might be really helpful. - What about bias resistors for the bases?

In many cases, the circuits themselves were the best that could be done cheaply in that simpler era. They introduced side effects that became part of the sound we expect. In some cases, the circuit was built almost entirely for the side effects. This make re-engineering more complicated, to say the least. 

I'm not well versed in the limits of voltage swing we can get from a one transistor amplifier. Is 1/2 voltage really all it can do? peak to peak?
I would've thought it could go, say, 4V each way, but obviously, IDK.

It gets complicated. For a single transistor with a grounded emitter and a custom-tweaked bias, you can get nearly all of the power supply as a peak to peak output. That has to be tempered by the fact that in the real world, you're going to need some emitter resistor to get stable operating points, and that you can't hand-tweak biases very well in production; as well as not being able to guarantee that the input voltage will drive the base equally well with different biases. Sure, you might get 8V out of a single transistor that's carefully biased. The + side (for an NPN) will cut off sharply and the - side of the signal will be mushily distorted.

Of course, a typical opamp can't go to the rails either. A TL072 will lose 1.5V typically from both sides of the power supply  and can only do 6V p-p from a 9V supply. That's why I mentioned rail to rail opamps. They get you that extra 3V of p-p swing. Single supply opamps like the LM324 will do something in the middle, being able to swing to within 50mV of the minus supply but only about 2V of the + supply.

What's really going on is that the circuit has gain at the resonance. You have a hot pickup with a few volts of signal. The circuit tries to amplify that by the open loop gain of the transistor stage at the resonant point. The Vox inductor-based  circuit has a gain of 20db or more at resonance, and even with rail to rail output amplifiers, 9V could only accept 900mV in without distorting.

The circuit can be tweaked to have lower gain overall, so that the resonant gain is perhaps only +6db to +10db. The feedback then drops the non-resonant level back to a net loss of signal. That's only going to be OK if you bypass the whole thing; otherwise the average signal level will be lowered. It will probably also be subjectively quieter than the bypassed signal if you do bypass it.

Down at the bottom of this, you're running too high a signal level from your guitar for 9V powered effects with any gain at all. As you note:

Also, as I was saying, my guitar output is hotter than I even need/want it to be. - A 14k DCR pickup, & 1Meg volume, is really kinda overkill. I can distort my delays with it. - So I'm going to dial that back a bit.

Most delays will have problems with that big a signal too.

You have to be careful in cutting back your signal level. High-output pickups generally have high inductance from the larger number of turns of fine wire, and that makes them prone to treble loss. That's one of the reasons that humbuckers are generally thought to be duller than single coils. And it makes them more prone to treble loss from loading. If you want to avoid this kind of issue, you're going to need to think about what kind of treble loss is acceptable and deal with it by either buffering before padding the signal down or lowering signal with loading and taking whatever losses you get.

[matt239]

You have to be careful in cutting back your signal level. High-output pickups generally have high inductance from the larger number of turns of fine wire, and that makes them prone to treble loss. That's one of the reasons that humbuckers are generally thought to be duller than single coils. And it makes them more prone to treble loss from loading. If you want to avoid this kind of issue, you're going to need to think about what kind of treble loss is acceptable and deal with it by either buffering before padding the signal down or lowering signal with loading and taking whatever losses you get.

I have given that some thought. I actually went a little overboard trying to see how much level & brightness I could get out of the guitar. At this point, I can afford to give a little back. I think I can live with two 500k controls, that's still less loading than most stock guitars. - And adding resistance in series can waste a little level without disproportionately rolling off treble right? - I can certainly experiment with the guitar wiring. I can test that out on the breadboard.

Thanks for explaining more about the transistor & op-amp voltage swing. I knew they didn't go rail-to-rail, but I wasn't sure why, or by how much.

Am I correct in thinking that to get Q1 biased correctly, without setting the gain higher, would take some more resistors than are in the stock circuit?
- Like biasing resistors for the base of Q1?

The + side (for an NPN) will cut off sharply and the - side of the signal will be mushily distorted.

So, does this mean, if we set up the gain to be reasonable, & not distort a lot for most guitars, that we might want to bias Q1 a little low, to get a gentler form of distortion on peaks? - and avoid "splat" artifacts? - Clipping the low side, not the + side?

Plenty of people expect/like a little grit from a wah. - I like that sound too, but I also like a clean wah sometimes, so I've been working towards that.

Just increasing the Q1 emitter resistor (to say, 1k ) and the input series resistor, (68k to 100k) gets us maybe 90% of the way there. & might be enough for most people / guitars.
It changes the filtering, less bass.. Can sound a little thin. But we can increase the value of the feedback, or "sweep" cap, & I liked increasing the value of the feedback, or "midrange" resistor.
- But changing that emitter resistor does raise the bias point..

It's easy to get close to what we want, harder to get that last little bit of the way there..

[matt239]

A couple of questions:

1: Does the Q1 emitter resistor "Re1" only change bass because it changes the gain?

- If we alter the gain by changing the collector resistor, "Rc1" will it have the same effect? Or is there something special about Re1?

The Electrosmash analyses for Crybaby & Vox wahs, have graphs showing this mod increasing bass. So it seems it's not just perceived as more bass because it's louder.

- I went ahead and simmed the circuit. (for DC at least.) If we change the 82k Rb2 to 100k (In some cases it's already 100k) and make Re1 1k, and Rc1 28k It gives me EXACTLY 1/2 voltage on the collector of Q1. (With what i guess is supposed to be an average or nominal 2N5088)
We can also get very close by leaving Rc1 @ 22K & making Re1 680R.

- Is there any reason to think changing Re1 is "better" than changing Rc1?


2: I don't really understand how Rb1, & Rb2 function. - I know it's basic..  I think I understand why increasing the value of Rb1 raises the voltage @ Q1's collector, but I don't understand why increasing Rb2 DECREASES the Q1 collector voltage?

[R.G.]

You have to think about this on two levels. One is how the individual resistors affect Q1 as an amplifier, its gain, input and output impedances and so on before feedback is added, then on the next level as what the feedback network does to the basic amplifier's operation and response.

Changing Re and Rc change the open loop gain of the basic amplifier and its input and output impedances before feedback. Then the feedback is applied, and this shapes the frequency response, as feedback interacts with the impedances and gain.

That's one reason I went to the opamp model of the circuit in the first place, way back when. The opamps let me set the amplifier gain and impedances independent from the quirks of a transistor to see what just the feedback does. Then I could add "imperfections" to the interior amplifier in the form of lower input impedance, lower internal loop gain, and so on.

The response of this network does have some quirks in the interaction of interior gain and impedances and feedback loop response. I'll go model up the transistor version when I get a chance.

[R.G.]

The values of the input cap and the cap from the collector of Q1 to the wah pot have a huge effect on the amount of bass in the output. The original design must have been tweaked to use these to cut a lot of bass out of the response. Making them quite large, about 1uF each, makes the shape of the response be much more like the opamp circuit's response, which I DC coupled. This also accounts for some of the loss of signal at bass settings of the wah pot - the input and output caps are strangling the bass response on top of the peaked low-pass nature of the resonance.

[R.G.]

Bypassing the 470 ohm emitter resistor with 4.7uF makes the open loop gain of the Q1 amplifier much higher - and makes the resonance very, very peaky. About 40db peaky, too much of a peak to be useful except for strange stuff.

[R.G.]

Forgot earlier - the 470K/82K from Q1's collector offer an independent way to adjust bias point. They divide the collector voltage down, then feed the divided DC voltage on the collector back to the base through the 1.5K resistor and inductor resistance, which is what sets the bias point on the collector. You can tinker the value of the "82K" to change the collector voltage without changing the emitter or collector resistors, preserving the Q1 gain from these two.

[bloxstompboxes]

GOLLY. Four posts in a row from R.G. You really got him interested in something. lol.

[R.G.]

I'm OK. The frenzy will pass soon.

[matt239]

Well, it's an interesting and quirky circuit, and very popular, but not always well understood. - Right up R.G.'s alley. Always good to hear his thoughts.

Forgot earlier - the 470K/82K from Q1's collector offer an independent way to adjust bias point. They divide the collector voltage down, then feed the divided DC voltage on the collector back to the base through the 1.5K resistor and inductor resistance, which is what sets the bias point on the collector. You can tinker the value of the "82K" to change the collector voltage without changing the emitter or collector resistors, preserving the Q1 gain from these two.

Ah, I get it now! Thanks! I was thinking about feedback, but not quite seeing how those two resistors form a divider & feed Rfb, from their midpoint.

Everybody wants to know what values "sound good" in a wah, but of course it will depend on the properties of the transistors, especially the hfe of Q1, & the value of the inductor, - not to mention personal taste.. Not much attention is given to Rb2, but it might be smart to socket it, or use a trimmer, so the bias can be adjusted independent of your preferred gain arrangement.

Excellent results, IMO, by increasing the values of the caps you mentioned, R.G..
I've been using the Boomerang/Noomerang values for a while; .047uF input cap, & 1uF for the caps that couple the wah pot, (C-out, & Cef)
- Much less "constricted" & "nasal," but still "wahs" nicely.
For anybody who has not heard Boomerang wahs, I highly recommend a listen.

[matt239]

So the gain of the filter is highest, when the wah is in it's heel-down position?
Is that accurate?

If so, that would explain getting distortion only in the heel-down position, rather than it being "too much bass?"

And then if we moved the wah pot gear a notch or two FORWARD, it would not ever reach it's highest gain, and not distort there?

(Or we could add a resistor in series with the pot to ground..)

And if that's too bright, we can use a larger value "sweep" cap..

(Hello again! 😉)