DR QUACK build report

[Transmogrifox]

I breadboarded a Dr. Quack last night, and was quite disappointed to find it to be finickey an rough.  It seems to produce an undesirable distortion on the lower sweep range, though it is not too terribly bad once it breaks the threshold.

Another problem is it is hard to set it to be "milked" into a certain frequency range, it goes real high, fast, or it stays low, and it's quite a balancing act to get it into an ok setting with sensitivity and Range.

I did some mods to improve its performace:  I added a 1.2k resistor in series with the 2N3904, which seemed to help some.  I also added some extra capacitance to the envelope detector to lengthen the release time.

I still never got rid of the distortion that it produces, however it sounded very sweet when I modulated the envelope detector output with an LFO :  wahwawahwawahwawawawawawah

very cool...if only I could get rid of the undesirable distortion.

I also put it into my "Haw-waH"  feedback thing, and it produces a very cool envelope phase sound.  Very nice, and it almost sounds like a chorus when it is modulated quickly.  I like it for that.

Has anybody else had trouble with the Dr. Quack getting distorted and being finickey?

Is there any magic to the 47 Ohm resistor in the op amp's power supply that may rectify this?  I omitted it because it seemed superfluous.

[smoguzbenjamin]

Maybe you've got undersignalitis, try increasing the feedback resistor in the EF opamp. If that doesn't work, try something else

[Mark Hammer]

With ANY half-wave rectifier like the DQ uses - and this is true of anything that uses such a rectifier, whether its a filter, compressor or noise gate - the trailing garble as the filter settles down towards the end of a strummed note is a relfection of ripple in the envelope signal.  Essentially it is the filter being subtly modulated at near audio or audio frequencies.

More technically speaking, it sounds like crap.

The traditional approach to solving this is either to use a much better envelope follower (which would require a redesign of the DQ), a control element which is much more forgiving and able to hide such flaws (like a photocell), or by stretching out the decay time to smudge over the small perturbations in envelope signal.  

In this case, your salvation lies in swapping the 10uf capacitor between the envelope detector and transistor for a higher value such as 22uf or even 33uf.  Note that although this will "cure" the envelope ripple problem, several new problems will crop up:

1) Since the filter will now take longer to settle down, the "normal" setting/mode will seem to take forever to come back down again.

2) The slowness of recovery will make it less responsive to successive note/strums if they come too soon.

3) The attack time is a joint function of the series resistor and cap to ground, so that increasing the cap value will also slow down attack time.

For these reasons, my own solution is to use a switch to kick in a second cap in parallel so you have two decay cap values (e.g., 10uf and 10+10uf), and a 500r to 1k pot as a variable attack resistance.  This will let you make sound choices about matching time constants to style needs.  If you pick fast enough, you won't get to hear the ripple so you won't care about it.  If you need to hear the full sweep cycle, and can afford to wait longer before you pick, add the second cap.

The pot is wired as a variable resistor in the following way: take out the 100R resistor and run leads from those pads to the two outside lugs of a 500R to 1k linear or log pot, with a 47r resistor straddling the wiper and one of the outside lugs.  When set to "minimum resistance", it will yield about 43 ohms (with a 500R pot) and an attack time about twice as fast as the stock value.  When rotated the other way, resistance will increase and attack time will slow down.  Kick in the second cap and it all increase proportionately.  

Note that increasing the resistance of the pot will reduce the apparent sensitivity of the unit because the increased resistance limits current.  The smart thing to do here is increase the 2M2 feedback resistor to provide a bit more gain so you can still get decent sweeps at slow attack time settings.

Try it, you'll like it.  I promise.

[smoguzbenjamin]

So it's an attack speed switch. Wahoo it looks like the doctor Quack is being perfected many years after its original appearance

What I'd do is have a large attack gain, tweaked so that the max CW turn of the sens pot is at the limit of the useable range. Gives you diversity. In a while I'll be breadboarding this out and tweaking it untill it sounds like 7th heaven. I'll post my results on my website but you'll have to be patient.

[B Tremblay]

Wahoo it looks like the doctor Quack is being perfected many years after its original appearance

A number of Mark's excellent suggestions are present in the Nurse Quacky at runoffgroove.com...

[Transmogrifox]

Thanks for you input guys.  I was thinking of replacing the 2N3904 with a  CMOS switch and pulse-width modulating it at 100 kHz, making it like a cross between MXR envelope filter and Dr. Q.  I know I can make a better envelope follower using the pulse-width trick.  I think I'll use a precision full-wave rectifier and more parts, just because I'm anal about having a very controllable envelope filter.

[Paul Perry (Frostwave)]

A full-wave rectifier is only going to be "twice" as good, which may or may not be worth while.
Whether one of those higher order switched-cap filters would be worth using, I cant say, but maybe someone will give it a go.

[Mark Hammer]

The quality of envelope detection shares a lot in common with the quality of simultaneous translation.  The faster you try to do it, the sloppier it gets, and the more effort you put into doing it right, the farther behind you fall.

In that sense, I think for performance purposes it is probably foolish to spend too much time and effort in pursuit of the "perfect" envelope follower (although Harry Bissell's design comes about as close to perfection as anything I've ever played; pity it uses more parts than most pedals do in their entirety), and smarter to focus in on what it is about the envelope control that you want to optimize. or emphasize.  

Indeed, that is pretty much the tack taken by the overwhelming majority of envelope-controlled filters.  In truth, most of the envelope detection circuits I've encountered in E-H effects are pretty shabby as generic envelope followers, but the fact remains that any one of them can sound fantastic under the right circumstances, and posted sound clips are focussed on those circumstances.  Even the more extensively thought out filter pedals like the current tube-based Zipper, as great as they sound, my first thought when hearing the sound clips is that they don't really adress the way I'd like to use such a pedal, but for the way they are being used they sound great.

That, I think, is the reason why envelope-controlled filters never really took off until recently.  For the longest time, guitarists avoided them like the plague because they required more deliberate playing and planfulness to bring out the best in them; something that runs anathema to the dime-it-and-wank mentality all too frequent amongst guitar players.  If you think of them as devices to suit very specific purposes - not QUITE as targetted as say ring modulators but in that direction - you learn to live with them more comfortably.  Like spouses and girlfriends, no filter or envelope follower circuit is going to be all things for any occasion.  Learn to love them for what they do well and you get along better.

If you're up for it, though, a FWR follower will probably be less susceptible to things like old-strong-wobble (older strings that need changing tend to "beat"; the beats are detected as envelope fluctuations during the decay cycle).

[Transmogrifox]

Mr. Hammer:  You have been around this stuff and have likely tried about every conceivable trick under the sun, but I'm going to take a shot at this and see if I can clip the signal and try a few modulation tricks to see if I can translate some of the higher frequencies down, use some creative filtering, use up an entire breadboard and determine that you were right.

Thanks for your response. I can handle a one-trick pony if I can learn to make the pony do its one trick...and I think I have the trick of the DQ down on my breadboard now.