Adventures in Troubleshooting

[Mark Hammer]

In this corner, weighing 325 pounds, with parts unknown, "The Customer".  His tag team partners, at 7 foot 2 inches and 180lbs (he's tall and he's long, girls), "The Mystery Signal", and at 250 pounds, "Mr. Language".

In the opponents' corner, weighing in at 200 pounds, "The Meter", at 240lbs, but rarely to be fully trusted, "The Schematic".  Their partner, at 6 foot 8 inches and 350 pounds, "Experience Man".

Now let's get REA-deeeeeeeeee to RUMMMMMMM-bllllllle!!!!!!!!!

A friend called me the other day with an interesting problem.  He makes a very nice, reasonably priced CE-1 clone.  In addition to what else people might say about it, they remark on its pleasingly noise-free performance, part of which is probably attributable to use of a wall-wart, and another part of which is attributable to improved op-amps from 1978.

In any event, a customer called/wrote to complain about "clipping" in the pedal.  Now, all BBDs have limited headroom, and are the weakest link in any BBD-based effects pedal.  The usual response to this challenge is to either a) use a compander to sneak hot signals past the BBD, b) keep the input stage as a unity-gain buffer, and c) include some sort of input padding.

The CE-1 uses strategy 'C', having a 50k pot on the guitar input to turn down the input signal to a manageable level.  In this case, it appeared the customer found no relief from the "clipping" by varying the input level control.  This, of course, suggests that whatever this "clipping" was that he heard had little to do with overload.  At least we might assume so.  What we don't know is the sort of signal he was feeding it, under what conditions he heard it (or didn't hear it), how he plays, or what it actually *sounds* like.  He may have been feeding it two cascaded Blackfires for all we know, but we'll assume that it was a normalguitar signal.

But what is "clipping"?  As avid fuzz-o-philes can attest, distortion has an infinite number of possibilities, and can be produced in many ways.  In that sense, if somebody tells you they hear "clipping", that could include so many audible possibilities, that they may as well just tell you "It doesn't work right" and that would be about as specific.

While on the phone, we both dug up our respective CE-1 schematics and dove in to them, becoming stompbox detectives.  It was a bit like CSI: The Chorus Edition.

The CE-1, like a few other BBD-based pedals (but not many), uses a simple noise gate to address audio crap.  Yes, there is the usual lowpass filtering before and after the BBD to eliminate clock noise and aliasing, but at the end of the wet path, just before it gets mixed in with the dry to produce chorus, there is an FET to ground, acting as a voltage-controlled resistor.  Like any noise gate, it cuts the signal out (in this instance by shunting it to ground) whenever that signal falls below a certain level.  When you play, any residual clock and other noise is masked by the instrument signal so even though its there you don't hear it.  When you stop playing, ALL you hear is the audio crap, so gating it out is a nice feature.  The nice thing about using a gate in this situation is that the gating action applies only to the wet/delayed signal, and is not applied to the dry/straight signal.  So, note dynamics are preserved.  You never lose the onset or trailing off of a note.  All you really sacrifice is that the effect starts maybe a few milliseconds later than it might in an ungated chorus, and ends a few dozen millisecinds earlier.  Who knows, it may even be that the fading out of one signal, while another lingers creates a stronger impression of two distinct sources - a true chorus.

So why am I taking about the noise gate?  If you look at my site (http://hammer.ampage.org) and scroll way down, you'll find a noise gate project from Jon Gaines that appeared in "Modern Recording" magazine some 20 years ago.  The article is there in its entirety, but what you don't see are the letters that followed for the next few months.  One of the issues that cropped up was a discernible "buzzing" or "trill" as the gate cut out.  That buzzing was explained as the consequence of the gate responding to small, rapid, changes in the incoming signal during the decay phase, often in the audio range (i.e., >20hz).  The recommendation was to lengthen the decay time, and indeed, that cured it, and was my big epiphany about the mysteries of envelope-controlled devices.  I owe that project/article a lot.  Since that time I have also found out that using reasonably new strings also helps to avoid such false triggering of gates. Older, deformed strings tend to produce "beats" that constite a kind of tremolo effect that you don't hear that much but which produces triggering in the gate circuitry.

The CE-1 uses two cascaded "filters" for the envelope signal and is set up for a relatively fast decay. Look at the schematic and you'll see that diode D11 is followed by a .033uf cap to ground and a 10k resistor and .047uf cap to ground.  That 10k resistor and .047 cap form a lowpass filter that smoothes out the envelope ripple a little bit more but not all that much (6db rolloff at 338hz).  This is followed by a 330k resistor to ground.  This resistor allows for whatever charge is stored up in the .047 cap to bleed off to ground once the incoming signal stops/decreases.  The thing to bear in mind, though, is that caps can be anywhere within a +/-10% and sometimes 20% tolerance.  Standard resistors are within a 5% tolerance.  The amount of play that offers probably allows, once in a while, for a unique combination of cap and resistor to yield a decay time which is too fast for its own good.  Sadly, as much as those of us who diddle with envelope-controlled devices know that  finer control over the circuit is necessary to get it to comply with our playing style, the CE-1's envelope-follower is fixed in all its parameters, with no trimpots, or anything adjustable.

Now, this is what we *think* may be the issue, after a phone discussion, and without having the pedal in hand,  It's being shipped back for troubleshooting and we'll see how right we are in due time.  It may well be that the smart thing to due in future production runs is to incorporate a trimpot in place of that 330k resistor and just the decay time at the pointof production to minimize false triggering.  Given that it's not in the audio path per se, it will have no effect on the tone of the pedal.

What's interesting about this "adventure" is that:
a) the customer doesn't know enough to describe the problem in a precise enough way
b) the customer doesn't know enough about how the pedal works to identify a possible source of problem
c) we have absolutely no idea of the signal conditions
d) somethng learned from a completely different device (noise gate) turns out to be applicable to a another one (chorus)
e) component tolerances, especially when the design dances at the edge of what's feasible, can be important
f) never underestimate the value of trimpots

[petemoore]

Sequel...
 I mean, was a winner of this 'bout' ever definitively determined as such.
 Well, whatever the case there, your adventure Has taught us a thing or three about how to better prepare 'our contenders and participants'!!!
 Whoever decided to touch leads to different points across a resistive element, and thereby discovered the associated variable resistance...should have a national holiday in honor of this great achievement !!!