CLK input on counter/divider logic ICs

[WhenBoredomPeaks]

We all know a lot of octavers work by sending a guitar signal to the clock input of an IC (4024, 4520 etc.) and then we get the divided output in the form of a square wave.

What is an optimal "format" for that clock input signal? Simply lowpass the guitar signal? Or lowpass it steeper with an active filter? Boost it before to get clearer separation between 1s and 0s?Or convert it to a square wave via clipping because that resembles a "real" digital clock signal better?

Is the difference between CMOS and TTL matter? (since they handle the logic levels differently maybe they would track differently too)

[AdamM]

clock inputs (in fact most logic inputs) like fast & clean edges, so I'd say square-wave and maybe even push it through a schmitt trigger to get clean/fast edges. Never built one though, so have no direct experience of these types of circuit to refer to.

[greaser_au]

A fast risetime is what is required for clean operation. A schmitt trigger is the tool for  the job generally, unless you want switching unreliability (which in itself may be interesting in this context).

david

[WhenBoredomPeaks]

clock inputs (in fact most logic inputs) like fast & clean edges, so I'd say square-wave and maybe even push it through a schmitt trigger to get clean/fast edges. Never built one though, so have no direct experience of these types of circuit to refer to.

A fast risetime is what is required for clean operation. A schmitt trigger is the tool for  the job generally, unless you want switching unreliability (which in itself may be interesting in this context).

david

You mean putting my guitar signal through a schmitt trigger and connecting the trigger's output to the Clock pin of the divider?

Does my guitar signal need any of the treatments i described in the first post before putting it through a Schmitt trigger?

Is the 40106 will be good for the job?

[slacker]

There's really two separate things to concider. In order to drive the clock inputs you want a signal that resembles a logic clock, so you want to convert the guitar signal into something like a square wave, that gets close to the positive and negative supply. You can do this by boosting and heavily clipping the signal, look at some of Tim Escobedo's designs for various simple ways of doing this. The other way is to boost the signal and feed it through a comparator to square it up, you make comparators using opamps or use something like a CD40106, again see Tim E's stuff for an example.
That's the easy bit, the trickier part is making this square wave track your guitar signal. This is where the low pass filtering and other techniques come in, the idea being to filter out the upper harmonics so that you're left only with the fundamental. If you search on here for fundamental extractor you'll get a lot of info and opinions about this.

[WaveshapeIllusions]

I like comparators for this kind of thing. An opamp can be pressed into service as one, but it won't be as good. A dedicated comparator usually has some degree of hysteresis, which helps limit noise and false-triggering.

You will also want a certain degree of filtering before the comparator. I'd pick a corner frequency a little bit above the highest note you'd consider playing into the octaver. For bass, I picked something close to 300 Hz. On a guitar, maybe 600-700 Hz would be reasonable IMO. I think that'd be somewhere around the high D on the G string.

A 2-pole filter is probably the minimum, more wouldn't hurt. As long as the signal doesn't have any extra zero crossings, you should be good. If it crosses more than once it might glitch out since it'll look like a higher frequency clock. As long as there's just one rising edge per cycle it should be good.

Compression may help as well. Mostly for length of tracking. A comparator should put out a square wave of the same amplitude regardless of the input level (below the max input limits and above hysteresis of course) so the clock will be consistent in that respect. It will cut off rather abrutly below a certain level though, so compression will help extend the clock signal a bit. If you want to get fancy, the envelope signal can be used to add dynamics to the divided signal. Then it will be fading into silence by the time it start to sputter and glitch.

Hope this helps. I'm by no means an expert on this topic, but I've learned a bit from everyone around here.

[WhenBoredomPeaks]

I added Tim Escobedo's PWM before the clock input of a 4520 and i got good octave downs when i cranked the volume on the PWM.

On the other hand the Escobedo PWM's tracking is pretty far from flawless, the 386 just boost the signal like hell before going into the Schmitt trigger.

Could i improve it's tracking by filtering the highs on the 386's output? I don't even need that much gain probably.