Could a DF-2 be turned into a cheap harmonizer?

[Mark Hammer]

Watching a demo from Andy Martin of the Beetronics Swarm pedal, which would seem to make use of a 4046 PLL chip or two ( https://www.youtube.com/watch?v=7sBj_92Owks ).

It got me to thinking, could a Boss DF-2 Feedbacker-Distortion, which also uses a 4046 to produce octave-up that fades in, so as to mimic feedback, be coaxed into reliably producing other intervals (one at a time)?

I will note that the Beetronics Swarm coughs up other intervals immediately, while the DF-2 requires that one hold down the foot treadle in order to fade in the simulated overtone.  But fading in, say, a 4th or 5th, on demand, wouldn't be a particularly terrible thing.  I will also note that the DF-2 feeds a somewhat shallow modulation from an LFO to the 4046.  What I will freely admit to, though, is that I have no idea how one would create some sort of pitch offset for the 4046.

Whaddya think?

[denden]

Last time I used a 4046PLL was in a frequency counter I built some 30 years ago? What I remember is that you get a frequency out based on the 'divider' you use. Simple flip-flop divides by two, so you get Freqx2. You can get freqx4 by simply dividing by 4 (two flip-flops will do it). In my Freq counter I wanted to get the frequency displayed quick (0.1sec) so multiplied the input x10 (used a decade counter for divider).

So you could setup a number of dividers and select via a rotary switch easy enough. The only other issue I recall is that you may need to tailor the associated filter to the divider. Hmmm, in a guitar pedal that may not be as much of an issue. I was going for 10Hz to 20KHz range which meant the filter was 'broad' and took longer for the PLL to lock on the frequency and give the correct output. I had to incorporate a lock detected LED to make sure I was at 10x.

Hopefully someone more versed will come along ...

[anotherjim]

Several octaves alone are not all that interesting! Sequencing them would be better.
A 4017 decade counter can be used as the divider and the basic multiplier then depends on the count it resets on. That can be selected by switch or sequenced using a CMOS mux chip off a binary counter. The frequency range output can be much higher  - too high probably, but it can always be reduced with binary dividers. The point is that you can then get odd numbered multiples.

[garcho]

Cool post Mark. I recognize 1B is a non-inverting amplifier, 2B is a sallen key LPF (to shave off problematic harmonics and rumble for the rectifiers?), 3A and 3B look like rectifiers for both "sides" of the AC signal, but IC4 throws me for a loop, as does the flip flop. What's going on in that corner?

Cool schematic symbol for the analog switching, I've never seen that before. I glanced through the 4066 datasheet but didn't see it there.

Man, look at all that stuff just to turn the damn thing on. There's a whole fuzz pedal's worth of components in the switching logic. Of course, for the engineers at Boss, that was as simple and elegant as a voltage divider, and allowed them to use a cheap little plastic push button under that tank armor foot pedal. I know it didn't turn out to be the perfect solution (ye olde tone sucking, etc.) but still, as far as engineering a solution to the foot operated audio device, it's brilliant problem solving!

[Mark Hammer]

I gather another potential mod to this largely overlooked pedal would be some means to alter the rate of feedback onset.  Stock, it takes a few seconds from the point of treadle-press to when the simulated overtone begins (i.e., becomes audible).  I'll be a monkey's uncle if I could point to how to alter it, but I'll assume that if it isn't instantaneous then a cap or two somewhere is responsible. Being able to vary, even if it was just 3 rate settings selectable via a toggle, coupled with some simple means to change the PLL second note generated, would make for a much more interesting pedal.

[denden]

Ok, the flip-flop 5B sets up a simple divide by 2 into the PLL to get 2x the input freq. The loop filter which determines settling time and damping are made up of R59, R60 and C37. Pretty much stock based on CMOS cookbook, but the C37 is twice what the book starts at. Also and I think part of the big reason it takes so long for the note to kick in is the switch 8a between the two resistors effectively opening the loop. So when the effect is switched in we need the loop to stabilize before we get our tone. Just guessing here ...

Also normally the output frequemcy would be tapped off pin 4 of the PLL, what it appears they have done here is tap off the filter loop via C36 and R62.

If you want this thing to turn on 'quick', maybe leave 8a on all the time?

[anotherjim]

One thing they did was make the pll outputs only available after a delay to give it time to find and lock to the signal. Without that delay, it might get messy.

x2 square wave outputs could be made more interesting. The output from 5b can be made 25% pulse. I've an idea it might cost x2 diodes to do. Make a 2 input diode OR gate feeding into R71.

Choosing the OR gate (it would also work with AND), the pull-down resistor function is taken by the path through R71 and Q12 base - ideally, since the base will have to "float" off when both diodes are off, a pull-down resistor should still be used from the diode cathodes to 0v of 10k or more.
One input will be from 5b pin11 and the other either from pin12 or 13. One or the other might work best when the 2 octaves are mixed by the Overtone pot. You can either have 25% or 75% pulse which, on its own, either duty cycle should sound identical.
C41 and C44 could be made smaller for a high-pass filter effect that can make the pll waves a bit more sawtooth string-like.

I wouldn't want to make a clone of something so complex completely by scratch, although if there was a pcb available it could be one of those long  projects you can build up a bit at a time. Also, unless you can get a 4046 of the same DNA as the originals, a clone might not work as well.

[garcho]

I wouldn't want to make a clone of something so complex completely by scratch, although if there was a pcb available it could be one of those long  projects you can build up a bit at a time. Also, unless you can get a 4046 of the same DNA as the originals, a clone might not work as well.

9 ICs and 15 transistors, and searching for more options in pitch... maybe this is a great candidate for DSP?