E&MM Harmony Generator query

[Mark Hammer]

I have a fully-populated board for one of these, and was getting ready to finally stick it in a chassis.  Since I wanted to make sure I had done all the legending that was going to need to be done, so that I could clear coat it, I wanted to verify something that I noticed while staring at the schematic over lunch.

The HG produces the additional note/harmony by switching a rectified version of the signal inand out with a CMOS switch.  It first synthesizes a higher octave version of the note, and then divides that down appropriately, so as to be able to derive pitches that are above OR below the original fundamental.

If I look at the schematic, though, the CMOS dividers act on the switch in a binary fashion, turning it fully on or off.  The op-amp stage that follows the switch has no treble cut.  Indeed, there is nothing whatsoever to remove any objectionable treble from the signal at any point after the CMOS switch.

What I was thinking of was turning the 220k input resistor to the op-amp (R6) into multiple fixed series resistors (100k->100k->20k) and running caps to ground after each 100k to lowpass filter the added note and make it a little more melodious.  I figured using a 3-position switch, with filter settings (cap values) selected to optimize notes well above, near to, and below, the original.    That may require upping the value of R7 to get not quite so much attenuation out of IC3, for a better signal balance.  Of course, if this is entirely unnecessary, I won't bother.

So, the query to anyone who has already built theirs and fired it up is: did it strike you as being a bit harsh at all, and in need of tonal control on the synthesized portion?

[artifus]

i haven't looked at the schematic for a while but wouldn't (heavy) filtering occur before cmos to ensure cleaner tracking, the output being a 'synthesized' square wave? and square waves being as they are some filtering may well be appropriate to soften the output. i would certainly consider a simple resonant low pass for tonal shaping. i'd rather have it and not need it rather than find i need it and not have it.

*edit* i should add that i haven't actually built one but have looked at the schematic a few times, am guessing and may well be talking out of the top of my hat.

[Mark Hammer]

I was aiing for the least invasive mod, that could be easily made to work on the existing board.  The path leading up to the switch doesn't offer much opportunity, but the op-amp stage after the switch shows much greater promise.

[AdamM]

I built one about 20 years ago and it never got onto my pedal board for exactly the reasons you're describing - that synthetic square wave sound which didn't blend well, even when it was harmonising an overdriven/distorted tone. I often thought about some low pass filterign as being an option to try but lost interest before doing so. I also found it was not brilliant at locking - but I never tinkered with optimising the PLL so I don't know if that is an inherent design flaw or not.

I found the still (almost) populated pcb when clearing my garage before moving house recently. I kept it for the PLL....

[R.G.]

This was always one of those clever ones I wanted to build but never got time.

There are a few clever things in there, like chopping the envelope to impress an envelope on what is inherently a square wave from the PLL dividers. Messing with the harmonic balance of what comes out was not one of the clever things they put in.

I can never quit designing long enough to properly enjoy a pedal any more. I kibitz too much with the design.

On the HG design, I have thought of using a secondary PLL which tracks only the selected frequency but uses something like an XR2206 waveform generator to do it. The 2206 lets you choose square, triangle, or sine outputs; you use the square wave for the PLL and the sine/triangle for smoother output waveforms, or some mixer combination of them all.

That doesn't get the neato envelope trick. You could use the envelope from the HG envelope detector to run a VCA to impress the envelope on the chosen waveform.

This could get you an octave up/down/sideways output as a sine or triangle wave.

[liquids]

Mark - intelligent additions to the input of that circuit, for filtering, is really useful.  If you can house it as such with a simple filter such as in the microsynth at the input and THEN go to your populated circuitboard, you'll have much more useful results.  Even still, it's a bit of a toy, and a true re-design can be much more satisfying, but you'll have fun.  With the minor input filtering, it's a beast that becomes frustratingly hard to tame since, when it tracks, the sound is amazing.

As for filtering...square waves are where it's at.  Some filtering can be fun and all...but the way the thing can produce notes +/- 3 octaves makes it, in some cases, useless to filter without a lot of flexibility.

At that, from my spice simulations, no matter how many passive R-C filters you cascade, the effective curve will always be ~6 db/octave, and said filter can be duplicated with one (albiet not mathematically equivalent) single R-C filter.  So the cascading is really not necessary.  I'd turn the 25k/50k/100k pot as a SWTC plus a set resistor.  SWTC works here because the resistance there needs to remain static since it sets gain. Likewise, I'd advise a way to break the connection between the cap and ground especially for +3 octave situations.

[liquids]

This was always one of those clever ones I wanted to build but never got time.

There are a few clever things in there, like chopping the envelope to impress an envelope on what is inherently a square wave from the PLL dividers. Messing with the harmonic balance of what comes out was not one of the clever things they put in.

MarkusW did a creative design that utilized some of this circuit, though it was for for an 'intelligent ring mod' after all.  The bones, pre-ring modulation, is definitely an improvement on this whole circuit, IMO, though it still requires some creative thinking/circuit snipping to flesh it out. And I imagine you could improve it even further...

I can never quit designing long enough to properly enjoy a pedal any more. I kibitz too much with the design.

On a similar note, I seem to adore the endless torture/joy of auditioning every idea and tweak I can conceive of on a breadboard.  To the effect that I rarely flesh out a usable pedal for fear of re-imagining an improved circuit right when I am yet spending weeknights soldering and housing up the previously satisfying finalized design!  

On the HG design, I have thought of using a secondary PLL which tracks only the selected frequency but uses something like an XR2206 waveform generator to do it. The 2206 lets you choose square, triangle, or sine outputs; you use the square wave for the PLL and the sine/triangle for smoother output waveforms, or some mixer combination of them all.

There are some clever ways to re-condition a wave once you get it in sync with the intended note.  Though I find that, with true CMOS logic square wave, the most fun waves are in the area occupied around square, pulse, and saw waves.


I do wish you'd try this circuit, because I find these kind of circuits addictive, and imagine you'd have quite a bit to add.   And I confess that I also love that this kind of circuit is way over the head  () of the fuzzbox-by-number constituency.  

[Mark Hammer]

I am probably walking out from the shore onto the thin ice that is going to break under me, but here goes...

The Boss DF-2 also uses a 4046 to synthesize a note derived from the fundamental of the input signal.  What it does, that the HG doesn't attempt to do is wiggle that VCO with an LFO.  I'm wondering how difficult it would be to whip up a little daughter board and inject a little wiggle into the HG's master oscillator.

As for the advisability of a more sophisticated kind of filtering, I suspect you're right.  As I noted, though, I was aiming for the least invasive approach.  But equally important, while it would be a piece of cake to tack on a wide-range sweepable active lowpass on the output, you really want to preserve as much as you can on the clean signal, and use the filter only to soften the added note, so it doesn't distract too much.  And that means it needs to go before the mixing node.

On the other hand, if one was going to perf up a daughter board with LFO, I suppose you could also include a better filter as part of it.  What I'd really like, though, is a more complex signal, with multiple concurrent side-notes (e.g., octave and fifth), and maybe even some envelope control so that the side-notes don't all kick in or disappear at the same time as the source.

But now I've pretty much walked out past the ice and am swimming in frigid waters. 

[liquids]

I think the SWTC + set resistor subbing for the 220k will work nicely and be fairly low on the 'invasive' scale.  Worth trying, at least.  

I kind of forget what  IC9d does for all practical purposes...but putting that on hold (since it could completely invalidate the following): the nice thing about putting a SWTC in that particular location is that the signal going into 220k resistor should be low output impedance.  If that is the case, than with a reasonably (not huge) sized capacitor for passive filtering, when the SWTC has the least amount of resistance before the capacitor to ground, you will more or less get a RF filter - no perceived effect in the audio realm.  So in that case, there would be no need to have a switch to break the connection between the cap and ground, as it would be redundant if so.  

That said, I have never tried that particular filtering configuration of using the series input resistor to do double duty as a passive LPF in conjunction with a capacitor.  It may be flawless, or it may yield all kinds of unruly results in conjunction with feedback.  A little LTspice simulation would answer my questions right now even, but as of my typing this, I am feeling too lazy to switch computers to simulate it and find out.   If you're curious, say so and I will check it out.

As for the LFO wiggle, it's not unreasonable.  I just personally find it irritatingly mechanical, and, I find the PLLs track man-made vibrato ( ) quite nicely!

What you are talking about with concurrent side-notes is what I have been intending to do since summertime.  Once the fundamental is extracted in the typically fudged way and conditioned so that a PLL can more or less follow it (comparator), you've got a square note coming out...from there, why not feed the square to a CMOS sub-octave or two, and mix the two?  Feed the square to the harmony and/or octave up circuitry too, and mix it with the fundamental and sub-octave?  Why not feed it to a less-crude Escobedo SWSMS/PWM type circuits for pulse, triangle, and saw wave capabilities for a major shift in tonality?  

All that is really another realm beyond the confines of your PCB, but it's a really wonderful sound.  Dan Snazzelle aka loss1234 really maximized some of the above ideas and the PLL in general if your recall; he's a mad scientist that took this area to the moon and back, and has also produced an incredible effect along those lines and way more...I personally only want the  sub-octave and wave shaping, and in addition, an option to feed these evil sounding waves into a resonant filter with the clean signal as an enevelop or trigger.   So the harmony capabilities are really a digression.   However, I was working on these ideas for weeks and weeks, first striving for the ULTIMATE fundamental extractor, never getting beyond the PLL square wave because IMO the rest is easy once you have that.   And yes (stephen), I even built the oft-touted EHX fundamental extractor!  Sadly, I never reached the perfection I imagined and dreamed of with tracking, and walked away with my tail between my legs....this post has brought me back to the joy of the sound and fun of PLL triggering, even with occasionally glitchy-tracking.  The EM&M circuit can be either great fun and no more needed for some; likewise it can be so cool and yet frustrate you with it's tempermental nature and whet your appetite for something less impossible to tame.  If you finish the harmony generator and want to start over with a clean piece of perfboard, let me know.