harmonic engine guitar pedal

[mikeford]

Hello! Here is a circuit that may be not suitable for stomp-box material, but having built it and plugging up my guitar and running it to my amp, IT SOUNDS KILLER! The PCB is from https://www.circuitbenders.co.uk/forsale/harmonic/harmonic.html
Here is the schematic:

I am wondering what sort of bypass switching I should implement: Millennial? I tried a bypass switching that I could not recall,as I put it on the shelf, because whatever it was, it was quite noisy 9Clock noise) whilst the true bypass switch was engaged. Any thoughts? THANKS!

[R.G.]

Looks like an update of the E&MM "Harmony Generator"

[mikeford]

YES! it is. It sounds awesome.

[Marcos - Munky]

Could you post some sound samples? There's some people here in the forum which are interested in how this circuit sounds with the guitar - me included .

[PRR]

Does anybody follow links?

"The circuitbenders.co.uk Harmonic Engine PCB is a clone of the E&MM Harmony Generator."

No deception attempted; and this mouth-watering description: "chaos, with all kinds of bizarre gurgling digital squeaks and squeals vomiting forth."

And there are demo-videos ON the site which was linked.

[FUZZZZzzzz]

I bought the pcb about a year ago, but still havent build it. Thanks for the reminder!!! I think I had some trouble scoring some of the chips used, but that just as well been another project

They also sell a pcb for a clone of the Maplin Voice Vandal. Could be a nice twin project
https://www.circuitbenders.co.uk/forsale/phonictaxi/ptaxi.html

[mikeford]

I got to figure out how to kill the excess (clock?)noise of the circuit while it is in  bypass mode............ shielding the in/out jacks?

[Marcos - Munky]

Well, it indeed sounds awesome!

[R.G.]

Every octave generating method that creates a digital signal for the output will need some way of introducing an envelope mimicking an analog signal of some kind. Impressing the envelope of the original signal ala the Rocktave is one way, using the synth method of an attack-decay-sustain-release to force an envelope is another.

There are bound to be misfires and glitches, so you will need a gate of some kind to silence the blasted thing between actual notes in the original analog signal. Well, OK, unless you just like the noise it generates.  There will be clock noise in any circuit that uses a clock. Accept that it's going to be there, and design to minimize it. Gating helps hide it, but careful layout forces it down into the (analog) noise.

Generating octaves down by dividing with flipflops gives you limited harmonic content. A square wave contains the original frequency, then the third harmonic, fifth, and all the rest of the odd harmonics. If you want more flexibility, you would probably like the sound of introducing some even harmonics. Getting these involves figuring out how to generate them from the squared-up analog signal.

One good way is to use a phase-locked loop (PLL). PLLs lock an oscillator to the frequency of an input signal. A very common PLL chip for music use is the CD4046, and its successors, which you'll see in the HEGP. If you feed a PLL a frequency F, it forces its output oscillator to be the same as F. But you can play tricks on PLLs. If you feed the output oscillator to a digital flipflop, then use the divided down output of the flipflop as the "output" going back into the PLL, the PLL will force the divided down signal to be F, and the oscillator of the PLL will be running at 2*F - an octave up.

You can use a string of flipflops, and make a 4046 generate half a dozen or more octaves up. So you can get F (the original frequency) as well as 2F, 4F, 8F, 16F and so on. If you pick an output in the middle of the string of flipflops as the reference fed back into the PLL to be locked, you can get 8F, 4F, 2F, F, F/2, F/4 and so on. So the same circuit simultaneously generates both octaves up and down relative to the original frequency.

These are still square waves, but remember that business about square waves containing F, 3F, 5F, 7F and so on? You now have divider outputs with 2F, 4F, 8F... and you can add up the square wave with its octaves up with resistors and make waveforms with all the harmonics in various mixtures. So you can generate non-square waves with this. You can actually generate true sine waves this way, although that's more work.

Another interesting side story is that you can use non-binary dividers to generate harmony signals. In music, true musical intervals are in the ratio of simple whole numbers related to a basic note. So if a note has a frequency F, you can generate 3/2F, 4/5F, and so on, which are much the same as hitting harmonized notes on the guitar.

This multiply-up-and-divide-down-cleverly approach is what the E&MM circuit did. I have not analyzed the HEGP, but I'll bet that's what those divider chips are doing.

Another interesting thing to do is to use a PLL that can't lock to the incoming signal very quickly. It follows the input signal up and down, but slides up or down to it. It glides, like slide guitar, but electronically. I haven't seen a pedal that is set up to do this yet. But someone has to be doing it somewhere.  You can play the same tricks with this PLL I mentioned earlier - non-square waves and harmony.

The ideas here are straight from electronic organ and synth theory. They're a lot of work, probably more than any pedal hacker has the patience to go do. But there is a lot that can be done.

[thermionix]

But he's getting clock noise in bypass.  Is it true bypass?

[R.G.]

Could easily be. "True bypass" doesn't mean you get no noise problems.

"True bypass" means only that the input and output of the pedal don't load the signal line when it's "bypassed". It says nothing about the signal lines picking up capacitively or inductively coupled noise. Free space lets signals leak out of wires as radio waves, electrical fields, and magnetic fields. The further from DC the signals are, the more easily they leak out of wires, especially as electrical and radio.

Digital switching noises are by definition very, very high frequency. A vertical rise or fall of signal has components well into the many MHz. These fast edges leak out easily. It takes good power supply design and good layout to keep the leakage in the copper and away from sensitive components.

[thermionix]

"True bypass" doesn't mean you get no noise problems.

Right, but (in my mind anyway) makes it more likely that shielding input/output wires will help eliminate the noise.

[R.G.]

Not necessarily. A buffered output that is always driven, even in "bypass" with the clean signal is lower impedance and much more immune to noise. Boss, Ibanez, even the Univibe, all work this way.

[R.G.]

I just jotted down some thoughts on the octave up/down generation, and put them up at geofex. It's here:
geofex.com in the first link in "new".

[bean]

Oh boy thanks for posting this. I cannot resist building this thing.

[mikeford]

WOW! Thanks all!( R.G./Thermonix) I suppose this circuit is not "stompbox friendly", but it just SOUNDS so cool!
Brian, i have SEVERAL of your pedals on my board.all killer! Can't wait for your FV1 projects to come out.
I have built the Phonictaxidermist  circuit also: https://www.circuitbenders.co.uk/forsale/phonictaxi/ptaxi.html







It has a bit of 'noise' issues,but is a cool,non bbd chip  based echo/delay/bitcrushing/ring mod sounds also!

[BetterOffShred]

"True bypass" doesn't mean you get no noise problems.

Right, but (in my mind anyway) makes it more likely that shielding input/output wires will help eliminate the noise.

Wonder if true bypass with input or output ground would fix the bypassed noise..  worth a try.

[Marcos - Munky]

Or just use 2 poles of the 3PDT for the bypass and the other pole to switch off the power.

[BetterOffShred]

I get popping whenever I try that technique.. I kind of gave up

[anotherjim]

Freppo @ Parasitstudio developed some 4046 based pedals with portamento/glide. Many liken it to a Theremin effect.

The old EDP Wasp synth, which is always worth studying for its unusual design, used 4046 purely for the portamento. The pitch was made by a DCO so it didn't have a pitch CV with which to apply the usual way of slewing pitch change for portamento. A trick is to use the 4046 VCO control voltage as a key tracking CV for a VCF. In the normal use, if the 4046 loses input, the VCO pitch rapidly drops to 0hz and the CV to 0V. A VCF controlled by that CV also drops to complete cut-off, so it acts like a VCA.

In the case of something like the Harmonic engine, I've considered using a 4046 with loop counters and run that part always at higher multiple frequencies. Follow it with the Wasp idea of another 4046 for portamento  (which might smooth out tracking of the first 4046 some), and then use dividers with wave shaping for the usual complement of synth voices.