Guitar synths really, really need a good, reliable hex pickup.

...or some serious DSP to identify the notes coming in.

If you use a compressor in the detector you don't need a hex pickup for it to track well on *single notes*.

If you want it to track on chords the hex pickup helps but then you have to decide what
to string to throw away for a monophonic system, or, have a polyphonic system which is
far more complicated and getting beyond any reasonable sized analog circuit.

I don't use a compressor or any form of signal conditioning other than an opamp to get the guitar level high enough to give a decent signal to noise ratio. Tracking is pretty good, but I use a couple of tricks - sample a new note within a couple of cycles to get the frequency, then lock this for the note duration and apply hysteresis to the decaying/new note to give a clear start and end. For hammer-ons and slides, get the frequency and continuously track throughout the note duration, but limit note changes to within a bracket of 4 semitones to exclude jumping to harmonics. The limitation is in quantizing notes, but this is mitigated by applying glide to allow semitone-interval bends. Other than that, bends can be applied using an expression pedal (whammy style).

To sidestep the chord issue with my mono synths I generate the appropriate chord for each note depending on the key and this works well enough when mixed in with a regular guitar signal to give a rich backing. The ability to play a note, generate the chord on the fly and then be able to hold it while playing over the top is quite inspiring.

A basic polyphonic synth with a hex pickup doesn't need to be complicated. I started going down this route a few months ago but got sidetracked away from pickup design into condensing my previous mono builds into as small a unit as possible. I thought that instead of parallel processing there may be the opportunity to multiplex a hex pickup, generate 6x oscillator signals and then apply a common envelope shape and filter to the overall combined output.

All a bit too much to do exclusively analog, but more straightforward if the upfront signal detection is done digitally and 6x 1v/octave outputs used to control analog VCOs, VCAs, and filters. Because small processors are so cheap and easy to use it's feasible to parallel process on the basis of one per string, though I'm thinking that this could be reduced to two or three (say) postage-stamp sized RP2040 Seeed boards or similar.

Quote from: Mick Bailey on May 14, 2025, 04:32:04 AMA basic polyphonic synth with a hex pickup doesn't need to be complicated. I started going down this route a few months ago but got sidetracked away from pickup design into condensing my previous mono builds into as small a unit as possible. I thought that instead of parallel processing there may be the opportunity to multiplex a hex pickup, generate 6x oscillator signals and then apply a common envelope shape and filter to the overall combined output.

All a bit too much to do exclusively analog, but more straightforward if the upfront signal detection is done digitally and 6x 1v/octave outputs used to control analog VCOs, VCAs, and filters. Because small processors are so cheap and easy to use it's feasible to parallel process on the basis of one per string, though I'm thinking that this could be reduced to two or three (say) postage-stamp sized RP2040 Seeed boards or similar.

A "hybrid synth" approach would be possible too - get the digital side to generate the note frequencies, and then feed that to analog VCF+VCA. The digital side can also produce the required envelopes or LFOs or whatever to keep the circuit component count down.The only parts that are really analog is that output path, with just a handful of CVs to drive it (Cutoff, Res, Amplitude). This is a pretty standardised design ever since things like the Matrix/Xpander did it with VCOs and the Prophet VS did it with digital oscs.

Thanks for sharing the details, BTW. I'll have to have a good dig through that, but that's a fun job!

My previous build was a digital front end to generate the fequencies, then an analog VCA, envelope generator and filter. I just used a gate pulse from the processor, but it would have been better to have done the EG in code as in my latest build, but output a PWM envelope signal. Outputting  other control voltages as you suggest makes sense.

The single IC Monovoks filter is super simple, but really effective. I thought it would make a good standalone filter pedal as a guitar effect

I still say the best way to get the "gesture capture" information is to not lose it in the first place.

You know what fret you are playing on which string and with a hex pickup, you can get an amplitude for each string.

Making frets with six insulated conductive sections sound a bit easier and more reliable than mixing the signal together then trying to pull apart all the components of it.  Just multiplex them and send the six string amplitude outputs to the synthesizer and you have no ambiguity and minimal circuitry.

if you stick with old-school monophonic, the other issue is that monosynth leads generally sound best when someone is tweaking the knobs with one hand while playing a keyboard with another. that's tough to do when you need both hands to play guitar. sure, an expression pedal would help, but it's a pain to map one to multiple parameters on an analog synth.

i have tried this, plugging my guitar into a pitch and amplitude tracker to control a simple modular synth. the results were always a bit boring, a bit 80s prog in a bad way.

so it basically always makes more sense to do it with DSP, which will give you better pitch tracking, and polyphonic pitch track, thanks to FFT processing.

Quote from: amptramp on May 15, 2025, 03:36:28 PMMaking frets with six insulated conductive sections sound a bit easier and more reliable than mixing the signal together then trying to pull apart all the components of it.

Thomas Organ manufactured and sold a guitar with separated fret sections back in the 60s. It had a simplified electric organ inside the guitar. Strange but true.

Quote from: R.G. on May 15, 2025, 07:01:09 PM

Quote from: amptramp on May 15, 2025, 03:36:28 PMMaking frets with six insulated conductive sections sound a bit easier and more reliable than mixing the signal together then trying to pull apart all the components of it.

Thomas Organ manufactured and sold a guitar with separated fret sections back in the 60s. It had a simplified electric organ inside the guitar. Strange but true.

Exactly what you are looking for and the Thomas organ with its bandshaped square waves would be good for a clarinet but not so good with other voices.  We have the technology to take an input and provide an output which is any instrument you want, including instruments that don't exist in nature.  This may be the right guitar to start with.

Quote from: amptramp on May 15, 2025, 03:36:28 PMI still say the best way to get the "gesture capture" information is to not lose it in the first place.

You know what fret you are playing on which string and with a hex pickup, you can get an amplitude for each string.

Making frets with six insulated conductive sections sound a bit easier and more reliable than mixing the signal together.....

That doesn't sound easy to me. Fretting a guitar is one thing, placing six insulated sections, wiring them up and then making sure the whole thing doesn't separate with timber movement looks to me to be very involved.

QuoteThat doesn't sound easy to me. Fretting a guitar is one thing, placing six insulated sections, wiring them up and then making sure the whole thing doesn't separate with timber movement looks to me to be very involved.

But not impossible if Thomas Organ managed it in the 60's. Then again, I can't find any pictures of it so it clearly never caught on.

Is the idea of using isolated fret sections to pass a current along the strings in order to detect which section is shorted? 24 frets x 6 strings = 144 cables (or 72 if you limit it to 1 octave per string) Ribbon cables? Two covered channels in the back of the neck, one either side of the truss rod - should be room...
Access under each fret for cable connection should be possible either with careful CNC work or, in my case a ruler and chisel

Then, there's the electronics - each fretted section is known but, what to do when a fretted string is bent, going from one section to another? Actually, as I type that, I think I know. And, it would be different for each instrument. What to output? MIDI?

Of course, I'd still have to learn how to play the other instruments parts.

I'd better stop thinking about this now. It's time I took my meds and return to the other rabbit hole I've been down for too long...

The Thomas guitar organ was a right nightmare to work on. The frets each had a wire running down inside the neck to the circuits. The grounded strings contacted a fret section in what amounted to an organ key switch. Thomas had a lot of experience with organs, so they didn't much blink at the amount of circuitry involved.

Thomas also put effects of various types inside other guitars. One of my many rabbit holes was digging out the on-board effects circuits. I have repro data on all the ones I managed to find.

Thomas typically had really-forward vision on guitar effects and amps. They were, after all, the origin of the wah pedal. They did a lot of clever stuff that fed into the music industry, but never became the main stream. This has to be tempered by Thomas not having a clear view of how musicians would actually use guitars and amps. Their onboard-effects guitars and amps just kind of missed.

Any of you folks ever worked on one of those Fretlight guitars?  Strikes me that the terrain under the fretboard would be as challenging as those Thomas/Vox guitorgans.

Quote from: R.G. on May 16, 2025, 10:16:19 AMThe Thomas guitar organ was a right nightmare to work on. The frets each had a wire running down inside the neck to the circuits. The grounded strings contacted a fret section in what amounted to an organ key switch. Thomas had a lot of experience with organs, so they didn't much blink at the amount of circuitry involved.

That might be true, but these days we could do it with a scanned matrix and use a lot less wires. We might not even need divided frets if we don't have to power them all simultaneously.

24 frets might be three 3-to-8 decoders to power them up. That's six lines from the uP, three address lines, and three chip selects. The decoded lines are 24 wires up the neck, although most of them don't go even half way (the gaps between frets are bigger closer to the headstock, remember). On my LP clone, the highest 12 frets are barely more than 1/3rd of the neck, so only 12 wires run up the upper 2/3rds. Then you read 6 bits of a byte from the strings to see which ones are touching it. That's one input port on your uP, of which you're only using 6 bits. These days, that's not even a run around the block to stretch the legs, let alone a full marathon.

So, 12 I/Os and a few decoders. No divided frets needed. Thomas Organ didn't have cheap microcontrollers to work with!

It seems like an interesting idea.

Hagstrom Patch 2000 seems to have done it with no divides frets, perhaps first. Watched some Dick Denney vids and jennings AC30 story (where they kicked AC30 down the stairs). Teisco Del Reys Barney Kessel bodied guitorgan is coolest looking imho.

And what about bass guitar synth? My wifes sisters son is talented pro bass player but he can play keys too so he played syn bass parts with keyboard.

EDIT Looks like that Hagstrom is not poly

How I thought divided frets could be done is treating each string as a keyboard bus and the frets as contact points on a constant-current resistor string. Just 6 wires needed, though the frets would need connecting with maybe 1/8W precision resistors, all the same value.

I built Ray Wilson's keyboard controller that uses this method and it works perfectly, though being analog is more involved than using a processor.

A gate pulse also needs to be generated for each new note.

I thought a detachable phenolic fretboard as used in the Bond guitars would make the electronics accessible. I'm no stranger to electronics in necks and the though of a post-assembly failure has always been a real concern.

One reason for not wanting to go down this path anyhow is that it isn't a DIY project that most people would want to undertake, so developing the electronics would probably be a dead-end after I'd built a one-off for myself. I much prefer to develop electronics that anyone who can build a stompbox could replicate.

Even winding a hex pickup would be a challenge for many, but a hall effect or optical pickup could be assembled on prototype board and held in place without modifying the guitar.

People have mentioned DSP and FFT, but I can't find any DIY examples of a successful polyphonic synth build. My own experiments with FFT haven't produced anything remotely usable. If anyone has a practical working synth that can accept a regular guitar signal, I'd be really interested.

My own method of capture is based on taking measurements of the first full cycle of a signal. This means that the overall acquisition time is based on the period of the lowest note. With a standard tuned 4 string bass the period is doubled, so latency becomes noticeable. I do have ideas to reduce this by detecting the frequency based on a half cycle, but it's much more prone to error.
 

Boss's synth pedals are polyphonic I believe and they accept a regular guitar/bass signal. Rip them off I guess.

A design using a microcontroller is probably the most feasible/affordable. I've seen projects for taking the Ray Wilson type of keyboard and converting the output to midi.

Is a 'frequency detection' circuit similar to a 'frequency to voltage converter' circuit? Just throwing that out there in case it sparks an idea, as all this is way over my head.

Wouldn't it be cool if Boss sold a guitar->CVs+gates chip that we could use like the ubiquitous PT2399 and FV-1. 

Quote from: MaxPower on May 18, 2025, 07:25:43 PMIs a 'frequency detection' circuit similar to a 'frequency to voltage converter' circuit? Just throwing that out there in case it sparks an idea, as all this is way over my head.

For me frequency detection is just identifying the frequency and then it can be used to generate a voltage, midi note, or used to look up any other value from a table.

A couple of years ago I developed a cheap and simple Arduino guitar frequency to voltage converter that uses a DAC to output a 1v/octave output as well as gate and trigger pulses and I have this built up as a modular synth module.

https://youtu.be/nA1bWnuWTjQ?feature=shared