Infinity Polyphonic Sustain System

[Cycfi]

Hello Everyone,

If you've been following the in this forum, it's very much alive and kicking. I avoided posting marketing related ads as proper etiquette dictates so recent activities may seem sparse in that thread. On the contrary, the project has matured and we've begun limited production last April.

Anyway, as mentioned in the hex-thread, "the goal is to have hexaphonic sustain drivers as well. That, and with extensive processing for each string, will give us musicians full control over the dynamics of the guitar. I know hexaphonic sustain has been done in the past with the Moog guitar, but that was a very expensive gear. I want something more affordable. And I want a system that can be adapted to just about any guitar. This IMO is the holy grail and I know this is very difficult to do right, but every journey starts with the first step."

So... we have a New Project! We have a very early prototype of the Infinity Polyphonic Sustain System. Here's the link to the short demo:

http://www.cycfi.com/2014/05/to-infinity-and-beyond/

Again, I would very much love to hear your thoughts and gather ideas while the project evolves!

Like before, this will be an Open Source Hardware project, all the designs (schematics, PCB layout, software, bill of materials, CAD drawings and source code) will be freely shared, 100% free.

[curtisa]

Fantastic! Do want!

[Prostheta]

I look forward to it also! The video makes it sounds very natural most of the time.

[Cycfi]

I look forward to it also! The video makes it sounds very natural most of the time.

In that short demo, the MCU controls are bypassed. The MCU should be able to control the harmonics of the driver to make it as natural or as synthetic as you want depending on how much injected harmonics you allow to feed into the strings. I'd love to feed it other sounds as well, like say, bow noises :-)

[Prostheta]

There's scope for a lot of things when the system is open and able to be tinkered with. For example, a pressure-sensitive ribbon controller might be interesting to include in there somewhere. Pressure controlling the level of feedback and the X axis a filter or blend from one harmonic profile being fed into the strings to another....this could get insane.....

Given that this is not a monolithic project you have all the opportunity in the world to patch in different things happening between the pickups and the eventual drive signals.

[Cycfi]

There's scope for a lot of things when the system is open and able to be tinkered with. For example, a pressure-sensitive ribbon controller might be interesting to include in there somewhere. Pressure controlling the level of feedback and the X axis a filter or blend from one harmonic profile being fed into the strings to another....this could get insane.....

Given that this is not a monolithic project you have all the opportunity in the world to patch in different things happening between the pickups and the eventual drive signals.

Exactly! That's similar to what was just discussed in the comments section in that link. Yes, that's why I want it open. I'm extremely stoked about this. I'm more excited than ever before!

[curtisa]

In that short demo, the MCU controls are bypassed. The MCU should be able to control the harmonics of the driver to make it as natural or as synthetic as you want depending on how much injected harmonics you allow to feed into the strings. I'd love to feed it other sounds as well, like say, bow noises :-)

It's certainly a great start. Very interested to see where this heads.

Does the system rely on your hexaphonic pickup or can any pickup be used alongside your "driver"?

I notice that there's still a little bit of superimposed overdiven signal in the sound (eg, 0:52 - 0:58 in your video). This was something that seemed to be completely unavoidable in the old DIY sustainer that a lot of us here at Project Guitar built years ago. There were many theories behind the cause of this at the time - proximity of the driver to the pickup, driver amp being overdriven, driver being overdriven, low-tech of driver circuitry, induction/coupling of driver signal back into sensing pickup etc - but no real fixes. Is one of the areas of your project to tackle this "interference" in the sustained signal and make it operate as cleanly as possible? The hour or so I spent playing a Moog guitar at a store about a year ago seemed to imply it was possible to do completely clean sustain.

[Cycfi]

In that short demo, the MCU controls are bypassed. The MCU should be able to control the harmonics of the driver to make it as natural or as synthetic as you want depending on how much injected harmonics you allow to feed into the strings. I'd love to feed it other sounds as well, like say, bow noises :-)

It's certainly a great start. Very interested to see where this heads.

Does the system rely on your hexaphonic pickup or can any pickup be used alongside your "driver"?

I notice that there's still a little bit of superimposed overdiven signal in the sound (eg, 0:52 - 0:58 in your video). This was something that seemed to be completely unavoidable in the old DIY sustainer that a lot of us here at Project Guitar built years ago. There were many theories behind the cause of this at the time - proximity of the driver to the pickup, driver amp being overdriven, driver being overdriven, low-tech of driver circuitry, induction/coupling of driver signal back into sensing pickup etc - but no real fixes. Is one of the areas of your project to tackle this "interference" in the sustained signal and make it operate as cleanly as possible? The hour or so I spent playing a Moog guitar at a store about a year ago seemed to imply it was possible to do completely clean sustain.

Yes, it requires the Neo polyphonic pickups. I'm not sure what other way to do it. You'll need a polyphonic pickup to have polyphonic sustain. The Roland GK3 maybe? But the Roland divided pickups are passive with a very weak signal.

Yes, there is the "overdiven" signal. And yes it can be eliminated. As I said, in that demo, the controls are bypassed. In that demo, the drivers are actually driven by a synthesized square wave and a class D driver using PWM! It's not an analog feedback system. It's a phase and frequency locked loop with an MCU controlling what gets fed back into the drivers. A square wave gives you that synthy sound. A sine or triangle wave gives you a more natural sound. Combinations of some odd and even harmonics gives you something in between (natural to synthy).

There's also this issue of inductive coupling too, yes. Yet, the hex pickups have a very low susceptibility to such interference by orders of a magnitude. If I were to use a strat pickup, for example, the inductive coupling would have been maddening! That would mean though that the system might not be usable with other types of pickups. I know that you can't use any other types of pickups with the Vo system too due to this interference. I still have to figure out that one. At any rate, note that the drivers and pickups are fully exposed. It may be plausible to constrain the magnetic field of the drivers with some isolation (maybe mu-metal sheets).

Edited May 14, 2014 by Cycfi

[Prostheta]

Not using magnetic pickups would defeat the object of the hexaphonic pickup too :-)

[Cycfi]

Not using magnetic pickups would defeat the object of the hexaphonic pickup too :-)

All it should take is a little nudge. That's what I always say. I think current breed of sustainers are inefficient and use too much power. If you pull at the right moment with just the right amount of force, you can get something to oscillate indefinitely. That's the essence behind sympathetic resonance. With just the a little amount of force, at the right frequency, you can make a bridge collapse, for example.

Why am I saying this? Well, to me, control is crucial. That's what I am developing right now. We've come a long way since last year when I started working on this. At first I tried an analog feedback system just like fernandez sustainer or sustainiac. It worked but was sooo unwieldy when you have to have 6 of them. Analog AGC does not cut it with 6 drivers. Why? Analysing the signals using a scope, I see that the phase is almost always wrong. Why? That's due to the delay between the driver and the pickup. You can use analog phase shifters to get the phase correct, but that will make it even more cumbersome. Because of this, you have to use brute force to get the string to oscillate, and that wastes too much power. That may be OK in a monophonic sustainer, but multiply that by 6!

I've now abandoned the analog approach in favor of a digital approach with an MCU doing the phase and frequency locking and synthesising a waveform that's fed back to the driver. The MCU can do the phase corrections, analyse the envelope of the input and control just the right amount of signal to drive a string to oscillation. I can even make the driver sleep for a few millisecond intervals (output disabled), esp. with the heftier strings that are very easy to pull. That saves energy. The best part about the digital system is that I can feed any kind of waveform back as long as its coherent with the input. Recently, we've tried square, pulse, triangle and sawtooth. Wave tables would be cool, for example! How about samples of bow noises or wind blow noises? How about the human voice? Guitar or Piano samples? That might be cool. And, needless to say, there are no nasty squeals that plague analog feedback systems. It's just pure sympathetic resonance!

[Prostheta]

Do you have any diagrams of how the magnetic field is aligned by the driver? I've always wondered if aligning it better might improve sensitivity. Same as how you wouldn't pick a string parallel to its direction, rather perpendicular to it?

Additive synthesis would be great fun. I miss my analog synths :-(

[Cycfi]

Do you have any diagrams of how the magnetic field is aligned by the driver? I've always wondered if aligning it better might improve sensitivity. Same as how you wouldn't pick a string parallel to its direction, rather perpendicular to it?

Additive synthesis would be great fun. I miss my analog synths :-(

Wow, I didn't think about that. Tell me more! What alignment would make it better? I only assumed that the magnetic pull is perpendicular to the string. Now that I think about it, it makes sense.

[Prostheta]

No idea to be fair, since I don't have experience in this regard. The idea came from how strings are put into motion - a pick or finger tenses a string and releases it whilst a bow induces what, torsional oscillation? I figured that focusing the induced motion in useful directions would be more efficient. Maybe even a pair of drivers for each string, kind of like how a Jazz bass pickups looks, with a 180° phase difference between them?

I don't know what I'm talking about though. Maybe I'm off course here.

[Cycfi]

No idea to be fair, since I don't have experience in this regard. The idea came from how strings are put into motion - a pick or finger tenses a string and releases it whilst a bow induces what, torsional oscillation? I figured that focusing the induced motion in useful directions would be more efficient. Maybe even a pair of drivers for each string, kind of like how a Jazz bass pickups looks, with a 180° phase difference between them?

I don't know what I'm talking about though. Maybe I'm off course here.

Each of our drivers are rectangular and covers at least two strings per string. That way we can track string bends (see the demo). After a few iterations, we found it to be the best configuration (so far). A push pull configuration made sense at one point and we tried it. There's this kind of tug of war pulling the string. Alas, it didn't work out quite well. Maybe we didn't have the proper configuration. Perhaps I'll try again.

Here's a question for those who used the Vo system (Moog): On the Vo system, I notice that his drivers are small and round (like the Neos). I wonder how well it tracks string bends. Does it work well for bends? I only see it demoed on acoustic guitars. Here:

[curtisa]

From memory it worked fine for all "normal" guitar playing styles - bends, tapping, strumming, hammering, pull-offs etc. The only quirk the Moog guitar had was that it gave best results when using strings with a high steel content (I assume because it made moving the strings via electromagnetism easier when there was lots of ferrous content). They even go as far as saying that you should use Moog-branded strings for the guitar, or D'Addario Pro Steels as a lower-cost alternative. The guitar I tried had some generic strings fitted, and its sustainability was subsequently a little weak.

[Cycfi]

From memory it worked fine for all "normal" guitar playing styles - bends, tapping, strumming, hammering, pull-offs etc. The only quirk the Moog guitar had was that it gave best results when using strings with a high steel content (I assume because it made moving the strings via electromagnetism easier when there was lots of ferrous content). They even go as far as saying that you should use Moog-branded strings for the guitar, or D'Addario Pro Steels as a lower-cost alternative. The guitar I tried had some generic strings fitted, and its sustainability was subsequently a little weak.

Thanks. It's still a quirk with the latest VO-96 version.

[elmo7sharp9]

Steel-wrapped (versus Nickel) strings made my AGC-less Bass Sustainer spring into life - it's a significant step up in efficiency.

But it really shouldn't be a requirement.

For 6-string guitar, Nickel-wrapped strings are all that's available in most stores now. I'm guessing that a nickel 9-42 set, in regular tuning, is what most potential users will have on their guitar, and want to continue using.

I'm impressed with Cycfi's "synthesised excitation signal" approach and must ask - have you tried to replicate Moog's "mute" response?

[Cycfi]

Steel-wrapped (versus Nickel) strings made my AGC-less Bass Sustainer spring into life - it's a significant step up in efficiency.

But it really shouldn't be a requirement.

For 6-string guitar, Nickel-wrapped strings are all that's available in most stores now. I'm guessing that a nickel 9-42 set, in regular tuning, is what most potential users will have on their guitar, and want to continue using.

I'm impressed with Cycfi's "synthesised excitation signal" approach and must ask - have you tried to replicate Moog's "mute" response?

Thank you!

Yes, I want to be able to use ordinary strings. That was a requirement from the very beginning. As for sustain dampening, no I won't do it the way Moog did. Moog, and the newer Vo-96 system uses pure acoustic synthesis and advertises zero post processing. In my opinion, that is not necessary. You can do a lot with post processing on individual strings including control of attack and decay. With post-processing, it's easy to chop an envelope to achieve the muted banjo like effect. That's what I intend to do. An advantage of this approach is that it is simpler, requires less power, and does not require special strings. You only need to get the string sustaining, plus introduce some harmonics along the way. There's so much potential in polyphonic processing that the Vo system shuns.

Edited May 15, 2014 by Cycfi

[elmo7sharp9]

"it's easy to chop an envelope to achieve the muted banjo like effect."

I keep forgetting the Polyphonic scope - the ability to treat each string separately...

The Vo muting, then, is a rather "moot" science-demonstration of the power of the system.

Oh, if only there was some way of using the edge of my hand to tame the string vibration .

Regarding previous comments about using a Roland hex pickup as a source:

I have done this, in combination with a monophonic power amp and driver coil,

in order to achieve a more consistent response across the strings

(Fully polyphonic sustain was beyond my grasp - I used trimmers in a summing amplifier to achieve balance).

The GK2A, that I used, is not PASSIVE... It has active buffering for each micro-humbucker on board,

with different low-pass filtering for each string (to aid Fundamental Detection).

The GK2A takes a +7v/GND/-7v supply in normal use - it DOES work with a 9v battery and a resistive divider to provide the GND tap.

Regarding the possibilities of stimulating a string with samples etc...

We can only stimulate the harmonic series that is physically present in the string.

This usually results in the string sustaining a close-to-sine tone - a musically uninteresting result,

once the novelty wears off.

I have had freak occurrences of a string sustaining BOTH the Fundamental and the Octave-plus-a-Fifth

harmonic - a musically interesting result that can surely be made repeatable by controlling the proportions of each harmonic in the stimulus signal.

It would be interesting to have a slowly-shifting transition through the harmonic series as a note sustains. This is already achievable, using an E-Bow, by pinging a "weak" harmonic and observing the string vibration "morph" to a pitch that has a stronger resonance - but the effect is a one-shot, one-way and monophonic.

The effect would be far more impressive on a Polyphonically Sustained chord - a Tamboura-like drone of incredibly complex texture. Naturally, I'm imagining different rates of change on each string...

It's a just a short step away from what you've achieved already...

I look forward to future developments.

[Cycfi]

Regarding previous comments about using a Roland hex pickup as a source:

I have done this, in combination with a monophonic power amp and driver coil,

in order to achieve a more consistent response across the strings

That makes sense.

The GK2A, that I used, is not PASSIVE... It has active buffering for each micro-humbucker on board,

with different low-pass filtering for each string (to aid Fundamental Detection).

The GK2A takes a +7v/GND/-7v supply in normal use - it DOES work with a 9v battery and a resistive divider to provide the GND tap.

Thanks for the correction. I guess it's the same for the GK3. Are the low pass filters not bypassable? Are there schematics (or at least block diagrams) available? Do you know what the cutoff frequencies are?

Regarding the possibilities of stimulating a string with samples etc...

We can only stimulate the harmonic series that is physically present in the string.

This usually results in the string sustaining a close-to-sine tone - a musically uninteresting result,

once the novelty wears off.

I have had freak occurrences of a string sustaining BOTH the Fundamental and the Octave-plus-a-Fifth

harmonic - a musically interesting result that can surely be made repeatable by controlling the proportions of each harmonic in the stimulus signal.

Exactly! I've been experimenting on driving the strings using various harmonically rich waveforms and the results are astounding! The string is a very resonant filter and passes only its fundamental frequency and its harmonics. It is a comb filter that adjusts to your playing. I'm excited to try more types of waveforms and even noises.

It would be interesting to have a slowly-shifting transition through the harmonic series as a note sustains. This is already achievable, using an E-Bow, by pinging a "weak" harmonic and observing the string vibration "morph" to a pitch that has a stronger resonance - but the effect is a one-shot, one-way and monophonic.

The effect would be far more impressive on a Polyphonically Sustained chord - a Tamboura-like drone of incredibly complex texture. Naturally, I'm imagining different rates of change on each string...

It's a just a short step away from what you've achieved already...

I look forward to future developments.

I'll keep you guys informed. This is my holy grail and I am super excited about this!

I'll need your support! It's an uphill battle trying to convince people to think beyond the 50's "Mojo", relic'd guitars, over-priced "boutique" pickups, etc. It's 2014 now and we're still locked in the past. Guitar players are super conservative but I'm pretty sure there's interest in high tech guitar and I want to be able to connect to them, such as in this forum.

[elmo7sharp9]

"I guess it's the same for the GK3. Are the low pass filters not bypassable? "

On the GK-2A, It's the simplest single pole RC filter, between the pickup and an op-amp input.

Snip the capacitor's ground leg and the filter is defeated.

"Are there schematics (or at least block diagrams) available?"

http://www.vguitarforums.com/smf/index.php?topic=73.0

There's a GK3 diagram as well.

"Do you know what the cutoff frequencies are?"

I have never calculated them. I see that on the GK3 diagram,

each of the 6 channels is identical. My memory of per-string values comes from earlier dedicated Roland guitars (G303, G505 and the G707).

Prior to the GK-series, Roland's dedicated synth guitars

had similar onboard filtering.

When used with the GR-100 (No synth - just Polyphonic Chorus and Compression),

the signal was crisp and clear- no ill effects from the rolloff.

Interestingly, the earliest Roland guitar synth had POLYPHONIC SUSTAIN.

The GR-500 guitar had strings that were isolated from Ground at the bridge.

The strings were grounded at the nut and, when Sustain was switched on, each string carried a 15 volt

sine wave of its detected fundamental pitch.

The strings passed over a broad permanent magnet, located where the Rhythm pickup is on a Les Paul.

http://en.wikipedia.org/wiki/Roland_GR-500

It's an unusual configuration for a sustainer, but not unique...

Watkins Electric Music made an ultra-rare guitar, called the Fifth Man, around 1967.

It featured a similar arrangement, but with opposing permanent magnets mounted at each side of the neck joint. The guitar was self-contained and the circuit was powered by a heavy-duty (and heavy!) rectangular torch battery in a cavity on the rear of the guitar.

http://www.youtube.com/watch?v=NqBaW1qMs4w

Edited May 18, 2014 by elmo7sharp9

[Cycfi]

"I guess it's the same for the GK3. Are the low pass filters not bypassable? "

On the GK-2A, It's the simplest single pole RC filter, between the pickup and an op-amp input.

Snip the capacitor's ground leg and the filter is defeated.

"Are there schematics (or at least block diagrams) available?"

http://www.vguitarforums.com/smf/index.php?topic=73.0

There's a GK3 diagram as well.

"Do you know what the cutoff frequencies are?"

I have never calculated them. I see that on the GK3 diagram,

each of the 6 channels is identical. My memory of per-string values comes from earlier dedicated Roland guitars (G303, G505 and the G707).

Prior to the GK-series, Roland's dedicated synth guitars

had similar onboard filtering.

When used with the GR-100 (No synth - just Polyphonic Chorus and Compression),

the signal was crisp and clear- no ill effects from the rolloff.

Thanks for the links. I just recall now that I've had these links before. Man, I'm getting old! The one who wrote that (Elantric) even sent me the same links sometime ago while developing the Neo pickups.

I see now that this rolloff is the main reason why people are complaining about the roland pickups' sound when sent directly to a multichannel audio interface.

Interestingly, the earliest Roland guitar synth had POLYPHONIC SUSTAIN.

The GR-500 guitar had strings that were isolated from Ground at the bridge.

The strings were grounded at the nut and, when Sustain was switched on, each string carried a 15 volt

sine wave of its detected fundamental pitch.

The strings passed over a broad permanent magnet, located where the Rhythm pickup is on a Les Paul.

http://en.wikipedia.org/wiki/Roland_GR-500

It's an unusual configuration for a sustainer, but not unique...

Watkins Electric Music made an ultra-rare guitar, called the Fifth Man, around 1967.

It featured a similar arrangement, but with opposing permanent magnets mounted at each side of the neck joint. The guitar was self-contained and the circuit was powered by a heavy-duty (and heavy!) rectangular torch battery in a cavity on the rear of the guitar.

http://www.youtube.com/watch?v=NqBaW1qMs4w

Now that *IS* interesting. Have anyone tried that here? Talk about feedback! That's feedback gone wild. Perhaps some filtering will tame that feedback. I wonder how well it behaves though, especially when playing clean with lightly played and long sustained notes (none of that in the video). That demo sounds like an uncontrollable mess, with lots of noise and some overpowering microphonic feedback.

[elmo7sharp9]

Now that *IS* interesting. Have anyone tried that here? Talk about feedback! That's feedback gone wild. Perhaps some filtering will tame that feedback. I wonder how well it behaves though, especially when playing clean with lightly played and long sustained notes (none of that in the video). That demo sounds like an uncontrollable mess, with lots of noise and some overpowering microphonic feedback.

Look at the alignment of the magnets... the WEM Fifth Man sustains the side-to-side vibration of the strings, the Roland sustains the more usual up-and-down vibration.

From memory, the Roland has a Common-Collector driver for each string with a different current-limiting resistor value for each channel (Presumably to achieve some balance across the strings - I don't see there being a significant variation in resistance-to-ground through different string gauges!).

I tried it out on a piezo-equipped acoustic, using a variety of magnets at the neck - couldn't get sustain at all - tried different power amps, too. The usual "Too Many Variables To Account For" that DIY Sustainer enthusiasts know well.

[Cycfi]

I tried it out on a piezo-equipped acoustic, using a variety of magnets at the neck - couldn't get sustain at all - tried different power amps, too. The usual "Too Many Variables To Account For" that DIY Sustainer enthusiasts know well.

LOL. I know exactly what you mean. I have this hall of shame drawer full of non-working junk. I'll post that sometime just for laughs :-)

Anyway, I think I'll stick with what I have now. One variable. Refine. Rinse Repeat. I think I'm closer now than ever before.

[Cycfi]

Hello Y'all,

The Infinity Project... We've been talking about it here. Here are my thoughts on how the project will proceed, the goals and objectives, the implementation strategy, how it compares to current systems, etc.

http://www.cycfi.com/2014/06/the-infinity-project/