Diy optical pickup for guitar -- is it possible?

[sorveltaja]

There is surprisingly low amount of information about the subject on the net. Yes, I'm aware, that there are commercially made optical pickups for guitars and basses.

But that's not what I'm after. What I'm after, is to expand the soundscape of electric guitar, in analog way, doing it using(hopefully) easily available components.

To be specific, my aim is to make a hexaphonic pickup, that has separate output for each guitar string.

In the past, I made hand-wound coils for each string, but their output was very weak, so I buried the project.

I have to say, that I'm not an artist, or electrician, so I probably butcher every one of the rules involved.

But, besides of that, I can see the reason, why there is so little info about making optical pickups on the net.
When there is practically nothing to look for, you have to try it for yourself. That's where it gets really challenging. Not only that, but hexaphonic pickup needs same circuit for each string.

I have done some testings with different setups, using 3mm infrared leds, and -receivers. Somewhat engouraging, as the crosstalk between strings is extremely low, not even noticeable, when listening to the output. But those infrared leds, and -receivers do not produce equal output. More on that later.

Unfortunately no pics yet, and I know this thread is useless without them. Will post them soon.

If nothing else, I hope to bring out positives and negatives, that noob like me faces on project like this. I have no idea, where it leads, but anyways, journey has started.




 

[sorveltaja]

Here is the current setup, that has 3mm infrared leds and phototransistors paired, so that they can be plugged in or out separately, if it is needed. IR components are press fitted.

Parts holding them are 3d-printed, also press fitted to the base, as they would be very tricky to machine, because every pair has to have different height, corresponding to that string.

Meaning, that the led's and phototransistors "eyes" should be at the very same level with the string, to create a maximum "disturbance" in the infrared light, that travels from the led to the receiving phototransistor, providing the output:
 

Hopefully bit more clearer picture, as in the above pic, the ends of the IR components have been painted to prevent the "shine through" -effect:
 

   

There are plenty of configurations, of how the IR components could be arranged. At the moment, I use the horizontal one, because the string doesn't disappear from the IR beam, when bending it.
Vertical one cuts out immediately, unless there are multiple IR component pairs for each string.

For testing purposes, I'll try to make every consisting parts, and stages as "modular" as I can.

This earlier contraption has practically the same configuration of components, as above, but they are all part of... one part.
Addition to that, I used superglue to secure them in place. All seemed jolly good, until I tested the outputs. There was nothing to write home about.
 

Closer inspection revealed, that all the IR component's "eye" -surfaces had gone dull, because of the cyanoacrylate fumes .
As one might guess, none of those parts are "serviceable" anymore.

Fortunately basic IR components aren't that expensive, neither is the PLA for the 3d-printer.

Next comes the "interfacing" with a guitar, while testing. I haven't yet found a way to "surface mount" all the possible electronics, without disturbing the guitar's delicate, inner self.

If one wants to take the step, to test out the concepts mentioned, be sure to have a sacrificial guitar, that doesn't mind the few extra holes drilled through it.

I think the interfacing is very important, to assure, that the IR components stay in place during all the tests.

I made a cavity to the back of the guitar, to have enough space for two small circuit boards. I used paint to attach aluminum foil to the surface, just in case, that the grounding is needed:
 

There are 12 holes, that go through the wood:
 

 
That way, there is something to attach the phototransistor's leads, instead of them moving wildly around. Longer leads belong to the IR leds, and are wired in series, instead of earlier, parallel configuration.
--------------------------------------------------------------------

Feeding the IR leds is somewhat confusing for me. While wired parallel, there was big differences between their interaction with phototransistors, so much that every one of them required "tuning", and still didn't deliver other than unwanted noise for the receiving component.

Series config offered pretty much the same results. All that was done using plain DC for the leds.

Seems that it wasn't enough to keep the phototransistors busy enough, to keep them from observing the background noises.

More on that in the next posting. 

[RussellT]

I'm curious, why is an optical pickup better? And aren't you making it more difficult by working so close to the end of the string?

Russell

[russ57]

I'm curious as well, but I guess it's just another thing I've never heard of.

I would expect you would need a very narrow beam in order to detect the movement of the string, so as it vibrates it can cut off the beam. So the beam would need to be at right angles to the vibration of the string.
Is it vibration parallel to the body or perpendicular? I would think parallel?





Russ

[philf]

I too wondered what the advantage could be and found this:

https://www.willcoxguitars.com/lightwave-optical-pickup-system/

"Music at the Speed of Light
The optical pickup is a proprietary type of transducer which utilizes an infrared emitter and an array of photodetectors for each string. The emitter casts a shadow of the string onto the photodetectors. As the string vibrates, the size and shape of the shadow changes accordingly and modulates a current which passes through the photodetectors. This current is the analog electrical signal which represents an accurate depiction of the vibrating string."

I do wonder if the ordinary listener would be able to detect any difference.

Phil.

[Will_D]

I would think that the optical method will work with any string as opposed to the electro-mechanical pickup that only works with a metallic string

As I am in no way musical the above may be pure BS

[sorveltaja]

Thanks for the replies. RussellT, optical pickup isn't necessary better than the ordinary electromagnetic one.

Reason, that I chose optical pickups, is that they are much more compact, and ready to be used as sensors.
Electromagnetic pickups require very precise winding of really thin copper wire, to produce good output. Naturally, space between the strings limit the size of those type pickups.

Not only that, but when one wants to make a polyphonic pickup, six separate elements are needed for guitar.

What do I mean by polyphonic pickup, and what is its purpose?

The consept comes from commercial guitar synthesizers, that have a specific, polyphonic pickups.   
Although that kind of pickups can be bought separately, unfortunately they require a model specific synthesizer device to work, which are rather expensive.

Difference between mono- and polyphonic pickups:
 

So the polyphonic(or hexaphonic) pickup has six outputs, which each could be processed independently. For example, signal from one string could be panned to the left, and signal from other string to the right in the stereo field. Or each string could drive different oscillators. Almost endless amount of options.

What comes to the current setup, I placed the optical pickups near the end of the strings(bridge), so that they are not in a way, when plucking strings.
Obviously, string vibration is very faint at the end of the string, but even that minimal movement is enough for the IR receivers to pick up even the tiniest deviations in the IR light, that the leds produce.

Russ57, that is exactly, what I thought at first. That idea came from optical encoders, that use slotted wheels to interrupt the IR beam.
Then came the idea of using narrowed laser beams as a light source. But I think, that the light source should be infrared, according to commercial optical pickups.

Infrared laser is definitely no-no, as it is plain dangerous, making serious damage for the eyes, without proper safety precautions.

Philf, thanks for adding the quote. It pretty much tells the essential things about optical pickups.

Does the ordinary listener detect the difference? Not necessarily.
But If the guitar or bass has only optical pickup(s) in it, there isn't any magnetic pull from the electromagnetic pickups, that might dampen the sustain of the steel strings.

So I assume, that to tell the difference, you have to be either musician, or any other person, that had to develop their auditory senses, while working with the music.

Will_D, That's not BS. When using optical pickup(s), the string material can be something like nylon or rubber, or practically anything in a form of a string, that vibrates.


 

[BillTodd]

 If you want linearity from optical sensors, you might try to get a difference signal from two sensors  (such that the motion of the string covers one while exposing the other ) the differential approach will also remove overall illumination variance .

might also be worth modulating the light with a carrier and then synchronous rectification to recover the signal to remove any external interference.

fixed and tapered apertures around the light source and sensors may also help

bill

[sorveltaja]

Bill, thanks for the suggestions. Could the differential approach be used for more than one pair, at the same time. Like for all the six outputs?
I'll have to admit, that it is grey area for me, even though there might be just simple math or logic involved.

But on the other hand, I have made a very simple square wave generator, to feed the IR leds, instead of DC. It is based on a 555-timer IC, that runs at ~120KHz.

That frequency is just a sum of the components, that I had on the shelf:
 

 
I have noticed, that by using plain DC for those leds, they tend to hog a lot of current, without actually "feeding" the receivers. I see those receivers, as bird cubs, that scream for food all the time.
When they are fed hundreds(or thousands) of times per second, they don't have the time to observe the environment, or make noise.

I'm not sure, if it is any form of modulation, but rather pulsing, to keep the receivers busy, so that they notice only the aberrations in the IR beams.
Also, current consumption is minimal, when compared to DC.

Some specs: my current setup has 3mm IR-leds (IR204), and 3mm IR-receivers (PT202B).







 

[BillTodd]

Your generator is effectively sampling the position of the string (like  strobe light) , it is sufficiently fast (at 120kHz) to catch all audible frequencies (has to be at least twice the maximum desired frequency or it will produce differential , alias, signals ). You may be able to reduce power consumption or reduce noise by reducing the on period while increasing the drive current (such that the average and peak current is within the diodes range). (look up shannon sampling theory)

An effective differential pickup would require at least one opamp per string (but they come four to a box so it's not such a problem).  The hardest bit is the optical layout , are you relying on reflection from the strings or trying to have the beam cut by it?.

I wonder if you could use a pair of standard opto switches at slightly different heights to catch the top and bottom (or left and right) of the string??

[sorveltaja]

Although the square wave generator pulses the IR-leds, which are now in series, there seems to be a lot of deviation between their output ranges.

At the very beginning, I tested those IR-leds in parallel configuration, just to find the same differences.   

So today I tested them with an adjustable DC source, still in series. All of them have different voltages, or "sweet spots", where they give the most output for the receivers.
Bit confusing, but I guess they(IR204's) weren't made to be that accurate, after all, what comes to the operating voltage and current.

One thing, that I might test with them, is to use potentiometers to "tune in" each led. If that doesn't work, it's just an end of part one, as I have already few other types of IR-leds on the shelf.

Bill, I took a look of the shannon sampling theory. Somehow it makes sense, but I haven't found yet a way, of how to implement it in practice.

I'm not quite sure yet about the optical layout. For example, Opik pickups by Light4Sound use reflective technique, which is one option.

Current setup focuses on disturbing of the IR-beam, instead of cutting it, as it could require microscopic precision.

Using the optoswitches is also one option. My local supplier has some Omron optical switches available, which I guess, could have better tolerances, than the discrete IR-components.



 

[BillTodd]

how are you measuring ir output? 

The ir laser in a cd pickup uses a second monitor diode to detect the laser level, that is fed back to the laser current to keep the laser output level constant with temperature , aging etc.

It seems more likely to me that the variance you are seeing is due to optical path differences.

[sorveltaja]

I think the IP-receivers' outputs could be measured with multimeter, but the readings come and go so fast, that it's a bit hard to follow. 

When plucking a guitar string, there is first the highest peak, then the signal slowly fades away, so it hasn't that much of constant values.

Without having a storage oscilloscope, I judge the results just by listening. That way it's easier to observe, if the signal is too weak, or is it too strong, causing distortion.
I have tested the IR receivers one at a time, and the signals from them goes first through the unity-gain op-amp(9V battery powered), then to the laptop's mic input.

Optical path differences, yes they likely have an effect. By using variable DC-supply(30V/3A), that displays both voltage and current, I got some results.
From left to right, E1 is the thickest string, while the other E is thinnest.

When using plain DC, the "window", where the leds gave any usable output was really narrow(numbers above the letters)
 

 

One (or at least me) could expect more output, if more voltage is used. But no, outside of the above voltage ranges, there was no output at all.

I must be missing something essential here. Perhaps the receivers have something to do with those narrow windows? I Can't tell.

Then comes the test, using pulsed square wave at ~120KHz.
Generator schematic:
 

   

As can be seen, there is 7812-regulator, meaning, that it provides regulated +12V voltage for the 555- timer IC.

Results are very confusing. Again, numbers above letters present voltages fed to the circuit, to find out the sweet spots of the IR-leds outputs:
 

For some reason, E1 has the broadest response, as it gives good output, when supplied from 11 to over 20 volts.

All the others have very narrow output ranges, that works only below the actual, regulated voltage.





 

[BillTodd]

at a guess the larger string is obscuring the most light and allowing the transistor to operate in its linear ( non-saturated) region.

ATM you seem to have the string just wobbling  about infront of the transistor, you really need to get only the edge of the shadow to fall on the transistor die . Sketch out a scale diagram of light source , string and receivers , to  workout where the best placement for each should be. you may need to add apertures to source and receivers to get a crisp shadow.

You are making it hard for yourself if you don't have the right tools. Without an oscilloscope you are blind . You don't need anything expensive, the little pocket devices sold ln eBay would be more than good enough for your purposes.

BTW the 33u in the output that drives the leds , is making them see an ac drive . It is not necessary .

[PekkaNF]

To my experience pretty much all optocomponenets are pretty unlinear and they have large variance between componenets even in same manufacturing batch.

I would start checking iif all those optodiodes will need individual current or voltage source. If they are laser diodes, they normally have their own drivers and they behave very diiferently from normal LEDs. My point is that each emitter needs certain driver and you might need a separate driver for each channel.

Also receivers are most likely unlinear and strings/optical baths are diferent -> you need some processing for each channel. I can't estimate if with one opam you can bias, linearize and amplify signal. They must be tuned individually. No idea how to do that.....different frequences and plucking the string might not be exactly constant------probably to keep un plucking until statistically representatice smaple cloud has been reached.

Pekka

[sorveltaja]

Bill, yes the string shadow, and the placement of the sender/receiver pairs needs more testing.

In fact, one of the commercial optical pickup uses aperture to "tighten" the width of the receiver's "eyesight": https://patents.google.com/patent/US20110265635   

Oscilloscope has been on my buying list for years. I have waited for the "justification" to purchase one. Perhaps the time to get one has come closer.

I breadboarded the 555- timer circuit, using a bit different configuration, that I found on the net. So now it gives DC pulses, instead of AC.
But the results are pretty much same. The leds still have their narrow, individual ranges, that produce output.

Pekka, it seems to be very true, that they have plenty of deviation. Actually, what I'm planning to do next, is to somehow feed each of the IR-leds separately, from the same voltage source.
Maybe at first by using separate trimmer pots for each one, to see, how it works.

 

[sorveltaja]

I tested the previously mentioned configuration, using trim pots for each sender/receiver pair. Yes, it kind of works, but the receivers' outputs are still at a very narrow ranges.

On those narrow output ranges, it's more like an on/off switch situation, where the outputs are either non-existent, or are overly strong, causing distortion and a lot of noise.

I don't mind the distortion, as all those outputs, when I finally get there, are to be converted to CMOS- level square wave form.

But with that much noise, it seems to be very difficult to achieve with the current mechanical setup:
 

As I see it, is that when the sender/receiver -components stare each other in the eyes at so close distance, the receiving one goes mad, and gives overly saturated output, even if it isn't amplified in any way.

After all, the optoswitches have similar configuration, but are perhaps designed to be nothing more, than on/off -devices.
---------------------------------------------------------------------------------------------------------

I'll put the current concept aside for now, and move on to the next mechanical configuration, which concentrates to the reflection of IR-beam from the vibrating string.

To be honest, I'm not going to re-invent the wheel. Instead I'll use the info from the previously mentioned patent.

As it is for commercial product, there obviously isn't any "specific" details available.

[BillTodd]

The photo transistor is just like a normal transistor, its transfer characteristics are determined by base current. Even in a simple amp circuit the base current is carefully controlled , usually by a feedback mechanism (.
e.g. emitter resistor)  .  The photo tranny's base current is supplied indirectly by the impinging photons,( and it doesn't really care what colour they are)  there is no feed back , even with emitter resistors.

You have to either have to provide a very stable light level  (practically impossible given the circumstances ) or some means of feed back  to control the level.

[WeldingRod]

You might be able to get a steady string motion signal by using a speaker near the string, or driving a conventional pickup with a function generator.

Sent from my SAMSUNG-SM-G891A using Tapatalk

[sorveltaja]

Bill, yes the phototransistors(at least the ones I have tested(PT202B)) seem to be quite tricky to work with, as they all have different, individual operating levels.

WeldingRod, that's a good idea. There is actually a commercial device called "E-bow", which allows the player to electromagnetically "excite" the strings, but only one at the time, to make the string to "sing" as long as the device is applied to it.

Hmm.. after a quick search, there seems to be a plenty of those who had done "diy" -versions of that device, so I guess it shouldn't be that hard to build.
That could be nearly ideal one to produce constant signal from the string(s), which could then be more easily measured as an output.

In the previous post, I mentioned that I'll be moving on to the reflective approach, as mentioned in the patent. But not so fast, as the saying goes.

While taking the current setup apart, which consist pairs of 3mm sender IR-leds(IR204), and 3mm receiving phototransistors(PT202B), I Somehow got to this crazy, 'mad modder' -mood, and tested, how the actual emitting IR-leds(IR204) react, when they are treated, as receivers.

To my surprise, they provide quite a lot of output, when considered them to be emitters. Also they are bit more forgiving, than the optotransistors, what comes to the operating levels.

So I tested a pair of IR204's, one being the sender, and the other being receiver. Didn't seem to work well with plain DC, so I switched to previously used, pulsed source.

Crude circuit, that I'm currently using to amplify the signal:
 

It is powered by split power supply. But so far, there is good output, although with some hum. But that is mostly due to the jungle of wires involved, when testing.

[sorveltaja]

After testing several kinds of circuits found on the net, considering sine wave(amplified output from the receiver) to square wave conversion, it doesn't seem to be that straightforward. 

One of them being a zero crossing detector, that, as I assumed, should be sensitive enough for the purpose:
 

But there wasn't any output at all. So I removed all the parts from the breadboard, just in case, that there was mistakes, and built it again from the scratch. Still the same result.

Next thing was to test with multimeter, does the 'preamp' -circuit(mentioned in previous post) give measurable 'peaks', as the electromagnetic pickup does, when the string is plucked.

No, it didn't. Be it AC or DC range. Nothing. But the output is still there, when connected to the laptop's mic input. Somehow it sounds stronger, when compared to the 'high output', dual coil electromagnetic pickup.

If there isn't measurable AC or DC values, then what kind of output is it anyways?

[russ57]

The first thing to check is the frequency range of the multi meter.
Secondly, the meter will measure the 'rms' value of the signal, this is the average value, not the peak. And it probably assumes a sine wave.
If your signal is a series of 'spikes' rather than a sine wave, then the rms value could be very low.
Audio design really needs an oscilloscope. You can pick one ok for audio off any Chinese trading site if not eBay for under aud50. £20?
(Turning a sine wave to square is easy. Just apply massive gain) .

Russ



Russ

[sorveltaja]

Russ, yes I've been searching for the oscilloscope for a while now. Maybe something like the Nano DSO series, with colour screen.
Some of them even have two channels, like SainSmart Mini ARM DSO202 Nano.

Most of them on the Ebay are located in China, which means that the delivery times tend to be rather long. So I'll probably pick one of the few ones located in Europe.

I'm tempted to go and buy a full size, entry level Rigol, which seems to be quite popular. But perhaps it's not a time for going that far yet.

What comes to my description of the signal of the plucked string, when measured from the electromagnetic pickup's output, it was incorrect.

It actually isn't in a form of spike(s), but more like an envelope.

I'll probably butcher the subject by trying to explain it, but the first picture on this page hopefully shows, what I mean by the envelope: http://www.muzique.com/lab/pick.htm

[PekkaNF]

I really would recomed entry level oscilloscope. Even old analog would be whole lot better than without one. I have had one old CRT 10MHz model that I bought 20€ and gave away, it would have been fine for this.

New DSO:s maybe easier to use, they normally can "find" the wave form with a push button. If you have nothing connected or no signal you normally find mains frequency or quantisizing noise.....but that is learning curve.

With low level audio signal you can't even blow up the scope, when you mess with grounds.....normal scopes have only one ground = chassis, this ir nice to acknowledge.

I had Rigol 1052 (old model several years, sold it with 250€), I think, New owner has been really happy with it. I bought Rigol DS2102E, somewhere bit over 600€ that time with serial decode etc. I wanted bigger display and some triggering options.

https://www.batronix.com/shop/oscilloscopes/DSO.html
https://www.batterfly.com/shop/

They offer few times a year free upgrades of software features, like triggering and decoders (if you are into serial protocols etc.)

Discrete devices with their own displays, settings and ergometrics are just so much nicer to use than the boxes that need a laptop and weird ground (trough the laptop?).

Pekka

[sorveltaja]

This post is an off-topic, but sort of related, because it considers the recommended kind of measuring device, oscilloscope.

After pondering, and watching countless amount of Youtube review videos about oscilloscopes(mostly EEVblog's, and also reading his forum),
I came to the conclusion, that the desktop one could offer a lot more functions and options, than I could ever need or fully understand.

One thing, that I must say: This kind of project (optical pickup for the guitar) definitely does not require a fancy 'full-featured' desktop oscilloscope, for the experimenter to observe signals.

------------------------------------------------------------------------------------------------------------
Totally off-topic:

When I'm talking about having a desktop oscilloscope, it's more like a part of my own journey, as I have wanted to have one for years now.

Then came the question: which brands are available in finnish stores? Owon's cheapest model(SDS1022) has nice, big 7 inch screen in it, but for some reason, there isn't too many reviews to be found of that online, so the reputation isn't confirmed in any way.

After all, Rigol is the one, that I ordered. As Pekka said "..but that is learning curve". Probably a steep one, that holds true for me, when I get that device in my clumsy paws.
-----------------------------------------------------------------------------------------------------------------

Back to the original subject. As Russ mentioned, 'Turning a sine wave to square is easy. Just apply massive gain'. But how much gain can one apply to a single op-amp, without making it self-oscillating? Or is it something, that requires multiple stages?