My Electromagnetic Harp

[oolfur]

Hello,

My name is Úlfur, and I am a musician living in Iceland.

I spent last year building a prototype of an "electromagnetic harp" that is computer controlled.

What it basically does is drive 16 different strings with hand woven drivers and pickups, controlled via a capacitive touch interface (polished pennies seen on the front).

The strings are on the inside of the instrument, and it can generate its own music when you dont feel like playing yourself.

It sounds like this ::

https://soundcloud.com/ulfurh/segulharpa

Right now I am working on version 2 - adding a **** ton of features - but first and foremost i am trying to optimize the driver system for better performance.

This morning I was doing some google research - and I came across psw's "Sustainer Ideas" thread - I am literally shaking from excitement !!!

First of all - I read the first 90 pages of the thread and i applaud you guys for all your amazing discoveries!!! - I am incredibly happy to find that other people that have been tackling the same problems as I have, with such incredible results! i wish i could have been part of the discussion sooner.

The reason why i am starting this thread is to make contact with people like psw, lovekraft and ansil - hoping they would be interested in a dialogue concerning this type of instrument.

psw's idea of an acoustic synthesizer sound incredibly intriguing - and I have had a similar idea for the next version of my instrument. i have had some great results driving audio through the 16 different strings for a kind of "Acoustic String-Reverb".

it sounds a little something like this ::

https://soundcloud.com/ulfurh/vocals-through-kassagitar/s-XNTkY

the early prototype I got working a year ago was made only using LM386's and crude hand woven coils. here it can be seen playing generative music on a grand piano ::

I implemented this system into the prototype harp casing seen above, but right now i am working on a design which utilizes the LM386 as a preamplifier for piezo pickups, sending the signal into a more powerful LM1876, that drives some super expensive electromagnets i bought online.

this allows for high fidelity signal processing of audio through the strings of the new harp.

also, there is a rubber membrane on the inside with a small magnet mounted in the middle that is being driven with the master signal from all piezos combined, producing a rich and beautiful acoustic "rubbery" reverb. the instrument can therefore be played as an instrument by itself, or used as an audio effect for signal processing (early tests indicate AMAZING results on vocals and guitar)

if anyone who has dived into this business extensively is interested in working out some ideas together - i would love to chat!

more pictures and info on my harp can be found here, as well as my contact info :: ulfurhansson.com

all the best!

úlfur

[ansil]

very very very nice my friend. this far surpasses my string box i used to keep on my amp to do stuff like this. i want one now i just need some money. if i was wearing a hat i would indeed be taking it off

[oolfur]

hey, thanks a lot!!!

one thing i've been wondering about for the new version

before the LM1876 driving each "channel" (string) i need something to preamplify the signal from my handmade pickups.

what I have been using are some LM386's, but they havent been the best preamplifiers... they tend to go crazy and oscillate and arent effective enough until they reeeeaaally distort the signal. thats with the gain set at 200.

in the meantime i use my zoom h4 to preamplify the signals while i prototype.

are there any other simple circuits out there i should try? maybe like a TL072 headphone amp?

[ansil]

use a single lm386 to provide you with a stable v ref and try a tlc 2262 mosfet opamp

[oolfur]

hey ansil

that sounds like a good idea

do you know of any schematics that i could use for reference? not quite familiar with the 2262...

maybe im experiencing these problems because of the power supply - im regulating a 25v supply to 9v for a LOT of these opamps.

is there a certain etiquette i should keep in mind when multiplexing these circuits?

[Cycfi]

Wonderful idea! I too am going into "acoustic synthesis" starting with the hexaphonic pickup project thread in this forum. I'm very interested with your results, especially the drivers. I'd prefer class D amplifiers for efficiency, BTW.

[Prostheta]

Welcome onboard! I'm going to have to digest this at greater length since my brain is a little broken today. Wow, though. This is seriously 1000% cool!

[oolfur]

hey thanks guys!!! happy to have found this place...

@ cycfi i have been reading your webpage all morning! its awesome, never heard of hexaphonic pickups before... sounds like something i should consider for my new instrument.

my drivers and pickups are handwoven with a disassembled drill/bobbin. the drivers are NOT "e-core" shaped, since this design seemed to work better with neo-dymium mags on the front. here are some pics ::

maybe the elongated design is not optimal, i will have to make further experiments - it seems like some of you guys have A LOT more experience in winding drivers than me... please share your insight if you like!

here they are on a grand piano, smaller ones are 220 ohm pickups, longer ones are 8 ohm drivers, clad in some red shrink tubes.

later i decided to make the pickups humbucker style, they work even better.

- so, since im not that smart in the diy amp department - whats a class D amp?

[ansil]

i can draw you up something if you like. are you using one preamp and multiple drivers i take it. also the 386 can handle up to 22 volts on some of them. at 9v they can get squirmy.

[oolfur]

at this stage, im actually thinking of running as many as 16x lm386 to preamplify 16 pickups...

what is the most economic way of preamplifying a guitar pickup signal? is the LM386 definitely the best way to go?

thanks for your interest in this!! i hope the outcome will rock your socks off...

(p.s as i am prototyping the LM1876 - and they are quite pricy - will they be destroyed if i run them without any load from time to time? or is that only the case with tube amps?)

[oolfur]

a simple rectifier or another type of fuzzy harmonic distortion would be a nice option on the preamplifier, to play with different waveshapes driving the strings. the octavia clone seems a little heavy and complex to make 16x... any ideas for an approach on this?

also - here is an idea that you might find interesting -

i haven't been happy with the acoustic sound levels of previous systems i have made. the best i have so far is very subtle, and im hoping i could at least reach the sound levels of say, an acoustic guitar.

as i am getting sleepier while sketching up ideas, i thought maybe physically driving the string bridges would further induce responsiveness through the magnetic actuation coils. does that make sense?

then i would have a driver (kinda like a buttkicker) under the bridges, physically vibrating the strings (the sustainiac model c principle).

it would be something like this :: string - pickup - preamp - LM1876 output A - driver coil on said string

output B - buttkicker under bridge

maybe this approach would lead to the buttkicker taking over the nice weird characteristics of the magnetic feedback.

any thoughts on this?

[ansil]

as far as a preamp goes. you can utilize a single lm386 wired up with inputs grounded and the output will be half the voltage supply so you would have your typical voltage bias use a dual or even quad opamps as they are pennies on the dollar and use those for the amplifiers. you could even use cmos hex inverter for 6 preamps since they are not cascaded into each other the noise level should be quite low. as far as a circuit for it goes you string a pot from output to negative input and set your gain by that. throw in a few caps for freq control and your off and running. if your strings are tuned at a specific freq range then i would try to tailor the freq range of the preamp to that range you will get a much more effective driver section then as its only amplifying what it needs to. ie if you have say a high e string of a guitar which is roughly 329.6hz there would be no need for less than 300hz amplifiying

[Prostheta]

Placement of the drivers is probably quite useful to experiment with also. Is the induction of a fundamental note easier near the greatest point of movement (first antinode, halfway down the string) or does that require beefier driving because of the greater string deflection? Would doubling or tripling up drivers at the second and fourth antinodes induce more pleasantly appropriate harmonics rather than one driver carrying this out?

Unlike a guitar you are not restricted to driver placement because of the playing logistics of the instrument so there far more scope for invention.

[DarkAvenger]

Wow! That's pretty cool!

A class D amp is a more efficient way of amplifying vs a class AB amp like the LM386. My main issue with finding a suitable class D amp is that most are smd and designed for cell phones, etc. rather than driving a string at full power. I still have a class D test chip laying around... hm. Somewhere.

Anyway, for your application a class D amp or class AB isn't an issue, go with what works best. I assume your going to plug into the wall no matter what, you know with the whole 16 power amps and all, so why worry about efficiency? It's mostly relevant when relating to battery operated devices.

[Cycfi]

Wow! That's pretty cool!

A class D amp is a more efficient way of amplifying vs a class AB amp like the LM386. My main issue with finding a suitable class D amp is that most are smd and designed for cell phones, etc. rather than driving a string at full power. I still have a class D test chip laying around... hm. Somewhere.

Anyway, for your application a class D amp or class AB isn't an issue, go with what works best. I assume your going to plug into the wall no matter what, you know with the whole 16 power amps and all, so why worry about efficiency? It's mostly relevant when relating to battery operated devices.

Those are very good points actually. I thought at first that this would be a portable instrment, but it seems it's not. So DarkAvenger is right, it does not matter much.

[Cycfi]

hey, thanks a lot!!!

one thing i've been wondering about for the new version

before the LM1876 driving each "channel" (string) i need something to preamplify the signal from my handmade pickups.

what I have been using are some LM386's, but they havent been the best preamplifiers... they tend to go crazy and oscillate and arent effective enough until they reeeeaaally distort the signal. thats with the gain set at 200.

in the meantime i use my zoom h4 to preamplify the signals while i prototype.

are there any other simple circuits out there i should try? maybe like a TL072 headphone amp?

I won't use an lm386 as a preamp. It's not designed to be used that way. You are right, it will oscillate like crazy if the gain is set high. And you need a high gain since you have very low impedance (220 ohm) pickups. I'd use an Op Amp wired balanced like an XLR mic preamplifier. You can see lots of schematics of these if you do a search for "balanced preamplifier schematic". As for the Op Amp, yes, the TL072 is better suited. I have here a list of some more Op Amps: http://www.cycfi.com/2013/04/choosing-good-audio-op-amp/

[DarkAvenger]

Why not use/make step up transformers? Or for that matter, wind larger pickups, higher output, thus less amping needed. I don't think it will alter the result too much, maybe to a slight degree but nothing too dramatic.

However, keep the drivers the way they are.

For the pickups, once the impedance (not resistance) is greater than about 25k, you will need a small jfet preamp to lower the impedance when inputting to an lm386(50k input?). However, with a larger input you will need less amplification, eliminating oscillation from high amplification and reducing the power needed in addition to the possibility of reducing noise in the system which in turn will improve your results.

Just some thoughts to leave you with, the inductance of a coil is related to the square of it's turns. Inductance relates to output. A coil less than hmm... say, 1k ohms? is probably wasting output with very little benefit. There are quite a few other variables that effect such things such as design or wire gauge or metal mass nearby, I'm not quite sure where a DC resistance measurement would be exactly... but that's my guess for your design given the facts I know. The best test will be to wind something up an try it!

[Cycfi]

Why not use/make step up transformers? Or for that matter, wind larger pickups, higher output, thus less amping needed. I don't think it will alter the result too much, maybe to a slight degree but nothing too dramatic.

However, keep the drivers the way they are.

For the pickups, once the impedance (not resistance) is greater than about 25k, you will need a small jfet preamp to lower the impedance when inputting to an lm386(50k input?). However, with a larger input you will need less amplification, eliminating oscillation from high amplification and reducing the power needed in addition to the possibility of reducing noise in the system which in turn will improve your results.

Just some thoughts to leave you with, the inductance of a coil is related to the square of it's turns. Inductance relates to output. A coil less than hmm... say, 1k ohms? is probably wasting output with very little benefit. There are quite a few other variables that effect such things such as design or wire gauge or metal mass nearby, I'm not quite sure where a DC resistance measurement would be exactly... but that's my guess for your design given the facts I know. The best test will be to wind something up an try it!

The higher the inductance, the lower the cutoff frequency and the darker the sound will be. There *is* merit with low impedance pickups below the 1k range. No wonder Les Paul himself favors a low impedance pickup of 10 Ohms! (see Les Paul’s Favorite Les Paul – The Recording Model). He sends the balanced output direct to the mixer. I've had very good results with small, low impedance pickups in the 100 ~ 300 ohm range. Here are some preliminary spectrum analysis of low vs. high impedance PUs on the Strat and the Les Paul and a Bösendorfer grand.

[DarkAvenger]

It's fairly hard to compare such different designs. Impedance and inductance and a number of other factors influence the sound.

In practice, testing will prove more vital since I honestly don't know where that cutoff point will become noticeable. There is also still the question of whether high frequencies are desired and to what point. The simple answer, preference.

[Cycfi]

It's fairly hard to compare such different designs. Impedance and inductance and a number of other factors influence the sound.

In practice, testing will prove more vital since I honestly don't know where that cutoff point will become noticeable. There is also still the question of whether high frequencies are desired and to what point. The simple answer, preference.

Oh it is noticeable. The pickup itself is a second order low pass filter. HIgh impedance designs have the cutoff point in the audible range. It's the thing we associate with the color or flavor of the pickup (e.g. dark, bright, etc.). It's easy to trim high frequencies with a second order low pass filter to shape the sound later, but not the other way around. IOTW, you can remove what's there but you cannot add what's not there to start with (i.e. higher frequency harmonics). That way, the player has the flexibility of tone shaping rather than having the pickup itself hard-wire the tone for you.

[Prostheta]

<edit: I just realised how much of what I wrote here was said perfectly well by Joel! Oops. Just goes to show how fuzzy my head is right now>

Agreed about how noticeable the "character" of a pickup is. I would add that the low-pass filter characteristic of a pickup usually possesses a significant bump at the resonant frequency before the highs fall off. In addition, eddy currents can affect the response by adding a drop just before this resonant peak. Whilst I am only vaguely aware about what causes eddy currents and their effect, I know for certain that this whole "resonant peak and falloff" is the most characteristic form of a pickup's "profile".

Although not directly relevant in this case since you are not modelling a guitar pickup, it might we worthwhile noting that it is possible to simulate a prototypical "humbucker sound" or "Strat pickup sound" using filtering to emulate these response curves. I am sure the application that Line6 use in their Variax range takes a more complex approach than this simplified example, however it IS certain that the source is fairly wideband in the first place to provide a source from which to perform subtractive and additive processing. You can't add back what wasn't there, as many studio engineers will point out.

Again, this comes back to what you want from the setup. You're not looking for guitar sounds or playing the instrument in a guitar-like manner. The sound is more than likely meant to be a basis for subsequent processing and manipulation (perhaps even with a degree of feedback to the drivers) so low turn count wideband pickup coils make more sense. There is no need for heavier winds similar to guitar coils. As the example of Les Paul's personal instruments goes to show, guitars don't even need them....we're just stuck with designs which haven't changed since the 50s.

I am not thinking too straight this week due to illness so bear with me if I appear to be contradicting myself or going off-track!

I would perceive this project as a fantastic basis for creating more organic sounds from a digital source. A digital mouse organ?

Just kidding. From the video it looks like some form of MIDI is driving a synth which in turn is used to drive the strings of a harp (or grand piano) which is then picked up by pickup coils for the final sound. I think that if the driver and pickup unit had an interface (virtual or hard) to make it configurable as to how much of the pickup signal was fed back to the driver stage (both pre and post processing) then it would add in a wonderful temporal element to the instrument where sounds can evolve in the analogue domain despite the control and processing being purely or partially digital. You could do all kind of stuff with that....