A hundred wrongs make a dozen rights: Iterative design and 3d printed circuits of the v2 hand units
Before-over a year ago- it was much easier to blog progress and process simultaneously. it was mostly technical at that time and i was reading out of books and forums mostly. i just wanted stuff to work and i could express that pretty well, with the odd/random swim into esoteric concerns here and there. but as i have noted in the previous few posts, it seems to be more of an all encompassing thing where technical, artistic and weird shit, all hang out and come up with surreal shit to hit me with , just to see what i do. i allow myself to go off on tangents at times just to investigate the space it inhabits and as such, it can get strange. sometimes for a few days and sometimes for weeks and during this time, i am playing with contradictory information just to see what comes out of the conflict between them. it is intellectually stimulating but when turned into words-spoken or written-it devolves to bullshit so i wait until these idea conflict algorithms play themselves out.
When they do finally converge and mutate, the results are bits of the intense parts of the conflict, but with all the weird shit washed off, or at least recombinated by the process of abstraction into numbers and form. one such conflict has been a design puzzle that has occupied most of the last 3 months; designing a biomechanical interface for interacting with a gestural sonic construct. of course i already have that, but another concept has been setting up shop in my head;
that perfection is a moving target and highly subjective.
a year ago, the exo-voice hand units were, to me, perfect. they fulfilled their function as well as i could conceive of at that time; finger placement, palm positioning, joytoggle functioning, and accelerometer positioning were as perfect as i could imagine at that time. but, of course, there is no static "perfect". perfect is a moving target; a significant point on a process trajectory. as such, last years "perfect" is this years thing to fix.
last year, i uploaded the exo-voice/beatjazz system as an open source project and wondered why i wasn't seeing more of them in the wild. being honest with myself i knew that it was because it was because it is a pain in the ass to print. it takes forever to print everything and requires loads of screws, bolts, tubes, felt and hot glue. the cardboard version that preceded it was much easier to assemble but its internal wire-nest irked a design "voice" that i only just realized was there. 3d printing is definitely the defacto hardware expression vector for the foreseeable future. so if this part of the expression, on equal footing with playing music (which was the primary expression previously, with every other vector relegated to hobby or support ) then it was necessary to look at how hardware was being expressed as a shareable, playable construct.
theoretically, at the time, if i designed it right, it could be assembled right off the printer, but it took a half dozen trips in wrong directions to find a right one. starting at the "wrong" place has many advantages if you can learn to not get attached to these stages.
I knew it needed to be;
- more sturdy-the v1 was prone to glitching over time as the components came loose over time. also, there was the constant fear that dropping it would permanently damage the radio an/or the arduino (which has not yet happened)
- snap together-it need to be assemble-able right away, without tools.
an epiphany came in the form of the 3d printed gun that was making the rounds on the interweb this last spring. once i got over my distaste for what it was, i quickly saw why it was important; because it was designed to not be able to be stopped or controlled; print, assemble, nail, bullet, fire. that was the key! design something that comes out of the printer, working! I experimented from this perspective, producing many "wrong" prototypes. mostly just creating a simple piece that i could hold in my hand then measure with calipers to get a feel for where things could be placed within the design, but many of those "things" were never meant to be used in the working models. they were simply there to explore the idea-space. physical meditative artifacts in a sense.
in early august, after the 3rd wrong iteration along the line of what i had been experimenting with for the previous 2-3 months, i started over from scratch. all of the "wrong" vectors had kind of converged to create a sort of ghost image of something that could work. this was a true version 2, only related to the previous version by basic function alone. all other aspects were completely re-designed but encompassed 2 years of investigation into this field.
to understand why the version 2 design is better than previous versions, it helps t understand why it exists. I used to play a Yamaha WX5 wind midi controller. this is a synthesizer controller that is played like a sax or flute rather than like a piano keyboard. Yamaha still sells them but hasn't investigated the design or concept since the late 90's. there were predecessors as well; the WX7 and the WX11. i played the wx7 and owned a total of 4 of these controllers over the course of 16 years. the wx5 was much better than either of the previous models. it was more sturdy, and allowed for the selection of midi program changes and control changes using key combinations. there was a thumb wheel for the right thumb that let you control one thing when rocked upward and another parameter when rocking downward (which I used for stepping up or down thru the layers of loops I created when performing live). on the upper part of the "horn" two keys were also freely assignable and it is these two that i used to control looper record and overdub in the early days of experimenting with live looping. in addition, i felt it necessary to replace the stock indicator leds with bright blue versions. i didn't understand synesthesia much beyond lsd at that time (erm, from I've read...) but i noticed even at this stage that the light became an important part of the expression.
when i began the project in winter 2011, i had no context for what i needed, wanted or could achieve, so i tried to emulate the functionality i had with the wx5 based system. the first mockup was paper mache with cheap switches for keys and a big gaming joystick on the bottom, painted gold it seemed cool because it was never meant to be played, but i could already tell that the components-battery, arduino, xbee, wires and lights, were not going to fit in it so when i created the first working prototype, i added enough size to shove everything inside it. i never even played that one outside of my flat. i played it exactly twice- once for the YouTube video documenting the project. almost as soon as i played the first notes on it, it was obvious how it was "wrong";
- thumb placement was painful
- accelerometer placement wasn't useful, but merely functional in the simplest terms
- the handbrace broke almost instantly (paper mache)
- the inside was a nest of random wires
it instantly created the design conditions for the next prototype that i would use for almost a year
the cardboard version
I call it a version rather than a prototype because it did actually get published in MAKE magazine as a project so it deserves a v0.5
this joint was so easy to come AFTER the first prototype. on paper, a long-ish triangle placed the fingers in an ergonomically perfect position. the forefinger was right above the thumb, creating a "pinch" space with the accelerometer in between. the inside was still a nest of wires, but the reinforced shell was solid and the 3-way harness held the hand pressed firmly against the palm points, originally made of foam and bottle caps, to keep the keypad position stable underneath the fingers. the second benefit of this version was that it was cardboard-easily modified and if modified too much, easily and cheaply replaceable, which allowed me to experiment with emulating the wx5 keys that i had been using for loop control. after a bit of experimentation, i decided on using joysticks on the keypad side and the thumb side. this consolidated the functions into a smaller space. then it just seemed natural to replicate it on the right hand controller, even though there were no functions to assign to them. it was just that the symmetry made sense. and this was where the leds also deviated from wx5 emulation by using rgb leds to convey much more info, but nothing at this stage alluded to any synesthetic synchronicity. even without it, it was still "perfect" for what it was…until it wasn't...
3d printing changed EVERYTHING. the design process became another computer based expression vector. i didn't know much at the time and had many, many, "wrong" ideas to explore here. first being the attempt to create something that was a shell that i would put components into. this displayed its wrongness before the print even finished. it took too long to print, wasted too much plastic and was distractingly ugly, so i thought i would make an internal frame to mount things in and build around it but once the frame was built, i liked it by itself and kept it that way. but from this, the idea of designing purely from a function aesthetic came. it was always there but it wasn't obvious and i didn't have any reference other than the reprap itself whose pure function based aesthetic is still a source of inspiration.
the version one carried over everything from the cardboard version, but with the precision of 3d modeling, so every component had a precise place to be. the palm points became adjustable, it had integrated led holders and the wires became part of the design by having them sprouting off a tiny circuit board attached to the arduino. it was perfect. i played it for the first time at a black history event in neukölln right after finishing assembling it in the dressing room 5 minutes earlier and it was breathtakingly intimate at the time. it felt like tools for doing something "else" else. eventually the upper joystick got placed into the frame, underneath forefinger and once the first gestural system software started working, the rear facing leds became secondary parameter feedback(edit mode), freeing me from having to look at the computer at all. yaaaay!!!
so, as i stated earlier, last years perfect is this years issue to solve. the circuit board on the v1 started "acting up" regularly to the point i didn't trust that i could get thru a show with it not fucking up and erasing my loops. the hard palm points were beginning to make my hands hurt and the flat key layout was starting to make me feel as if i were finger drumming on a table. in addition, this was always going to be an open source project. i wanted people to build it themselves but i knew from experience that it was a pain in the ass to print and assemble. it took alot of screws and bolts and hot glue and other tediousness, just to assemble it, let alone play it, and that is not even mentioning the headset, which i will cover next week ;-) so knew it need to be be completely redesigned.
the basic needs and desires of the previous year were still there but they were not sufficient for the coming exploratory stage of the project. simply being ergonomic and colorful are the norm since v1. v2 had to represent growth from many perspectives. being an artistic endeavor in most respects, meant that it not only needed to be more playable but had to display an evolving form that moves toward a sense of perfection and efficiency. the form itself needed to tell a bit of story on its own, whether being played or not. a story that hints of lineage, of having ancestry. of being “from” somewhere.
the v1 felt really good to hold but with the v2, I wanted to investigate a feeling of holding “energy” with my hand and manipulating it with my fingers. this meant, in my opinion, that this version should promote a grasp that was more rounded, like holding an apple-sized ball of energy. and if we investigate this metaphor further, eventually one comes to the expectation that energy is probably more “feeling” than form or shape, so in the center of the 3 palm points is a lilypad buzzer to provide vibratory haptic feedback. the beauty of this component is that it can vibrate like a mix between a tiny speaker and the vibrate-function in most phones, so it varies its vibration rate dependent on input. at the center-palm position it occupies, it vibrates my entire hand and can be felt up near my elbow as well as taking advantage of the RN-XV’s high speed bi-directional Wi-Fi. it feels exactly as it should; like parameterized energy in the palm of my hand and under my control. but that is jumping ahead a bit…
the question; what was “perfect” now? this flies in the face of “don’t rock the boat” or “never touch a running system”. stability or lack thereof, had become an expression vector. I came to the realization that its kinda my job to poke this thing with a stick. to explore boundaries that it may not even be fully “necessary” to explore. how do I create a perfect biomechanical sonic gestural interface that is also easy to print, sturdy, snap-together-easy to assemble, and cool?
luckily the version 1 was still very cool and very playable so I could (and have) play it until the new version was stable. first criteria was to shrink it a bit and give the arduino and the Wi-Fi radio a bit more protection, while getting rid of all those little edges and screws that used to catch themselves on my clothes and prevented me from more physically kinetic expression. from the beginning, the accelerometer was always going to be moved back to its position underneath the index finger, but with this iteration, the index finger is angled into a more natural curved angle with relation to the other fingers, which are also angled in the opposite direction. the thumb and the forefinger each have their own array of switches that allow the finger in question to “be” the joystick rather than dealing with the mechanical latency of a physical joystick. the thumb array has a guide groove that helps with mental orientation of the control layout. by providing little grooves and indentations (inspired loosely by braille), the fingers have a location map of function areas.
on the printability tip, I created the first of the v2 “wrongs” with a center pin that went all the way thru and connected everything but I could see how much just simple hand tension would flex the temporary METAL pin in the prototypes so I scraped that idea and went with more of a slot together design that would make ach piece interact with at least two other pieces so each would increase the strength of the associated connection places. THIS IS WHY I HAVENT BEEN BLOGGING! holy hell, you cant imagine how many ways you can do this wrong especially when you don’t know how to do it in the first place. eventually though, I created a key-like mechanism that locked the keypad mount to the back plate. then the battery mount slotted in in a way that secured the keypad mount. then, as a bit of an experiment, I got rid of the perfboard circuit board and designed the circuit into the front panel, as a bit of a jab as the overly mechanistic aesthetic of man-machine interaction representation on popular media. no “shielding” or aggressiveness. I just thought it would be beautiful with the light illuminating to from behind. this piece connects to the keypad and the battery mount as well as the back plate. then the handle slots into keep the battery mount from sliding out and clicks into the other end as well. all together, the design is solid as if it were glued together while being easy to disassemble.
and light expression….oooooh, hexagons. in 3d printing, you have the choice of making solid plastic objects but its usually a waste of plastic to do so, so you can have what is known as a “fill”- a pattern that is printed inside the object that takes up enough space to add structural strength but without having to use so much plastic. these fill patterns are an art in themselves but when combined with transparent plastics, then not only look really really cool, but they refract light amazingly! with the v2, I embedded LED holders into the design. they are part of the form. and the thin layers and hexagon fill conspire to create something that is beautifully functional and expressive. the element lights, the Wi-Fi radio lights and the edit mode lights mesh at times, to create a surreal storm of modulating color which supports the aural narrative and provides precise parameter feedback. and its pretty too.
I don’t want to go too much into the new fractal synthesis system I have been developing for the last forever, but there is much less of the “sea of beats” vibe from a couple of years ago and more of a sonic matrix; the ability to place a sound in a perceptual “slot” and construct with it and around it. each tweak, from or to any point, increases the resolution of the entire idea especially when you recognize it.
I wanted to share this trajectory. in the past I was afraid I was rambling on to much but I figure its easy to unsubscribe if its too much and I feel more honest than simply trying to say cool shit. sometimes its not cool…sometimes its neurotic and wordy and geeky and tiringly involved…and without giving too much away (because I still have to build it), if the contributor mask design (yes, contributors get a mask and not simply a headset….told you it’d be worth it to be patient) is as bad ass as the sketch I just made, I will say, without a doubt, it will be the absolute sickest thing you have ever seen pop out of a 3d printer. no straps, tubes, screw, hot glue…it only even touches your head at five points! but that’s the next post