Son of Stealth
The Stealth Wind Controller
In the early 1970s I began experimenting with homebuilt electronic music synthesizers. It was great fun, but I quickly became discouraged trying to play them from a keyboard. It was difficult to get much musical expression using keyboard control, and, in addition, I lacked keyboard training. Being a clarinet player, I decided to try to build a controller that I could play like a traditional woodwind instrument.
I was faced with two possible paths. The first was to try to duplicate a clarinet as closely as possible, either by building a controller with exactly the same fingering system or by attaching switches to a real clarinet. For technical reasons, this seemed like a daunting task, although it has since been done, both for clarinet and sax.
The second path -- the one I chose -- was to make a new instrument. I reasoned that as long as the fingering system was close to traditional woodwind fingerings (which are all quite similar) I could learn to play it. Not being constrained by the physical requirements of an acoustic instrument, I hoped I could find a way to make a simple but effective and expressive controller. There was even a hope that a new instrument could be better than traditional ones. For example, the action could be faster because there would be no need to have long levers moving big pads.
Defining the Task
A wind controller (or WC) has two main components: a mouthpiece and a keywork system. Blowing into the mouthpiece produces an electronic signal that can be used to control the loudness (or volume) of the final sound and also to change the brightness or other characteristics of the tone. Pressing different combinations of keys with the fingers produces an electronic signal that determines the musical pitch being played and that can additionally control other sonic properties. For example, high notes can be relatively brighter or darker than low notes.
The main challenges in building a mouthpiece are:
My first WC was crude but gave encouraging results. It was built with simple tools -- hacksaw, file, electric drill, lots of five-minute epoxy, etc -- and various bits of scrap metal and wood. A detailed writeup was published in the Electronotes newsletter in 1978, about a year after the construction was completed and I had had a chance to make a few improvements and to experiment with different wind-controlled synthesizer patches.
My original mouthpiece was designed to work as follows: The player blows through a passage in an aluminum block shaped with a protrusion that goes comfortable into the mouth. The passage leads the breath into a cylindrical chamber with a thin rubber diaphragm glued on top. As the breath force increases and decreases, the diaphragm moves up and down, and its position is monitored with a beam of light skimming along its top. A photocell measures the intensity of the beam, producing an electrical voltage dependent on the force of the breath. A small hole drilled into the chamber from the bottom allows breath to flow through the mouthpiece, as in a traditional wind instrument.
For the keyworks, I started with mechanical switches (microswitches) glued to a piece of wood. Flexible strips of springy metal mounted over the switches activate the switches when pressed. Recesses cut into the wood place the switches on comfortable arcs, rather than on a straight line as in most woodwind designs. To convert the switch closures into a control voltage I designed an effective electronic circuit requiring just one inexpensive electronic chip (a single operational amplifier).
In this system each switch closure corresponds to a change in pitch by a certain musical interval (up a major second, down a minor second, up an octave, etc.) When more than one key is pressed the corresponding intervals are added together. The keys and their intervals are arranged so that pressing down one finger at a time (starting from the top) produces a descending C scale, just like on flutophone or upper register clarinet. Thumb-operated switches produce octave intervals. Since any combination of fingers produces a valid note, it turns out that any given note may be produced by several different fingerings. With the addition of a few extra pinky keys, I was able to get a serviceable fingering system.
The current incarnation of my WC evolved in a series of steps over many years. It's been operational since September 1997, although construction is not totally done. The pictures shown here were taken in the unfinished state because the features should show up better than after it's all been painted black. The main thrusts in developing the present design are summarized in the following.
Ergonomic Body Design --A major goal of the present design was to reduce the mechanical instability and the stress caused by the right-hand thumb support generally used for woodwind-like WCs. By experimenting with cardboard mockups, I found that spreading the hands apart to make a triangular (as opposed to linear) support vastly improved the stability and comfort in holding the instrument. The design incorporates L-shaped support ears mounted to the sides of the body. The instrument hangs down in typical clarinet position, and the ears hook over the vees in the hands, which support the body by lifting.
Keyworks --I decided to use touch-sensitive switches rather than mechanical ones for the current design. Both approaches have pros and cons, supporters and detractors, and both can work well. I didn't have any particular motivation for changing over to touch control -- it was just something I wanted to try. It has turned out to be something I like very much. It's fast, and although a bit tricky to set up properly, it's been stable and reliable.
Modular Construction --What would the ultimate WC be like? This is a question that seems to have as many answers as there are WC players. One way to accommodate the enormous range of preferences among possible features would be to make a modular instrument. Such an instrument would have a main body with interchangeable, plug-in subassemblies to allow different keywork, mouthpiece, and other "subcontrollers" to be arbitrarily combined. The Stealth's mechanical construction was done along these lines (although the problem of modularity in the electronics has not yet been addressed).
Breath Flow --The original mouthpiece did not allow separate adjustment for flow rate and pressure. A second version had little flow valves, which worked but were too noisy. The new system uses two streamlined constrictions to perform the same function with minimal noise. The constrictions are press-fit into the mouthpiece body and can be replaced to change the aperture sizes.
Diaphragm Response --A rubber diaphragm has a highly non-linear response to pressure. In the current design, the diaphragm presses against a spring-loaded foot, providing an improved response curve. In addition a mechanical stop limits the travel of the foot, which simplifies calibrating the electronics and also discourages overblowing.
Mouthpiece electronics --The LEDs and photodetectors that monitor the diaphragm displacement have highly temperature-dependent characteristics. A servo-controlled electronic circuit was developed to provide a stable breath-control voltage output.