Self-playing musical instruments have been around for centuries. The greatest era for musical automata was the period preceding and overlapping the early development sound recording devices, from the late 19th century through the very early 20th. It was then that the wonderfully ingenious music machines called orchestrions were at their commercial and creative peak. It was then too that player pianos were ubiquitous, as were music boxes using either disk or cylinder mechanisms. In following years, with the increasing presence of hi-fi audio reproduction technology, the fortunes of musical automata ebbed. Yet it’s interesting to note that, despite their greatly diminished popularity, we are once again in an interesting time for self-playing devices. Developments in maker-friendly robotics and software, amplified by the steampunk and maker movements, have led in recent years to a proliferation of new explorations. Lots of very cool stuff is happening.

This article is about a particular type of music machine, namely, simple rotary devices in which a central turning element activates various sounding objects in its path. This description could apply to a great many music machines past and present including lots of commercial ones, so I’ll add another descriptor: this article will mostly be about rotary devices with a home-made, found-objects kind of feel – devices made not in pursuit of sophistication or perfection but rather in pursuit of quirkiness and character and makerly enjoyment. Such instruments typically play ostinati – that is, repeating patterns, with each repetition corresponding to a single rotation. The maker may add complexity and unpredictability by using multiple simple machines or rotating elements, or by tricks like using swinging and swaying elements which don’t move in quite the same way with each cycle. One of the very nice features of such simple rotary music machines is that they can easily be made adjustable or programable with tinkering by the maker or another user, so the same machine can be used and reused for the creation of different patterns, moods and effects.  

I’ve made a couple of these simple rotary instruments, but before describing them I’ll mention some of my favorites from other makers. 

First on my list is the wonderful work of Pierre Bastien  Pierre has created, in the words of his website, “his own orchestra called Mecanium: an ensemble of musical automatons constructed from Meccano parts and activated by electro-motors, that play on acoustic instruments from all over the world. […] a timeless-sounding orchestra, both futuristic and slightly dada, conjuring ancient traditions in its surprisingly sensuous music.” This description, with its surfeit of adjectives, is actually an apt and accurate portrayal. For performances and recordings, Pierre sometimes adds live vocalists or instrumentalists (including himself on brass and other instruments) working in understated sympatico with the Meccano machines.

Next on the list: Tim Phillips, and in particular Tim’s mechanical sequencer. From his website: “This is a mechanical instrument that can be built, customized and deconstructed while it is being played.” It’s the real-time manipulation of the sounds that makes Tim’s machines particularly great, along with the fact that the moving parts and their relationships are so engagingly observable.  The sequencers have multiple motorized gadgets which can work independently, but which also can be synced together by means of large wooden gears, easily configured and reconfigured by hand even as the machines operate.  

And one more music machine artist: Ernie Althoff. I’m thinking particularly of the machines he made in the 1980s and ‘90s using bamboo and other lightweight materials, powered mostly by the rotary movement of cassette machines. The wonderful thing about Ernie’s devices was the delicacy and fragility of their sounds, patterns and mechanisms. You can see the article we had on Ernie’s bamboo machines in the Experimental Musical instruments journal in 1990 here. There’s only a smattering of other information on these machines online, but for related works you can try googling Ernie Althoff’s metal machines or Ernie Althoff solar machines.


I made the first of my own rotary music machines, called Onstinato Machine, in the late 1980s. You can see it in the photo on the right, and it is also the featured instrument in this video.  This machine was pretty big, and fairly elaborate, but it retained the quirkiness and homemade feel that gives such instruments their character. The rotating component was a central upright pipe about five feet high, with arms extending out horizontally to activate sounding objects positioned circumferentially around. It was turned by a strong (and somewhat noisy) speed-controlled motor. A typical speed for it would have been about 10 or 12 RPM, which gives a rotation period of about 5 or 6 seconds. The sounding elements included four five-string zithers each tuned to a chord, made reasonably loud by virtue of large Styrofoam picnic coolers acting as sound radiators. There was a rumba-box-like four-tine lamellaphone, also provided with a styro radiator (a very large one to provide convincing bass). And there was percussion in the form of sandpaper scrapers, claves, tambourine and the plastic equivalent of a hollow log drum.  All the components – both activating arms and sounding elements – were finely adjustable in their position in the rotation, meaning that all was rhythmically programmable. It had a lot of fun features; for instance, the rotating arm which had the plectrum for the zithers was followed in sequence by another which had a paint brush to damp the strings of each zither at the correct moment after it had been sounded. Like all such instruments, it was fun to watch, especially since it wasn’t overly rigid and had a kind of wobbly quality which struck observers as comical. It was originally designed to accompany one particular song, and when first constructed it was configured to play an ostinato that served as the song’s accompaniment. It could later have been retuned and reprogramed for other musical patterns, but it never was: for its entire year life of 30+ years it never served any purpose but the accompaniment of that one song. It saw a fair number of performances though, culminating a year or two ago in the filming for in the above-mentioned video. Following the making of the video, having thus finally documented and memorialized the instrument, I dismantled it. Why? Because it had been occupying a lot of valuable real estate in my limited storage space (remember, it was big).

My more recent rotary music machine is much smaller; it sits nicely on a tabletop. In conception it’s much like the large Ostinato Machine described above: a central rotating pillar with arms outstretched, with sounding objects located circumferentially around. I’ve been calling it Citphto O.M., because that wonderfully peculiar word is written on the speed controller I got for it on Amazon. (Apparently the controller is produced, or at least sold, by a company called Citphto. The O.M. in the name stands for Ostinato Machine.) In making Citphto O.M. I wanted to create something smaller and more convenient, and also something truly easy to configure and reconfigure for different rhythmic patterns. To make reconfiguration and adjustment as easy as possible, I stole an excellent idea from Tim Phillips, maker of the Mechanical Sequencer described above. The instrument is built on a base plate of steel, and the sounding objects are given magnetic bases. The objects can effortlessly be placed anywhere on the plate and they’ll stay put. The base plate is 20” square. It’s marked with concentric circles which function as tracks for different rhythm patterns. You can have an arm reaching out a certain distance from the rotating upright so that it activates objects positioned within the circular track that is that distance from the center. (The 3-D geometry is actually a bit more elaborate than that. To avoid hitting objects placed on inner tracks, an arm for an outer track has to be positioned high enough to reach over objects in inner tracks and then down to activate objects in the outer track.) The plate is also marked with radial lines which divide the circle into various equal divisions; e.g., four radial lines spaced at 90 degrees might correspond rotationally to quarter notes in a 4/4 meter. These lines can serve as guides for object placement to help with good timing as the arms rotate past. The lines are color-coded to indicate divisions of 3, 4, 6, 8, 9, 12, and also, for Brubeck fans, 5 and 10. Some of these divisions may never get used, but there was a certain satisfaction, both conceptual and aesthetic, in laying them out.

Where my earlier ostinato machine had been heavy and clunky, I wanted everything about this newer one to be not only smaller, but light and flexible, easy to activate with minimum power. The wish for lightness was partly inspired by the delicate character of the Ernie Althoff machines mentioned above.  I had expected that this would make for a rather quiet machine, but no; the Citphto O.M. has turned out to be reasonably loud.

A few more technical notes on design and construction: 1) It helps to use very strong magnets to secure things in place. Neodynium magnets are available and work well. 2) One of the challenges in making Citphto O.M. was finding a suitably quiet yet strong-enough motor and matching speed control. The noise question arises in part because, unless you want to get into a lot of crazy down-gearing between the motor and the rotating pillar, you’ll need a gear motor, and these tend to be louder. Yet once I had spent some time wading through the weeds of available motor types, the motor question did not prove to be intractable. If you’re interested in details, be in touch by email and I can let you know what I came up with. 3) Along with suitable motor and speed control, the instrument calls for a face plate (to attach the motor’s drive shaft to the upright pillar) and a bearing or two to support the pillar and allow it to turn. It’s important to get the details on these right; for instance, the insert hole on the face plate has to fit the motor shaft diameter. 4) For good magnetism the steel plate shouldn’t be too thin.  The plate I used is 3/32” thick. This is the only thing in the machine that is heavy. 5) I used lightweight spring steel bands for various purposes in the activating arms and some of the sounding elements. These have proven very useful. They can easily flex to scrape across or dance over or sway and strike the objects they encounter, reducing the need for other sorts of more rigid mechanics, such as levers and pivots.  

As for the sounding elements: anyone making such an instrument can have fun dreaming up and experimenting with various sounding possibilities. The challenge for a small instrument like this is to come up with things that are small yet effective. Here’s what I ended up with: Small pieces of sandpaper for sandblock sounds, modified combs and threaded rods for flavorful scraping sounds of varying durations, small metal containers struck by hard beaters (with the containers tuned by gently hammering a slight curvature into their flat bottoms), pill bottles as shakers filled with BBs, a mallet with a small, hard head which strikes the steel plate, another such mallet with soft, heavy head, and lamellae – similar to individual kalimba tines – tuned to various notes.  The lamellae are particularly interesting: it might seem that they’d require a fairly elaborate mounting and that they’d too easily get pushed out of position in the playing. But thanks to the extraordinary strength of the neodymium magnets, I was able to make them really simple: For each one I used a piece of the spring steel band material, 1” wide, moderately heavy and rigid at .021” thick, and between about 3” and 5” long.  Using a vise, I gave the band a slight angle bend, maybe about 30 degrees off of straight, at a point along the band such that one leg of the angle covers the length of the rectangular magnets I was using. After placing a pair of the magnets side by side on the steel sheet, I put the bent steel piece on the magnets, leaving the angled-up portion as a pluckable overhang.  Believe it or not, this attachment method is strong enough for the overhanging portion to sound with clear pitch.  Not a perfect tone with long sustain, but good enough to provide the needed pitched element. 

Of course, many more types of sounding elements beyond those just described might work well. Given the easy interchangeability of elements on the instrument, it’s possible I’ll come up with more types in the future.

The finished instrument does pretty much what I wanted it to. It can play whatever sort of repeating pattern you program into it by the placement of sounding elements on the plate and configuration of activator arms on the rotating pillar. All is adjustable and tweakable, making it possible to have reasonably accurate timing, or slightly off timing if you prefer that effect. It’s easy to add, remove, or adjust sounding elements. The sound, as mentioned, is reasonably loud, and the motor admirably quiet. The period – that is, the time for one rotation – ranges, depending on motor speed, from about four seconds to most of a minute. Interesting to note that the effects you get from fast speeds are very different from those of very slow speeds: what is comes across as a catchy groove at high speed becomes an exotic exploration of textural qualities as slow speeds, especially if you have a lot of scraping sounds of longer duration. 


Thus ends my short overview of simple rotary music machines. They’re easy, fun, rewarding and highly recommended for home-builders.

Share This