Pneumatic Multiplexer- 3d Printed




About: Sebastian Morales is a Mexican born artist, engineer and researcher based in NYC. His work often exists in the confluence of sculpture, kinetics, the human body and code. His work can be seen at adorevolutio...

What if you could control hundreds of outputs with just a few inputs? I am not talking about Charlieplexing, or at least not in the traditional electronic way. I am talking of doing this in a mechanical manner where the outputs could be pistons, valves, bladders... in one word, actuators.

With this project I seek to create an efficient way to control large number of actuators in an independent fashion. Being able to move large numbers of components with a relatively low number of inputs.

Step 1: Motivation/Inspiration

By now, chances are you have at least seen one out of these 3 projects (see videos). All a little different from each other but all amazing in their own way. They all have a couple of things in common, nonetheless I'll focus just in one: They all move hundreds (sometimes thousands) of components independently.

For these type of projects to work quantity is paramount; quantity equals resolution and that higher fidelity allows the viewer to be captivated by the piece.

Through this project I wanted to build a system or a tool that would allow people to control large quantities of outputs using only a few inputs, in a more descriptive form, Outputs=(Inputs/2)^2.

Step 2: Initial Investigations

It all started with me thinking about gases, we all know gasses expand when heated and contract when cooled. However, in order to achieve a significant change in volume you need a dramatic change in temperature (Check ideal gas law), unless... Unless, the gas is changing state. When a gas change state from liquid to gas it expands hundreds of time its volume; a liquid drop can be enough to perform a hole lot of work, just think of a car engine for example. Of course I didn't want to use something that chemically transformed, that would require a system to keep adding fuel, removing waste, making everything a lot more complex than the original problem.

I started looking at all substances that are liquid at room temperature but will transform with just a small amount of added heat. They exist! Most of them are very flammable, many of them are toxic... I did however found some hope in refrigerants, in specific R123. This chemical has the nice characteristics of being liquid at room temp but just adding a little heat it will boil. Removing the heat would make it liquid again transforming its volume dramatically in the process.

Challenges with refrigerants ultimately swayed me away; refrigerants of these type are in the process of being phased out since they damaged the ozone layer, meaning that I would need to create a system where the fluid could be perfectly contained without leaks and recovered safely. I would need help to handle it and all of a sudden the simple solution I was trying to achieve was looking quite complex and not very responsible from an environmental perspective.

Time to think of something else, something that made a little more sense, something a little more logical.

Step 3: Air Logic

What is Air Logic?

Air logic is a field of pneumatics very similar to electrical circuits, where you can have relays, limit switches, AND gates, NAND gates, OR gates, amplifiers, and so on. The only difference is that electrical current is replaced by compressed air; think about it as a mechanical circuit (here is a great article describing air logic).

The idea was to create a grid where you could individually address any square in it by controlling the states of rows and columns. All the members in a row would be connected together by one line and all the members in a column would be connected by another line. A switch in each square would open only if the combination of pressure between both lines was the right one.

Creating my own valves:

In the spirit of keeping things simple I first started designing a NAND gate, where a rubber (BUNA) bearing would make a seal and only allow air out if the bottom line was low and the top line was high, (see diagram). After a couple of iterations I was still having trouble creating a good seal. Since the bottom lines had to be most of the time high, I was loosing too much pressure in the system making it hard to actuate.

Step 4: 3D Printing - Complexity Is Free!

As engineers and designers we often facinated simplicity, often forgetting that modern tools like 3D printing often offer higher degrees of complexity for free.

I was using the Object (resin based) printers, which have an incredible resolution and are able to create fully air-tight systems. Thinking of ways to create a better seal, I decided to go for a more traditional approach used in the world of seals... O-Rings. Instead of trying to make a seal with the Buna bearings I decided to use a piston type of valve, where three O-rings ensure a seal against the walls of the cylinder. Having high pressure in the bottom line will rise the pistons aligning the guides with the top line, if then the top line is high air will flow through the output.

Step 5: 3D Printing and Assembling

At first I was modeling and printing the fittings, this would work quite well for some time since you wouldn't have to tap each hole and place a plastic fitting to connect to the tubes. The problem was that after a while the fittings would become very brittle and break off, making it impossible to use that port without drilling, taping and inserting a new fitting. At the end making it a lot more work than tapping the holes from the begging.

I did a couple of iterations, until I was comfortable with the design (I though). For the testing I was using smaller grids of 4 by 4 for a total of 16 outputs. I was also using long animal balloons since these would very easily allow me to monitor air flow.

Step 6: Testing

I did a couple of iterations, until I was comfortable with the design (I though). For the testing I was using smaller grids of 4 by 4 for a total of 16 outputs. I was also using long "animal" balloons since these would very easily allow me to monitor air flow.

It took a while to get it working, lots of leaks in the process and a lot of learning on how to make 3D printed parts seal.

Step 7: Omnifarious

Omnifarious is a kinetic sculpture demonstrating the concept of shapeshifting objects. In this case a sphere would be able to transform its surface to render different geometries.

In this case the sphere (hexecontahedron) has 59 balloons, imagine what would happen if that number was in the thousands. The resolution would be enough to physically recreate a diverse set of geometries in a matter of seconds, changing the way we normally think of displays.

Step 8: Ideas Are Blooming...

There is still a lot of work to be done for this to really get to where I want it. Nonetheless my time as a full time artist in residence at Pier 9 has ended. It has been an extraordinary experience in so many ways that I won't even start listing.

I hope you enjoyed my instructables as much as I enjoyed making the projects.




    • Pets Challenge

      Pets Challenge
    • Fandom Contest

      Fandom Contest
    • Sensors Contest

      Sensors Contest

    15 Discussions


    7 months ago

    is there any possibility to get 3D files?
    thanks in advance



    2 years ago

    Which pressure can be used in a 4x4 (16 out) valve??

    Which flow output than can be reached?

    There are moulded or steel versions too??

    1 reply
    Sebastian MoralesNuvola

    Reply 2 years ago

    The pressure you can use is dictated by the properties of the materials used. In this case I was using Vero White (

    The flow will be dictated by your pressure and the diameter of your internal tubes.

    Keep in mind that this was a prof of concept prototype and not a fully working comercial product.

    If I were to re engineer it I would do several changes.


    2 years ago

    very very clever. very inspiring.

    i was wondering if you had any input on what one could change in your design to lock each location.

    what i mean is apply control pressure to x and y once, which locks something in place, then once pressure is released this something remains in place and the air keeps on flowing. this would allow for a faster control for the other pixels.

    maybe a steel ball gets pressed again a magnet, and with a 3rd control air flow the ball could be forced away from the agent?

    2 replies

    Hmmm, not really sure. I do see what you mean about the speed, it would actually be great and I did think about it for a little bit but gave up for simplicity's sake. At some points I started thinking about electromagnets (solenoids), these could do it all since you could cycle much faster (electricity vs air). Then you don't really need to latch since you could virtually open multiple at once. The problem is that the cost skyrockets defeating the purpose. Let me know if you come up with something!


    thanks for your reply.

    i have been thinking about this more and came across the field of fluidic logic and specifically this bi-directional fluidic flow control valve:

    if one would try to use this valve air would constantly flow our of all valves, but one could control if it comes out exit A or B, A being your ballon for example and B being some exhaust pipe.

    the only question now is how to control the control lines27 and 28.

    you are right solenoids are too expensive, and also if one would want to use water air or liquid control would be better.

    Sebastian Moraleszazu_247

    Reply 3 years ago

    The files are not ready to share since there needs to be more experimentation with tolerances to make proper seals. If you still want them send me an email or an inbox message and I can share with you what I have.


    3 years ago

    wow ! inspiring work !

    It is not quite a shuttle valve since fluid will not flow from one channel to another if the valve is closed. It is closer to a solenoid valve where instead of electricity you have fluid.


    4 years ago on Introduction

    does this pneumatic Multiplex array hold the same disadvantage as an electronic array whereas LEDs in a multiplex array would need to be quickly turned off and on in order to light multiple LEDs?

    2 replies

    Reply 4 years ago on Introduction

    I'd like to put it out there that this project is pretty amazing.

    With the problems with multiplexing, particularly problems I'm thinking may arise with air logic that aren't a big deal in electronics, is it possible to create significant pressure differences within one row? Does your current build automatically (unintentionally) lower the pressure of other cells in a row when one is opened?

    Unfortunately it does, the disadvantage is even greater since air moves significantly slower than light. In the video however, often more than one balloon inflates at the same time, this is because you can actuate entire rows at the same time. This would allow to move hundreds of components in just a couple of seconds. In theory you could also do all at once.