Thursday, February 17, 2011

How to build an air muscle and use it in a force-feedback joystick

In this video, I describe how to build an air muscle, and use it to create a force-feedback joystick

Link to the instructables article (not written by me):!/

Silicone tubing - McMaster 51135K281
Braided wire loom (polyester) - McMaster 9284K413


"The worlds forgotten boy." said...

Care to share the specs and source for the Manifold and Pressure Transducer setup? :)

Ben said...

Forgotten boy, I got the pressure transducer on eBay and the valves/manifold new from Clippard ( The transducer is a Fairchild T5700 series. There are a bunch on eBay now, but more expensive than usual. The model that I have takes 0-10V and outputs 3-120 psi.

The valves are 3-way normally-open Clippard "15mm series" valves - E315F-2C012 and are mounted on an aluminum Clippard manifold.

Good luck! -Ben

Anonymous said...

Pretty cool... this may be a great way to do force feed back in a driving seat simulator.

With the seat mounted on a floating frame the air muscles could pull up on corners to simulate bumps or tilt the platform.

Any thoughts on closing the loop so that the extension of the muscle is measured and the air is modulated to keep a particular length set?


Anonymous said...

A great, how to article, I must say. Your other projects are fantastic also, keep it up. Will be reading your stuff regularly now. Remember blogger template designer is your friend :-)(+adjust width to full). Within a few minutes your blog will be a thousand times better. Add rating and share buttons to your blog posts+ a labels gadget to side column for better topic navigation. My 2 cents. Are you on twitter?

d-sier said...

Hi, i have a, probably stupid, but anyway idea.
Would it be possible to use PWN on the valves to control the pressure instead of using the transducer?

Ben said...

Anonymous, sorry I am not on twitter. Also, I am not much of a web coder, but didn't like any of blogger's templates when I started this blog a few years ago, so I made some changes to the template myself. I actually prefer a very clean look without any gadgets, buttons, ratings, etc. I never use any of those things myself on other pages, so I'd rather not have them. I realize that my preferences may not align with the majority opinion, but then again, this is my blog and I get to do it my way. I might like to have a way to show a "list view" of all blog posts, but that's about all I would want to change.

d-sier, using raw PWM to control an on/off air valve may work, but it would certainly be noisy and difficult to control. A better approach would be to smooth the PWM with a resistor/capacitor network, amplify the resulting DC voltage and feed it into a proportional air valve (search With a pressure-sensor, you could have closed-loop control. It would probably be cheaper and easier to use an off-the-shelf component like the Fairchild unit.

Ben said...

Anonymous, closed-loop position control could work with the air muscle, but pneumatics are fairly difficult to control in that way in general. The problem is that gasses compress, so making the muscle (or air piston/cylinder) move a predictable distance is pretty difficult. Hydraulics have the benefit of using an essentially non-compressible medium, so motion can be accurately controlled from a pump/valve.

To attempt position control with pneumatics, you will probably want to diametrically oppose air muscles (or use a double-acting cylinder) and fill both sides with full-pressure air to prevent movement when the system is at rest. Then, you need a separate fill/exhaust valve for each side so that air can be added or removed as quickly as possible. It could work depending on how much accuracy you need. -Ben

Anonymous said...

@Ben re. Close loop for gaming platform....

I've been thinking about this and I believe that using gravity as the 'return', a 'distance to floor' sensor (ultra-sonic?) and a 3D accelerometer (Wii nunchuck) should give all the data to calculate the position/angle of the platform. You can then modulate the air going into each muscle based on current position.

Got some tubing ordered, so standback for a massive 'fail' ;-) Gotta try to succeed I guess.

Ben said...

Anon, that sounds like a great project. Let us know how it turns out, or if you need any help with it.

Anonymous said...

Want to know how to (not) scare the cr@p out of an 8 year old...? Better check the clamping on the ends of your air muscle. Tubing goes 'pop' real loud at 80psi :-)

It might make sense bonding the filaments of the mesh together with super-glue (or alike) underneath the clamp.

Next step(s) is to quantify the contraction and attach servo to air valve.

Keith Cooper said...

Simply awesome! I'm just getting into pneumatics, and I too was inspired by an Instructables tutorial ( I'm making this for my grandkids. I'm also adding a scare to Halloween this year with a skeleton that will jump out of a coffin. All of this got me looking at pneumatic muscle technology which led me to your videos and site.

Sorry to ramble on, but this is just so cool, and I had to say thanks!

aditya said...

Hello Ben your project and work is so excellent.But I have following query about the Electronic product.Can you please help in regard with that
My Query is regarding Electro-Pneumatic Transducer of the series Fairchild T5700 series.
This product as Micro compressor embedded (Inbulit) in the system.
Or this acts just controller.
Do we have to provide Air supply from out side.
I am actually looking for Air-Compressor will this product severs my purpose.
Looking forward to hear from you
Thanks and Regards

Ben Krasnow said...

aditya, the Fairchild T5700 does not include a compressor. You must supply compressed air to the T5700, which will then regulate that compressed air to a specific pressure. Choosing a compressor will depend on required flow rate, required pressure, electrical supply, noise requirements, vibration requirements, etc. In other words, it's an engineering task. For bench experimentation, simply use a hardware-store compressor for shop air tools.

Anonymous said...


Thanks for your excellent blog! Did you finish this force-feedback joystick? Would you consider showing a video of it?



Ben Krasnow said...

I finished building the joystick and sold it to a customer. Unfortunately, I cannot make a video about it.

aditya said...

Hi Ben
Thanks for the reply.I also constructed air muscle after seeing your inspiring video and it is working very fine.Now I would like to more about the valves.If suppose if I have multiple air muscle air ...The valves that showed in the Video which are connected to the Pressure transducer.How they will work?For each valve can we connect only air muscle??or for one valve we can connect three air muscles?Could you please help in this.Thanks in advance

Post a Comment