Let's Make Robots!

Flexinol Actuators

Prototype Nitinol muscle wire devices

I've been really interested in Flexinol lately (aka Nitinol Muscle Wires).  I've come to a nice stopping point where I've put together some basic actuators and a driver circuit so I figured it was a good time to post.

The embedded video shows two projects.  One is linear, moving in a straight line like a solenoid.  The other rotates on a pivot point like a servo.

For those unfamiliar, Nitinol is a shape memory alloy.  When made into Muscle Wire it will respond to heat and contract when it reaches a specific temperature.   In these circuits the heat is generated by running an electric current through the wire.

The control circuit for both devices is built around a PIC16F690 microcontroller and a ULN2003A Darlington array.  All the micro does for this demonstration is toggle power to the Darlingtons at intervals of about two seconds.   Flexinol actually needs a good bit of juice to do its magic- how much varies by the gauge of the wire.  Here I'm providing 200ma to the rotating actuator and 320ma to the linear actuator.

Nitinol is interesting stuff.  It's been around since the early 1960s and even though the potential for robotics is fairly obvious, you don't see it used in too many projects.  I've seen Nitinol described both as "a solution looking for a problem" and "a revolution waiting to happen."

Rotating Actuator


Flexinol has some very attractive characteristics.  Since the Nitinol shape memory effect does not rely on magnetism for movement, there is none of the electrical noise you get with servos and other motors.  Flexinol is also very strong but adds virtually no weight to a project.  The largest gauge available is .020" (.51mm) which can lift 7.85 pounds (3560 grams).  These projects use .004" Flexinol which will lift .31 pounds and .005" Flexinol which will lift .49 pounds.

Working with Flexinol does present some significant challenges.  Solder or glue won't hold for long so you have to crimp it.  I'm using #2-56 machine screws and nuts as adjustable crimps and they seem to be working out pretty well.  They're also good general fasteners and electrical conductors so I'm using them for all the attachments.  Regardless of how it is crimped, ultimate failure at the crimp joint is a chronic problem with Flexinol projects.  Another big problem is cycle time.  Nitinol needs to cool off in between contractions and with the heavier gauges that can take many seconds.  Cycle time is not as much of an issue with these actuators because of the relatively small gauge of wire.

Although it contracts with a lot of force, Flexinol doesn't move very far.  Depending on the configuration it will usually shorten by about 4-5%.  Both of these devices apply mechanical advantage to convert some of the available force to movement over a greater distance.  The leverage applied in the rotating actuator is probably familiar to most people. The linear actuator uses a less common technique called an angled pull to trade some force for greater distance. 

One significant challenge in building even simple Flexinol projects is that the wire does not return automatically to its elongated shape when cooled, but needs to be stretched back to the original length.  This stretching is usually accomplished with some type of spring or bias weight, or even another piece of Flexinol.  Finding the right spring for an application can be tricky.  It needs to be strong enough to pull the Flexinol back into shape, but not so strong that the Flexinol can't move to begin with.  There's some science to the selection, but when working with found parts or designing yourself, there's also some trial and error.  The linear actuator shown here uses a compression spring that was salvaged from a ball point pen which is performing quite well.  The rotating actuator uses a crude extension spring that I fashioned from K&S .020" music wire.  You'll see a lot of music wire springs in Flexinol projects.  Like the machine screws, music wire is conductive and in this actuator is part of the circuit. 

The main structures of the actuators are made from Plastruct styrene plastic.  Styrene is available from hobby shops and mail order.  It's very generic stuff usually marketed to scratch model builders, model train builders etc.  For these projects I used .080" sheets and strips, 3/16" diameter tubes, and .100" diameter rods.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

The first time I saw nitinol/flexinol my brain started spinning trying to consider all the possibilities, but on closer inspection nitinol is pretty much useless for battery powered robots, unless you are trying to make some really slow and inefficient robots.

To make an effective actuator you need a system that efficiently heats and then cools the muscle wire. With a simple wire in air, there is a trade off between how fast you can flex it and how fast you can relax it.

An interesting approach would be to keep the wire in a liquid bath, and pump the liquid between a cold and a hot reservoir. The heat could even be provided by the Sun.

Now that's a "Hmm..."

I wonder if there's any shop in Europe selling nitinol...

Nice write up and vid........

If you were to measure the resistance of the wire ....does the resistance change whilst you stretch or bend it manually.(ie without applying any electrical signal) ?

I am look for a materials that change resistance when flexed ...... this looks like one !!!

I'm not sure if there is much potential to use Flexinol as a flex sensor.  Different gauges have a different resistance.  The .005" is about 75 ohms per meter.  Flexinol manufacturer Dynalloy provides a lot of technical information at its website http://www.dynalloy.com.  According to Dynalloy, the resistance goes down as Flexinol is heated through its transformation temperature and contracts - partly due to the shortened wire, and partly due to the fact that the wire gets thicker as it shortens.  Following this logic, if you started with Flexinol in a contracted state (with or without electricity applied at that moment) and then stretched it, the resistance would increase very slightly as the wire became thinner.  However, you would need to apply heat to the Flexinol to make it return to the contracted state before you could repeat the operation.  Do these qualities seem like they would be useful in your project?

I just got some of those Nanomuscle rotary units. Hopefully they can prove strong enough with the current I'm willing to give them. "A problem looking for a solution" indeed.

There needs to be more video of this stuff in action. It just has that intangible "Hmm..." quality when you see it working. Nice primer on the material itself as well, nice work.

Thanks for the feedback JAX.  I agree that it would be great to see more Flexinol projects and video available.  I think one of the main reasons we don't see more people using it is because there aren't a lot of resources available to help them get going!  Obviously I'm just getting started experimenting, but I really do think there is some great potential here.  Hopefully you have good luck with your own project.  Are you working with raw wire or a pre-assembled actuator?  Are you trying to limit the maximum current, or just the total current consumption?  One method for saving total current without sacrificing speed is to apply an initial burst of high current to get the Flexinol to contract, and then pull the current back a touch to maintain the contracted state.  I've started playing around with a PWM circuit to get that kind of control; it's working but it's not quite where I want it yet.