Membrane Criteria

The toroidal membrane is the primary structure and source of friction of the drive system. This component was determined first because the design of both the brackets and contractile rings are affected by the properties of the chosen membrane. Four qualities were considered during membrane selection: rolling resistance, heat resistance, ease of manufacture, and ease of adhesion. Qualities were rated on a scale of 0-3 where 0 is unusable and 3 is exceptional. Three toroidal membrane materials were evaluated: Ecoflex 00-30, low density polyethylene (LDPE), and vinyl. The Ecoflex and LDPE membranes were fabricated in-house and the vinyl FFTs were purchased online under the product name “Water Wigglies”.

Membrane Evaluation

	Ecoflex 00-30	LDPE	Vinyl (Water Wigglies)
Heat Resistance	3	0	2
Roll Resistance	1	3	2
Ease of Manufacture	1	2	3
Ease of Adhesion	1-3	0	3
Total	6	5	10

Ecoflex 00-30: The Ecoflex toroidal membrane was fabricated by inverting a long thin-walled tube onto itself. While this membrane had superior heat resistance, it proved difficult to manufacture and invert. Furthermore, Ecoflex bonds well to itself and other silicon rubbers, but is difficult to bond to other materials. This would have resulted in additional design constraints on the brackets.

 

LDPE: The LDPE toroidal membrane was fabricated using thermally bonded film. This membrane had the lowest rolling friction of the materials tested; however, the SMA actuators melted through it almost instantly. This fact alone was enough to invalidate LDPE as a viable membrane material.

 

Vinyl: Vinyl FFTs are commercially available on Amazon.com and are reasonably invertible toroidal membranes. Like LDPE membranes, vinyl can be melted by the SMA actuators. However, these actuators melted in minutes rather than seconds. This was deemed acceptable for prototyping purposes. Rolling friction was significantly reduced by lubricating the FFT with dishwashing soap.

Future Work

Vinyl FFTs are the best of the three membranes evaluated; however, the melting problem proved significantly more detrimental than anticipated. Future designs will focus on the development and implementation of non-thermoplastic membranes. Currently, the viability of the Ecoflex membrane is being reinvestigated. The low scores in “ease of manufacturing” and “rolling friction” are potentially a result of the manufacturing process used to fabricate the membrane and not the material properties of the rubber.

Geometry

The brackets make it possible to connect the SMA contractile rings to the drive system without gluing the wires directly to the membrane. This fabrication method allows actuators to be reused or transferred between prototypes with minimal risk of damage. Additionally, the use of brackets provides more control over the interaction between the actuators and the membrane.

Our team hypothesized longer brackets would be more efficient at inverting the torus. It was assumed the long arm of the bracket would act as a second class lever, pushing the torus inward. Brackets were designed with and without the lever arm.

	Brackets without the lever arm are referred to as point brackets. Brackets with the lever are referred to as double brackets. The length of a double bracket is considered to be its center distance. Three brackets were tested, point brackets, as well as two lengths of double brackets, 0.25" and 0.375". All brackets were fabricated using an FDM 3D printer.
	In addition to bracket geometry, two brands of FFT were tested. These FFTs had 125mm and 110mm circumferences.

Experiment

An experiment was designed to quantitatively assess the contractile force needed to invert the vinyl toroidal membranes and determine whether this force was affected by bracket geometry. A drawstring-bag-like pulley system was used to simulate contractile force. The force and extension was measured using an instron material tester. In each test, vinyl FFTs were contracted about 45%. Forces were averaged over three tests.

Results

Results of the experiment do support the hypothesis that longer brackets invert the torus more effectively. However, despite the apparent correlation, the point brackets on the smaller torus required approximately the same force as the long brackets on the larger torus. This suggests the smaller torus is easier to invert overall. We were unable to evaluate the effect of double brackets on the small torus because of difficulties with membrane adhesion.

Final Bracket Design

Both membrane sizes could be inverted using similar amounts of force. However, the larger membrane required the use of double brackets. These were fabricated using a 3D printer and often melted during testing. As a result, they were deemed unsuitable for the prototype. Point brackets were fabricated from thin coils of spring wire. This bracket was impervious to the heat generated by the SMA actuator and selected for the prototype.

Minimum Force

To establish the minimum force required to invert the torus, a vinyl FFT was lubricated and inverted using progressively weaker actuators. It was determined the required force to reliably invert the torus is 0.3N.

Future Work

Development of SMA compatible double brackets is being investigated. Possible manufacturing processes include SLS 3D printing and resin casting.

Nitinol

The contractile rings are composed of coils of Nitinol (NiTi) wire. NiTi is a nickel titanium alloy that changes shape when heated. It accomplishes this though phase transition between austenite and martensite. In austenite, NiTi has an organised crystal structure where nickel and titanium atoms are held tightly in place. In martensite, the atoms are more loosely held than austenite and the structure is able to deform slightly without breaking any atomic bonds. When deformed martensite is heated it transitions into austenite. As the atoms reform into austenite the NiTi returns to its original shape.

The original shape can become stabilized through a process called shape setting. In shape setting, the NiTi is fixed in a deformed state and annealed to reset the structure of the austenite. The contractile rings were annealed at 390 degrees C for 20 minutes. [7]

 

Coil Actuators

NiTi can only recover from about 6% strain. In straight NiTi wire the stroke length is equivalent to strain; however, in a helical wire, the relationship between stroke length and strain is nonlinear. In helical wire, stroke lengths of 200% to 1600% can be achieved while deforming the wire 6%. The stroke length and strength of the coil actuator is largely dependent on four parameters: diameter of actuator wire (d), diameter of coil (D), initial pitch angle (⍺), and number of active coils (n). [8] There are caveats when designing SMA coil actuators. Stroke length of an actuator is typically inversely related to its strength. [8] Additionally, NiTi actuators are monodirectional and require an external bias force to be reset following actuation. SMA coil actuators need enough strength to overcome bias force in addition to the strength needed to perform the desired task. The net force is the force available to perform work.

Designing The Actuator

The contractile ring is simply a coil actuator bent in a circle. Through observations of the FFT it was determined the relaxed ring actuator should have a circumference of 110mm and contract down to 30mm. The ring must also be capable of maintaining a net contractile force of 0.3N throughout actuation.

The Actuators

The acutaor ring will contract until the net force falls below the 0.3N needed to invert the FFT. Through some trial and error we were able to design a coil actuator that roughly met all of the requirements outlined in the section above. The final actuator design has an actuation strength 0.62N and requires 0.28N to deform to a length of 110mm. Actuators were fabricated out of 0.15mm NiTi and had a coil diamiter of 0.9mm. They have an initial length of 22mm and can elongate to 110mm while experiencing 4% strain. The contractile ring is made by connecting both ends of the actuator to a loop of magnet wire. The magnet wire allows the two ends to be held together without shorting.

Electrical

The contractile rings are actuated by 20 volts from a regulated DC power supply. NiTi is extremely difficult to solder. Therefore, power cables were connected mechanically to the contractile rings though the metal brackets. Each ring is powered by an individual mosfet and the entire prototype is controlled using an external Arduino Mega 2560.

Future Work

Currently, the prototype relies on the inversion of the torus to re-expand the contractile rings on the front of the drive system. This strategy is unreliable. The team is currently working to develop contractile rings that passively expand when not activated.

When Water Wigglies are shipped, the outer layer is slightly stretched. This makes them favor one position when inverting. To adjust for this, role the torus inside out and store for a few days. This evens out the membrane and makes it suitable for use.

Lubrication

	Clean factory oil lubricant off the FFT. Slowly inject 10-12ml of dishwashing soap using an adhesive syringe. Do not roll or shake the FFT! If the water and soap mix, it will be difficult to glue the injection site closed.
	Make a patch by dipping a piece of printer paper in cyanoacrylate glue. Remove the needle and quickly apply the patch.                 If the patch does not leak, lightly shake the FFT to mix the water and soap.

 

NiTi Coil Actuators

An NiTi coil actuator is fabricated by wrapping a strand of thin NiTi wire around a shaft and setting its shape. Each 22mm coil uses about 34cm of straight wire. A single 5m spool has enough wire to make 10-12 actuators.

Clamp 150um NiTi wire to a core of 0.024in music wire. Insert into drill press. Slide 3D printed pieces onto the core and push up against the NiTi. Place NiTi into V channel of the piece and pull taught. Turn on drill press and push up lightly on the 3D printed piece. This forms the coil. After coil is complete, clamp the free end using a nut and bolt. Set the shape of the coil by annealing it at 390 degrees Celsius for 20 minutes. After shape setting is complete, remove coil and cut into 22mm sections.

Point Brackets

Music wire is considerably stiffer than the NiTi wire and therefore must be secured more firmly. To create the point brackets, use steps 2-4 above, using 0.029" music wire in place of the NiTi and a 0.042" core. Keep fingers away from the winding coil. If released, the remaining wire can whip around and cause injury. After the spring is compete, cut every three coils using a dremel. Do not cut the wire with pliers. This will cause burrs that can puncture the FFT membrane.

 

.

Attach Brackets

(Image 1) Print out template from the zip folder and cut on the dotted line. Poke holes in the center of all circles and roll paper into a tube. Secure with tape. (Images 2-4) Roll torus though paper tube, marking all holes with a marker. When all available holes are marked, continue rolling torus through tube, marking remaining holes as you go, making sure to mark both inside and outside of torus. (Image 5) Carefully glue brackets next to each marking, making sure not to place glue directly on markings. Note: Glue on markings may not adhere well to torus. (Image 6). FFT with all brackets attached.

 

Make Contractile Rings

Stretch NiTi coil actuator to 120mm. Thread coil through a single row of brackets around the FFT. To tie the ends of coil with magnet wire, insert magnet wire though the center of 3-5 coils on each side and tie a knot. Use a strip of electrical tape to secure and prevent the ends of the coil from shifting and shorting durring inversion. Continue for all rows of brackets you intend to mobilize.

Wiring

• Take a length of magnet wire roughly 3X torus length and sand off the enamel.

• Choose a column of brackets alongside the magnet wire knots. Thread the sanded wire through the full column of brackets, making sure the wire contacts the actuator wire in each one. This will be the common ground.

• Take a length of magnet wire roughly 3X torus length and sand the enamel off roughly 0.25" - 0.5" off the ends. Repeat with wires for as many actuators as you intend to mobilize.

• Connect these wires to the brackets on the opposing side of the knot, ensuring each is individually connected.

• Connect all free ends of these wires to mosfets, one mosfet per actuator.

• Supply the mosfets with 20 volts from a regulated DC power supply.