RIS-517 - Tweezer Squeezer, by Chris Stokes
One of the physicists at Rowland, Mike Burns, was looking for a way to hold micrometer-scale wires for experiments on electromigration. The requirement was to be able to grasp one of these wires with an existing micro-manipulator X-Y-Z positioning system, without introducing human-scale vibration. Tweezers exist which can accomplish this on a micron scale, but these are quite expensive, so we decided to see if we could produce something a little better.
Part of the requirements were for the tweezers to mount easily on and off the positioning system, and perhaps to be hand-usable. Initially, I tried a straight-forward approach with a small motor and a fine pitch bolt mounted to it.
I had difficulty with the motor size and torque, given the need to overcome stiction for the fine pitch we wanted to use. Eventually, the motor began to outsize the tweezers themselves and I started looking for a different solution.
Nitinol wire (often called Muscle Wire®) looked like a reasonable answer. The particular tweezer we chose, Technitool Stainless Steel #7's, required approximately 100 grams for closure. Maximum pull for 0.006" dia. Nitinol is 330 grams. That provided a decent margin for error, and should conservatively allow the wire run for a few hundred thousand cycles. Mike was working with thin wire, about 3 microns, so a reduced-distance tweezer opening of 0.15” would be more than sufficient. The wire contracts by about 5% of its length, so 20 x 0.150 gives a minimum 3 inches needed for closing the tweezers.
I planned to run the wire out and over a quarter-round piece of teflon, down through a hole in the tweezers, double back around and back up over the teflon quarter-round. The wire must remain insulated from the tweezers, so I mounted it on a nylon bolt set perpendicular to the tweezer's long axis. This bolt sits on tiny bracket that allows you to adjust closure and tension.
Terminating and joining the Nitinol was a bit of a challenge. Soldering is unreliable, as the Nitinol has a lot of oxide which makes things difficult. Clamping between bolts seemed suspect to me as the wire would be heat-cycling, so I chose to terminate with crimping lugs and then make a conductive contact to the lugs.
Three inches of 0.006 Nitinol gives about 3.8 ohms resistance. The Images SI folks say that 400mA through the wire when it's sitting at room temperature will raise the temperature sufficiently to make it contract. 400mA/3.8 ohms = 1.52V drop that I should expect across the wire.
I asked Win about a good controllable current source, and he recalled one he'd written about on a news group years ago.
Win's current source required a current sink of a few milliamps, which we implemented with a 10k resistor, but this wasn't quite effective at low voltages, so we made a better one with a 2N3904 aimed at about 4mA. This makes a more flexible single-supply current source. I used RadioShack's elegant little “power supply in a plug” (6V 1800mA, pin 273-1763).
The first problem I encountered was the fulcrum point. I made it out of a nylon screw cut in half to give a gentle turn to the Nitinol wire, and then epoxied it onto the bottom surface of the tweezer. The melting point for nylon is about 400F, but with pressure, and the thin wire, it seemed to eat into it at a much lower temperature, even around 170. The nylon fulcrum bolt then bound up the Nitinol wire and the tweezers wouldn't close.
To solve this problem I swapped that bolt out for a metal pin with a teflon sleeve. This gave a bit tighter radius, but better temperature resistance and less flex.
The next issue was very slow action, like 1/2 a second to close, and 5 seconds to open, plus about a minute to recover to a fully open position. I addressed this, probably in the wrong way, in retrospect. I changed to a thicker diameter Nitinol wire and ran more current through it. This did close quicker, but it still didn't open quickly, so I added a bit more resistance right at the closure point of the tweezer, by gluing in a bit of open-cell foam to the inside of the tweezers. This gets the tweezer started opening a bit quicker, more like a second to open, and five seconds to fully recover.
Next time I'd make a longer Nitinol wire run, I think. Start with a normally closed, reverse action tweezer that is teflon coated. Opening force for that would be more predictable than closure force. A longer Nitinol wire would also relax quicker.
The finished Tweezer Squeezer works well, and takes on the order of a 1/2 second to close, and about 2 seconds to open. The setup works best if the closure point is tuned by turning the pot, watching for closure, and then using the on/off switch from then on. Mike says he's eagerly awaiting the improved version.
In this version, I allowed for a longer run of the nitinol wire by zig-zagging it along the side of the tweezer. I epoxied a series of small copper tubes with teflon tubing as insulation. The wire routes back and forth like a shoelace. This gives a little more length and thus cools a little more quickly. Mike's impression is that it's much improved.