For the current iteration of SMAKS, we switched from a horizontal, straight, tube, to an arc. In order to do this, we had to bend our acrylic tube, figure out how to secure it, and add more solenoids to accelerate the magnet against gravity.
In order to create a smooth curve in the tube, we built a plywood bending jig against which to bend the tube and inserted a silicone insert into the tube to maintain the inside diameter. It’s important to maintain the inside diameter so that the magnetic projectile can still travel through the whole tube. We initially experimented with tube bending using a heat gun but found that direct heat in a small area led to melted, cracked, or kinked tube. Our goal was a long, smooth, continuous bend so we looked for a solution to evenly heat up the acrylic tube at once. Our solution was to soak the acrylic tube in boiling water. Boiling water gave us the result we were looking for, but it was a long and difficult process involving boiling multiple kettles of water at a time and the tube-bender putting their hands in the boiling water (while wearing heat proof gloves).
For the base of the enclosure, we continued with the basic laser-cut box design we had used for the second iteration, but made it taller and added finger joints to hold the box together. We also etched a pattern evoking the arc of the magnet’s path onto the box for added visual flair. To mount the curved tube, we laser-cut plywood ‘spokes’ with a straight section to fit into the box, a loop to slide over the tube, and a living hinge toward the top to allow the spoke to bend and line up parallel to the tube. When designing our living hinges, there were two parameters to optimize, the spacing between laser cut slits and the overall length of the hinges. Flexibility is always a tradeoff with strength. Removing more material from the wood makes the hinge more flexible but also makes the hiner weaker. Our hinges were structural elements of our design so they had to be designed to hold weight. But, at the same time, our hinges also had to be flexible enough to conform to the tube. The compromise that we reached was longer hinges towards the outside of the arc where the hinges need to bend the mos and shorter hinges nearer to the middle. Living hinges can be extremely delicate, and we snapped many spokes trying to test their fit on the tube and while assembling the system. We managed to get them all in place for the current iteration, but they wobble as the magnet passes through the tube. This wobble reduces the efficiency of the kinetic energy transfer from the electromagnets to the projectile magnet making it move more slowly.
We had used a 4-channel MOSFET board for our second iteration, but since the curved tube had 6 solenoids, we couldn’t control them all from one board. It was simpler and less expensive to just purchase a second MOSFET board than switching to 6 individual MOSFETs, we wired 4 solenoids to the first MOSFET board and the remaining 2 to the second. We kept the Arduino UNO and benchtop power supply from the previous iterations, but added two IR sensors connected to the Arduino to determine the location of the magnet in the tube.
Our initial plan for the current iteration was to switch from timing the delay between solenoids triggering, which was a lengthy process that did not result in perfect repeatability of the magnet’s motion, to using 4 IR sensors, one before each of the 4 center solenoids, to turn them on just before the magnet reached edge of the solenoid. We later realized that we might have to place the first 2 solenoids on each side too close together for the sensor to work properly, and later broke 2 of our 4 sensors trying to solder magnet wire onto them, so we switched to manually timing the second solenoid on each side and using the sensors to trigger the third one on each side. We were still using manual timing to do the initial calibration of the system and determine the best placements for the solenoids, so we had a version of our Arduino code to just loop through triggering the 3 solenoids on each side of the tube. After hours of calibration, we found that the manual timing actually worked better in most situations than using the sensors to trigger the third solenoid, although we were unable to get it to a point where the magnet always made it across the tube. Since the sensors were already mounted and we weren’t using them to trigger the solenoids, we decided to use them to instead detect whether the magnet made it all the way across the tube and repeat the firing of the solenoids if it didn’t, that way we would not have to wait for the entire loop to repeat before it tried to move the magnet again.
We now have a curved tube that can accelerate a magnet up and over the curve from one side to the other relatively consistently. The sensors check to make sure the magnet has made it across before determining which set of solenoids to trigger. The whole setup has a finished look and is fun to watch.
created with
HTML Designer .