This the final stage post on the topic of the project ‘The Giving Plant’. Last time I talked about how we finished the cans, but we also later added small water bottles on the bottom and flipped the fiberoptics and the LEDs by 180 degrees so that fiberoptics doesn’t block the light. At the same time we also started manufacturing the leaves as shown below
The leaves were made out out of two pieces of tyvek and a parallon-like material, plastic for the main stem and metal rods for the branches. This was a long winded and complicated process. Eventually these leaves broke almost 5 times in different places and had to be reinforced with wood and plastic. They were also very hard to work with in terms of access to the fiberoptics LEDs, which burned out several times and had to be replaced. Due to repetitive nature of testing, the wires came off multiple times and had to be re-soldered and replaced.
At the same time I started putting together the electronics – soldering all the components on two boards. First board was for fiberoptics LEDs (each had to be controlled with a separate MOSFET).
The second board was for pump switching, LED strips and LDR readings. As the strips were added later, they caused a lot of trouble in terms of component placement and code. Apparently, the library we used in the beginning – Fast LEDs, was messing with millis() function which I used throughout the code to control variables. Moreover, the libarary was randomly rewriting my variables, which was very hard to debug and took me 2 days to figure out that it wasn’t my fault. I switched to NeoPixel and everything worked like a charm.
The LEDs shown above are the ones we used and they required 5V. Initially, I tried to power them from Arduino, however most of the times that wouldn’t be sufficient to run them at full brightness, so I had to switch for external supply. I originally used LM7805 voltage regulator to drop the 12V supply down to 5V. I thought it worked fine, but I miscalculated one thing. LM7805 drops down voltage by dissipating excess heat. Over half a minute it would get so hot, that the LEDs would not be able to draw enough current and go red, and I burned myself twice working with them. Even after replacing 1 LM7805 with 3, they would still heat up so much that they would melt the wires around them. So instead of figuring out the problems with overheating, we decided to buy a 5V power supply and power the LED strips from the supply. That worked fine.
After I finished all the code, we tested everything and it worked fine, however was still partially disassembled. There was a lot of fine tuning and fixing small mistakes which consumed more time than it should have, including re-painting jobs, fixing broken pieces and putting wires through the poles. The poles were also drilled in random places to let the light escape to create irregularities.
A few days later we found a place to work with SMAs – Urban Glass workshop. We also read multiple tutorials for how to manufacture the SMAs, including quick annealing to make them softer and then full transformation in a 950F kiln.
We used the torch below to anneal the SMAs however is was very tricky, as it is very easy to overheat and burn the SMAs. Once you overheat Nitinol, it becomes very firm, as it changes the composition (I assume Titanium bonds with oxygen) and there is no way to fix it. So it took us 5-6 burned pieces before we were capable of annealing the wires. Once annealed, they become much more malleable and soft, which helps to hold them in a particular shape.
Once enough wires were properly annealed, we trained them using the kiln. We drilled holes in a metal slab in a desired shape, that would hold the wire in place for 20 minutes inside the kiln. After that we quenched the wires in cold water:
And then voila, all the wires worked just as expected:
When the wires worked and everything else was close to be done we were super excited, however we had very little time left. It was also the time when a lot of fabricated parts started breaking. The most issues were cause with constant redesigning of the pot to fit all the electronics and be modular and easy to remove, however once we put the pebbles on the top lid, everything went downhill
The cumulative weight of the pebbles crushed several wired and the main rope, which had to be replace. Moreover the pebbles had to be taken off and replaced with paper. Going back to the wires, we realized we had to do proper Physics to get the current and voltage we needed.
Using the properties of the wire and the spec sheet we’ve found that we need somewhere between 2A (roughly 45s activation) to 7A (roughly 2s activation). The problem, however was that wires we have bought had resistance that was almost negligible, around 0.2Ohm for a 7 inch wire. This was my major concern, as most power supplies will activate short circuit protection or simply drop the voltage too much to be supplying enough current through the wire. We, however did all our tests with 9V batteries and assumed we would work it out for a power supply later. Well, first off we started with trying to connect the wires. The problem was that Nitinol is non-solderable, and had to be connected with copper tape, which quickly heat up and melted the plastic around it. We later found a solution using semi-circular holder pins that can be crushed, making the two wires electrically connected. The holders would not heat up, however they took more space than copper wire. At the same time, we tried to sew the wires and see how they work. Below you can see one of the videos which both shows the capabilities and disadvantages of using Nitinol.
The wire unfolds the petal very nicely, however it also rotates unless glued (most people glue it with copper tape) and sewing doesn’t hold it well. You can also see that the tests are performed on a 9V battery which wasn’t safe and we ended up scraping this idea. Thus, each of us tried to focus on one task: trying to fix the wires in place, making longer wires and training them, creating the electronic circuits. Fast forward, all of those failed, extra wires we wanted were not delivered on time, all the power supplies we used had relatively significant internal resistance or simply treated the circuit as a short, and more secure sewing of the existing set of wires didn’t produce the desired results – they still moved freely. Having said that, I realized that we should have used thinner wires (we used 1mm dia.) that would have more resistance and would require almost 7 times less current for same activation time.
Having 1 day before the presentation we had to change the plan. The Nitinol was a great opportunity to learn and potentially stand out and we came very close to making it work, however unfortunately it required a little more time and efforts. For this presentation, we changed the flower to be static with fiberoptics in petals that would light up.
The LEDs that were connected to the petals were 5V leds this time, first of all to use a different color LEDs and also to be able to connected them straight to the arduino (220 ohm resistors were soldered straight on the wires).
The whole setup worked nicely, however was falling apart because of too many times it has been pushed in and out of the pot, and also some wires got disconnected during presentation. For that reason, we decided to move the board on the side of the lid instead of keeping it on the bottom, and also move the mist maker outside to prevent potential leaking during the show and be able to easily refill the water. The mist maker itself didn’t quite work as ideally as during the test, as the pipe most probably was bent and thwarted the mist going up the tube.
The last setup worked better, however it was still pretty hard to manage the wiring and the code. Tomorrow, we are presenting the final design, which unfortunately doesn’t have Nitinol working. I believe it there was better communication in the team and the final design was properly thought of, the project might have accomplished all of its goals, however, that said, we learned a lot of important lessons and in the ended managed to make something worth presenting. I will update this project one we record visitors testing the plant. Thank you for your time reading this!