The motivation behind this project was a personal peeve of mine- namely, that the temperature in my house can vary by up to 15 degrees, depending on the floor and room. So, my idea was a temperature control mechanism that had multiple sensors, and allowed you to see in real time what temperature you were setting your different rooms to.
The storyboard for my temperature control system
For this project, I needed a switch (to toggle between reading mode and setting mode), a potentiometer (to simulate the setting of temperature), an LCD for output, and a few temperature sensors.
I built this project in stages, adding one element at a time, and writing a test program for it, before building the final circuit.
My first stage used the switch and the LCD. The LCD simply displayed the orientation of the switch (on or off)
The first stage
From here, I added the potentiometer, and output its value onto the LCD. My project won’t actually control the temperature (obviously), but the potentiometer and LCD will serve to simulate the input and output of a temperature
Stage two- now with 100% more potentiometer!
Stage three was to add the temperature sensors. Initially, I used the digital temperature sensors from the kits (reading the analog out pin). However, I was unable to find a datasheet for the sensor, so I didn’t know how to convert the output of the sensor into Celsius or Fahrenheit, so I swapped to the LM35 3 pin temperature sensor, for which I could find the proper conversions. I wired one up, and output its value to the LCD
Stage 3- a single LM35 added (on the green breadboard)
With all the proofs of concept finished, it was time to build the final program. When the switch is flipped on, the program freezes the offset between the two temperature sensors, and then outputs a temperature based on the potentiometer’s value, plus the offset, in the range of 40-100 degrees Fahrenheit. So, if the potentiometer was in the center, and when the sensor was flipped sensor one read 60 degrees, and the second read 65 (so an offset of 5), the screen would say “t1- 70”, and “t2- 75” (the median value, 70, plus the offset of 5). When the switch was flipped back off, the screen would go back to reporting the temperature on each sensor.
It was here that the most difficult problem reared its head. Up to this point, it had been smooth sailing. However, the readings I got from the LM35s were extremely unstable, fluctuating by 20-40 degrees in the space of seconds. My first instinct was to take multiple samples, and average them out. I ended up using the mean of 64 samples. While this helped the stability, I still had far greater fluctuations than other projects online lead me to believe. The sensor should have been accurate to 0.5 degrees Celsius.
After further Googling, I placed a 1K Ohm resistor in between the data and ground lines of the LM35. This was a complete failure, and zeroed out my readings. I also moved the sensors to their own ground, away from the switch and potentiometer. This also failed to fix my errors
Finally, I carefully read through the datasheet of the LM35, and found that for stability, they suggested adding a 1 micro-farad capacitor and a 75 ohm resistor between the data and ground line. I didn’t managed to find exactly that, but I got 66 Ohms from 3 22 Ohm resistors, and found some 2.2 and .22 micro-farad capacitors. Between these, I had the most success with the .22 micro-farad capacitor, which finally stabilized my readings to something that was within the realm of possibility, though I couldn’t find a thermometer to double-check my readings.
The final circuit
In the video below, the system starts out in set mode, with a -3 degree offset. You can see the values change based on the potentiometer, always maintaining that offset. The, the switch is flipped (I have to set the camera down to do so), and the screen switches to a display of the temperature from the two sensors.
Were I to do this project again, I definitely would have used a larger breadboard, rather than two smaller ones. The small green board was especially cumbersome, as it did not have a power and ground line. Furthermore, I would have got straight to the LM35 datasheet, as that provided a wealth of information about recommended use of the sensor. I would also have brought a thermometer, because I was unable to test my final readings for accuracy.
Overall, I’m satisfied with the outcome. This was more of an exploratory project than anything else, and it was fun to work with Arduino again. It was quite interesting to contrast Arduino with the sort of microprocessor programming I’m used to, from my Embedded Systems class. Arduino was far easier to use, and required far less documentation to get basic things up and running, though it has more limited hardware than the DISPIC microcontroller we use in the class.
The idea for this project came when I realized one of my roommate’s birthdays was right around the corner. I wanted to make a personalized dice bag for her, since she inexplicably bought 6 sets of dice all at once. So, I asked one of my more artistically inclined friends to make me a sketch of one of the characters from our shared DnD game. My only real stipulation was that it had to have a place for the lights to fit in naturally. My friend delivered in spades, and gave the sketch you see below.
At this point, I discovered that the gray settings on the scanner were insufficient for my needs. There was too much noise in the image for me to properly colorized and refine it to something I could use. So, I traced over the image with a pen, erased all of the pencil, and scanned in black and white only, to get the clearest outline I could.
The inked image, scanned
To get the basic coloration down, I pulled the scan into GIMP, and used the fill tool to fill in areas with color.
This didn’t go so well. An side-effect of being a scan of a drawing was that some of the lines, though they look complete, had very small holes in them. So, when I filled, the fill to “escaped” through those holes, and filled the outside parts of the image. So, I went over every line in the image with the pencil tool, and fixed all of the small holes I could find, so my outline was a true outline.
Once that was finished, I experimented with a few colors, before finally settling on a gray cat, blue overalls, and yellow buttons/fireflies. I made one of the fireflies blue as a nod to Navi, from the Legend of Zelda. I also moved the fireflies closer, in order to lower the overall size of my embroidered design
The colorized image. The eyes will eventually be colored blue
Now began my struggle against PEDesign. PEDesign does not like having small, differently colored features, like the irises, or the claws. I spent quite a lot of time slowly enlarging these features, until they finally were stitched. I picked my 9 thread colors, and went to embroider.
My first embroidery attempt, mid embroider
The picture caption is slightly misleading. I put in the wrong color on my first try (you can see it off to the right), and scrapped it. However, the second attempt went far better.
Right up until the end
my second attempt “complete”
The machine had a bobbin tension issue, which caused the bobbin thread to show around the edges. On the overalls and shirt, this wasn’t too much of an issue. But when the machine went to do the black outlines, It looked really terrible. Furthermore, many of the outlines were larger than the features they were outlining.
The end result was that I decided I didn’t need about 85% of the outlines in the image. On fabric, the lack of outlines was far less odd that it was on the image. So, I removed almost all of the outlines from the original image, in order to enlarge the features, and avoid the bobbin problem (should it arise in the future).
My image, with only the most necessary outlines remaining
I had to increase the thickness of most of the remaining lines in order for them to be picked up by PEDesign, in addition to increasing the size of the claws. Since I was sewing on black felt (which was apparently recommended to me in error, as it’s not the greatest for embroidery), I decided not to stitch the outlines, instead allowing the black fabric to show through. This ended up being a very good idea, as it allowed me to enlarge small features, like the whites of the eyes, and the embroidery
Embroidery attempt 3- success!
Aside from the blue fairy, whose thread nearly pulled a hole in the fabric, and forced me to manually repair its shape, my third embroidery attempt turned out quite well.
For the soft circuitry, I spent quite a bit of time considering how I wanted to sew the circuitry on. I definitely didn’t want the gray conductive thread showing through. I considered sewing on a separate layer, only sewing halfway through the felt, and securing the thread with only hot glue. In the end, I decided to sew the lights onto a separate piece of felt, and sew that to my front panel in black thread. This allowed me to securely attach all of the electronics to fabric, while keeping the stitching hidden.
The circuitry patch attached
A peek at my circuitry
The patch wasn’t the prettiest, but it was functional, and wouldn’t be seen in the final project. The switch is facing out, so it can be flipped through the front panel (it can be felt distinctly beneath the fabric). One major issue with this design, I realized later, is that replacing the battery is now effectively impossible. However, with only 3 LEDs to drive, I’d wager the current battery will be going for quite a while.
With the front panel complete, I moved on to creating the pouch. Since the highlight of the pouch is the embroidery, I went with a dark blue lining and a black zipper. I made a relatively small bag, as I intended for it to be a dice bag for my friend. Mercifully, the creation of the bag went off largely without a hitch. The zipper had some sections on either end that were pas the stoppers of the zipper, and those holes had to be sewn closed, but aside from that. the bag ended up looking not too shabby.
the bag, almost finished
The final bag
Were I to do this project again, I would definitely have chosen fewer colors for the embroidery. Switching out 8 different thread colors was nerve racking, and made the embroidery take over an hour, so my failures cost me quite a bit of time. I would also have just forgone the claws entirely, as they are not very visible at all on the final bag, and cut down on the amount of thread the toes have. Better to remove the small features entirely, than enlarge them to the point where they show up, because even when they show up as small in the PEDesign render, they may be enlarged on the final fabric.
Those minor quibbles aside, I am quite proud of the embroidery. I may not have done the original sketch, but it took quite a lot of work (6+ hours) to transform the sketch into an image that could be satisfactorily embroidered, most of that time being zoomed far in, and using a 3-5 pixel wide brush to change lines and feature width. The embroidery is by far my favorite part, as it is what really made this gift stand out, rather than just being a amateur, though competent, bag with some lights. I believe the work paid off, and am very happy with how the embroider (and the entire bag) turned out.
And yes, my roommate did like it. Quite a lot.
To start this project, I brainstormed several possibilities, for each of the prompts. Among the ideas I considered are-
A pair of chopsticks with one end normal, and one end like a small fork
My head, grinning, pasted onto Michelangelo’s David
A forearm mounted phone/phone assistant that fits over the thumb, with controls on the portion of the palm, directly below the thumb
The idea I decided to go with was a statue of my hands, which would (ideally) double as a dice tower. From this base, I considered several designs, including a modern, angular support, and a contoured, sweeping base, though neither of these got to the sketching phase. My final idea was to go for a stone-hewed look, where it looked as if my hands were set in stone, as an unfinished stone sculpture. This had the added benefit of requiring the use of all three types of tools- Natural modelling tools (meshmixer and a little bit of blender), geometric modelling tools (blender), and 3D scanning (a Kinect + reconstructme).
a basic sketch of the sculpture
Since this was an art piece, it was less interested in solving problems or target audiences than it was in looking good. A storyboard would have just been somebody walking by, and looking at it (and possibly rolling some dice down it). My major concern, then, was that the statue look good. Second, was that dice roll down it, for which I did a few experiments with my own hands. The end goal for the “user”, so to speak, was to have a good looking statue, that had a secondary, not immediately apparent, function.
My first major challenge I faced was how to scan my hands. This posed a very tricky issue, as I had to find a way to rotate my hand 360 degrees, without it moving too much, and without getting any of my torso or head in the frame. In addition, the software was designed primarily for capturing human torsos, not hands. In the end, I placed the Kinect up high, and then rotated on one foot (so that I rotated without moving my shoulder), with my arm up as straight as I could. This was a good lesson in how finicky 3D scanning can be, and the necessity for a smooth 360 rotation when capturing models.
The scan of my right hand
In order to turn my two rough 3d scans into something I could use, I turned to meshmixer, a 3d design software with a lot of sculptris-like sculpting tools, but also 3d printing/scanning tools such as support generation, and, most importantly, “make solid”. This tool does its best to fill in any missing polygons in a model to make it solid.
It was here that I found that, as opposed to smoothing the model to remove all of the small pits and bumps leftover from the scanner, simply reducing the geometry worked very well, and gave a consistent, smooth finish.
The greatest challenge I faced in this project was separating the fingers fused by the 3d scanner. Due to the fact that the fingers were closer to each other, than the front of the fingers were to the back, the “make solid” tool would fuse the fingers, rather than construct them. I tried every combination of pinching or reducing brush I could, before finally hitting on a solution. I deleted the polygons fusing the two fingers, then separated one finger from the rest of the hand, then separately used the “make solid” tool on the finger and hand separately, before rejoining the two parts and smoothing out the joint. After that, and some inflating, flattening, smoothing, and creasing, I arrived at a satisfactory right hand model.
The 3d scan of my right hand, after cleaning it up
Upon moving to the left hand, I realized that every single one of my left hand scans had serious issues with it. See the images below for some of the issue. I realized that I would either have to find a way to fix one of the scans, or go back and do more scans, something I was thoroughly uninterested in doing. This was extremely discouraging, at least until I came up with a third option that trivialized the problem. I realized that my right and left hand scans were in almost the same position, with very similar finger orientation. I could mirror my right hand model, make a few small adjustments, and call it the left hand.
An example of the problems with my left hand scans. Notice how the thumb and index finger are mangled
My next challenge was generating a convincing rock formation to set my refined hand models in. For this, I turned to blender, as I wanted the rock formation to start out low-poly, and blender allowed control over individual faces, lines, and vertices.
My first attempt at generating stone was to start with a cube, and extrude, subdivide and pull faces until it looked satisfactory. In this way, the rock would slowly “grow” around the hand model, which I had suspended in air. This did not work so well. Subdividing faces made copies of lines layered on each other, which were a pain to move, and some faces would refuse to subdivide.
After doing some research, I found that blender, too, had some sculpting tools. Most crucially, these tools had a “constant detail” setting, meaning that using them would not create any new polygons. So, I made a cube that completely covered the hand model, subdivided each of its faces 4-5 times, and used the sculpt tool to get a very rough model of the stone.
The sculpting tool helped get the basic shape, but it was far from done. Firstly, the base was still mostly square, which was not the look I was after. Second, the sculpting tool left quite a few “spikes”, single vertices that were not reduced with their neighbors. So, I refined the shape by moving individual vertices, flattening spikes and creating a more natural shape in the rock. The limited polygons helped to create features like ridges and bumps easily, as I only had to manipulate 6-10 vertices to generate any one feature.
The back of the right hand, set in stone
The front of the right hand, set in stone
An example of the spikes left after sculpting
From blender, I took the models back into meshmixer, in order to smooth out the sharp edges of the low-poly rock and prepare the models for printing.
Both hands, ready to print
After printing the hands, sanding them down, and gluing them together, I found that I had failed to take a few things into account when designing them. Firstly, my models were smaller than my hands, so they did not trap dice as well. Secondly, plastic dice bounce a lot higher against plastic than they do against flesh. This meant that my desired method of rolling dice, down the left hand into the right, would not work. The dice simply bounced out. Were I to make this again, I would definitely make the hands larger, and I would also use a softer plastic. Jeff showed me a nonfunctional 3D printer that printed a plastic much more spongy than standard ABS. This would hopefully minimize bounce in the finished model. In terms of aesthetic, my original plan had been to print one hand in black, and one in white, but the black print jammed, and didn’t have enough filament to start over. Were I to do it again, I would like to have contrast between the hands.
Overall, however, I am quite pleased with
The finished models
The top of one model, pre-sanding
and post sanding
From the start, I wanted to make a pop-up card (style thing) with a scene from either movies or pop culture. I considered several options, before finally settling on the climax of Harry Potter and the Dealthly Hallows, where Harry faces Voldemort for the final time. This lent0 itself quite well to the sort of project I wanted to make, with a red LED for Harry’s wand, a green LED for Voldemort’s, and a yellow LED for the sun peaking through the window (that didn’t make it into the final version)
The Finished Project. Ignore the name tag/fallen fighter at the bottom
Rather than try and print out designs onto paper, then cut them out, I decided to go for a more stylized approach, with the characters being blank silhouettes, black for Voldemort, white for Harry, and Gray (later amended to blue) for the bystanders. Not the most subtle coloring, but one that I felt I could execute on, and make well.
I had three major goals with this project (in addition to looking good)-
Make the electronics as reliable as possible
Minimize the visible wiring when the card is open
Make the card sturdy enough to survive transport, and open well
I had had quite a bit of trouble getting my first couple of copper tape circuits to work, so ensuring the reliability of the connection was foremost among my concerns. I also did not want a lot of exposed wiring in my finished product, as I felt that detracted from the overall aesthetic.
To get the silhouettes, I found images on the internet, converted them to vectors, and cut them out with the silhouette, in a quite similar process to the sticker cutting. My major mistake here was leaving the blade at the vinyl cutting distance, rather than lowering it. This unfortunately left an outline on the cutting mat when i was done, where the blade bit into the adhesive.
Mooks all cut out, with harry in the corner
My original idea for increasing stability was to glue multiple sheets of paper together once cut out. However, James recommended that I use chip board as backing, and that worked so much better. It matched the background paper color I had chosen, as well, so it did not matter that the cutout wasn’t perfect (I had to do it with scissors). The chip board was necessary, as I planned on attaching the LEDs to the pop-out sections of my card, those sections needed to be able to support the weight.
Harry and Voldemort glued to chip board
The chip board served two of my main goals. It both made the card far sturdier, while also giving a way to easily hide wiring, because there was now a paper layer, with chipboard behind it. So, by routing in between the layers, I could hide it, except around the switch and battery, and near the LEDs.
I had faced two major difficulties in my first couple of tape circuits. The first, was that I had trouble sticking the LEDs to the tape. The second was that the sticky side of the tape was not very conductive, so joining separate pieces of tape often caused the circuit to open.
To solve my first problem, I decided to solder the LEDs to their copper tape. This was an especially good solution, as it not only ensured that the LEDs would have a strong connection, it also held the tape to the LED far more strongly than the adhesive could, which came in very handy when mounting the LEDs.
The LEDs, Soldered
To solve my second problem (how to attach separate pieces of tape without sacrificing conductivity), I came up with a method for making double-sided “band-aid” copper tape pieces. I cut out one long, and one short piece of copper tape. I then stuck the two together, sticky side to sticky side, and placed this piece on top of the joint, so that the short piece was in contact with both of the pieces being joined. The extra length of the long piece allows the band-aid to stick, while the shiny side of the short piece bridges the connection between the two pieces.
A copper tape “band aid”. The short piece is in the center, leaving both ends of the long piece sticky
Thanks to the soldering and the wire band-aids, wiring of the first two LEDs went smoothly. However, here is where I found that, even with two batteries, my green LED was far more dim than my red. In light of this, I decided not to add the third LED, which would have been behind one of the background windows, and acted as the sun. Adding a third LED, I feared, would further dim my already dim green LED.
Wiring finished. Just need to glue the LEDs to the pop outs
Here also arose the issue of the switch. I had to run a tight balancing act with regards to it. I used a copper band aid on top to facilitate conductivity, however the switch still had an issue where if the band aid was loose on top, the connection was spotty, while if it was tight, then the switch wouldn’t move in and out when I opened and closed the card. I experimented with several tab lengths, tab thicknesses, and band aid looseness/tightness. In the end, what I found worked best is to have a tight switch, and have the tab not enter all the way, instead being held against the copper band aid (see my final picture). This was where I failed in my goal to make the electronics reliable, because my final solution, while adequate, is somewhat unreliable. Sometimes, I will open the card, and the lights won’t turn on.
Aside from having to scrap the third LED, and having a somewhat unreliable switch, I am quite happy with how the project turned out. I had never worked with papercraft before, and found it rather fun. I am quite proud of how the scene overall turned out. Even without the LEDs, I think it looks good, and the chip board will help the pop outs stay rigid for far longer than regular paper would. Were I to do this project over again, I would definitely have used actual wire for most, if not all, of the electronics, as well as redone the switch. The copper tape was finicky in its connections, and often difficult to work with. As far as redesigning the switch, I think one possibility that would make it better would be to have the contact surfaces be vertical (so opening the card pulls them directly together), perhaps in addition to the horizontal surfaces I tried.
For my company logo sticker, I decided to do the Lego logo, since I’ve loved legos since I was little, and it had a clear 4 layer structure.
My other sticker is of a cat named Sakamoto. I found an image of him as a sticker in one of my messaging apps, and thought that it would make a cute sticker. It also fit the bill for a sticker, as the art style involved clear lines, and few colors. My primary tools were GIMP (for preparing the images), Inkscape (for transferring my images to vectors), and the silhouette cameo, for cutting the vectors out of vinyl.
I encountered only one major pitfall during this project – separating layers in Inkscape. Many times, the bitmap tracing tool would leave holes where white items (like the cat’s eyes) should be, rather than giving me a layer of white.
I tried two separate solutions for this problem. On my lego sticker, since I had concentric layers, I used the fill tool in GIMP to fill in the other colors (e.g. I filled in the black and white with yellow for the yellow layer, the white with black for the white layer, and the black and yellow with red for the white layer), leaving only a two tone image, with all of one layer filled in, which could then be traced easily
The Sakamoto sticker, however, was much more complicated. I couldn’t fill in the black outlines in the image, so I could not just fill in with one color. I used the “break apart” tool in inkscape to separate the eyes from their layer (this only works if the paths are disjoint), and edited other paths in order to separate other layers. The black outlines in the image gave me some trouble, as they were often included in the “black” layer, requiring me to heavily prune the black layer down.
If their image supports it, I would suggest others use the fill-in method, as it is simpler.
Overally, I am extremely happy with how my project turned out. If I were to go back to it in the future, I think I would try and get some colors that match a bit more closely (a darker gray, and a paler red). I would also use a credit card, or something like it, to smooth out bubbles in the sticker, as there were a few that made it into the final sticker. Aside from that, I believe that I successfully translated the image to a laptop sticker. I am probably the most proud of the layering job that I did on the Sakamoto sticker. I had only a general guide for where the eyes should go, and feel I got it very right. In addition, the head had to be placed very precisely in order to match up with the base layer, and I placed it nearly perfectly.
The finished stickers
My original Sakamoto image
Sakamoto, after I edited his picture
The layers of my Sakamoto sticker
The layers of my lego sticker.
From the beginning, I knew that I wanted to do something tech/sci-fi related. My first thought, since I had played it recently, was one of the keycards from the 2016 DOOM game. However, That shape ended up being more complicated than my limited skill with inkscape could handle, so I instead switched over to doing a keycard from the original Doom, and adding the new UAC (Evil Sci-Fi Corporation #476) logo from the 2016 game. I managed to find a mostly clean logo, but had to first whiten it in GIMP (think photoshop, but GNU) so that the bitmap tracer wouldn’t pick up the scratches and weathering, and then invert its color so that the logo itself was black, not the space surrounding it.
Since I was putting a name and logo into something that was not intended to have them, I had to fiddle around quite a bit before I was happy with where the name and logo went. You can see one of my earlier ideas, which I rightly scrapped, below. I went with the “First Initial Last Name” style because I felt it fit better with the idea of a big corporate name tag, and it fit better on the tag. In the end, I took most of the major features of the keycard, and found that they made a lovely little rectangle for my name to go in, and found space in the top right for the UAC logo.
I did my project out of mirrored red acrylic. In the lab, I was recommended to try rastering the mirrored layer off the bottom of the keycard, rather than rastering the acrylic on the top. This leaves the card with a very cool impression that the design is sunk into the card, because the parts that are rastered off are clear acrylic, while the rest is mirrored, while the entire surface of the card is smooth (and a fingerprint magnet, unfortunately). My one regret is that I also cut my piece from the bottom, and the heat from the laser slightly melted the edge of the mirrored backing, which is visible from the front (it can be seen along the bottom edge in the photo). Were I to do it again, I would cut from the top, and raster from the bottom. However, that is a minor gripe, and I am very happy with how my name tag turned out!
The final Design (Though I ended up increasing the name size)
An earlier design, with full name, a larger logo, and fewer features
The logo I cleaned up
the source material- Keycards from Doom
Cutting out the tag
The finished project