Anders Häggman

MAS.863 Portfolio

WEEK 14 // Final Project

Initially I had grand plans for what I would create: a rotating camera base, with a motorized dolly track. It soon became clear that this would be too much to tackle in my situation, but luckily my project proposal could easily be split into smaller sections. I chose to work on the rotating camera base, as I thought that would be more useful of the two, and better suited for the final project.

In it's current form, the camera base rotates very slowly, then stops to allow a picture to be taken, then rotates very slowly. The timing of the camera shutter is currently controlled with an intervalometer, but it should be pretty straightforward to have the microcontroller also control the shutter of the camera (and example of which can be found on Geoff Tsai 's page).

  • Level of previous experience:              
  • Time taken:                +
  • Tools used:
    CAD  //  3D printer  //  Laser cutter  //  Soldering equipment & Desktop mill
  • Got help from:
    Charles Guan  //  Will Langford  //  James Coleman  //  & many others
  • Main learnings:
      Make an effort to plan your schedule to leave enough time for different stages of the project.
      Get enough sleep. When you are tired you'll make more mistakes.
      Prototype and test as much as possible. Build sections at a time, and test, instead of building everything and then debugging as a whole.
[Image_Missing]
These are some of the original sketches I had regarding the final project. Initially I wanted to make a combination of self-propelled dolley track and turning camera mount, but it soon became clear that it would be unwise to attempt both at the same time. Therefore I decided to focus on the turning tripod head, and leave the motorized track for later.
[Image_Missing]
Starting to figure out exactly how I would make this in practice.
[Image_Missing]
From past experience, I know how easy it is to get blind to size and dimensions when working in CAD. Ideally I would have liked to make the final project in a more iterative fashion, but due to external factors, this became a down-to-the-wire project, and there was less time for iteration. However, as often as possibly, I would try and ground the design to the real world, to get a sense for if things were generally going in the right direction. This is a quick 'insanity' mock-up, just to see if the size I am proposing on screen makes sense in reality. The current CAD design is roughly the height of the tea container, and a little wider. Seems fine to me. Check passed.
[Image_Missing]
Similarly, for gaps and clearances, I would always try and measure something close by to get a sense for how big the features are in my CAD model.
[Image_Missing]
Dead-lines were approaching, and I had to send out some of the parts for printing even though the whole model was not complete. This posed a challenge, since I had to decide dimensions for some parts, and freeze that design, and then just make the rest of it work. I did not make life easy for myself here, and intentionally left very little empty space. Leaving a bit more leeway could have saved me a few gray hairs later on, but I was not treating this as a prototype, but rather a lottery ticket to a functional product or a failed attempt. In other words, if it would work, I wanted it to be 'good enough' that I would be able to use it for several years, instead of making something simply functional that demonstrates the working principle, but is not necessarily packaged very well, or durable. So, a down-to-the-wire schedule, and high risk approach. I would not have the most relaxing week ahead of me. On a side note, here are some of the first sections being printed. It is becoming real. The race is on.
[Image_Missing]
However, as with the quick physical mock-ups (such as the tea container), I also tried to prototype and test as much as I could along the way. On the right, a test piece that I created to check the dimensional accuracy and tolerances of the printer I was using. Unfortunately it did not arrive in time, and I had to freeze the final design for the bearing surfaces without any real knowledge about what the dimensions would be. I based my guess on discussions with several people who have used the machine before, and on my own experience. It was also helpful that the printer I was using was quite high quality, so I knew the dimensions wouldn't bee too far off.
[Image_Missing]
Another view of the test piece.
[Image_Missing]
A bit more ABS plastic accumulated at the starting point of each layer. You can see here the darker area where the metal shim that is used as the bearing surface scraped against the model.
[Image_Missing]
In critical areas (surfaces in contact moving in relation to each other), I used a knife to cut away the largest inconsistencies, and then used sandpaper to smooth the surface.
[Image_Missing]
As mentioned, this was to be a sprint to the finish, and in true form, my parts would arrive in two shipments, one four days before the final presentation, and the other shipment just the previous day. It was a tense moment when the first parts arrived, and started putting things together. Luckily, the parts fit extremely well.
[Image_Missing]
Meanwhile, the rest of my 3D printed parts were in a lye bath to remove the support material. I had checked on them a couple of times, but didn't get suspicious until they had been there for over 24hrs. I e-mailed the staff, and it turns out that the lye solution had not been changed in a while, which meant that it was not very efficient at removing the support anymore. The following morning, course staff would change the solution, and most of the support would be gone. I don't have much experience with the lye bath solution, it appears that it should be 'white-ish' in color. The solution we had was approaching brown. (This is a photo of the 'fresh' solution.)
[Image_Missing]
Even with the new solution, the support for some of the smaller parts which were in a mesh cage had not completely dissolved. I was strapped for time, so I called it 'good enough' and fished out the parts. Even though there was very little support structure left, it was incredibly hard to remove it. In retrospect, I would perhaps have been better off just leaving the parts in the bath, and spent some time on an extra trip back to the IDC to collect them later. However, the problem would be solved by Mr. Dremel, who would save me several times as the night wore on.
[Image_Missing]
Mr. Dremel in action.
[Image_Missing]
One of the many mistakes I committed due to time-pressure and lack of sleep (and just being human): not making sure I had the stock I was designing for. I had designed the section 'floor' to be made out of 6mm thick laser cut acrylic. However, once I had designed the parts and sent out the 3D printed ones to be printed, I realized that I did in fact not have 6mm thick acrylic. I had 5.25mm thick. Luckily, bristol paper was available, which was around 0.38mm thick, so I cut two shims to be placed under the 'floor' to lift it up, and prevent any problems with the remaining assembly.
[Image_Missing]
Almost all of the parts of my assembly. I later realized that I had forgot to add two laser cut pieces that hold the 3/8" camera / tripod attachment screws in place, but it was too late at night to for me to care enough to take a new picture.
[Image_Missing]
I 'dremeled' (is that a word?) flats into the shaft, so that the gears and connectors would better be able to fasten to it. I'm not sure if it's necessary, but I taped off sections that did not need a flat in them, and only made them where they would be needed.
[Image_Missing]
Amazingly good orange, toasted walnut and cranberry bread that my labmate Susan brought. This kept me going through the night.
[Image_Missing]
Getting a feel for the friction in the assembly. This was one of the hardest things to design for, and I didn't know of a good way to get a solid estimate for what it would be. Luckily, with the needle pin roller bearings that I had, and some ABS/metal sliding surfaces the friction was not too bad, but there was a bit of room for improvement, so I opened up the assembly again, added some oil to the bearings, and closed it again. Better. The friction is quite small, but large enough that the top won't keep spinning if you spin it and let go.
[Image_Missing]
Back to the mistakes part. In my rush to send out the 3D parts for printing, I had totally forgot to take care of a feature that I had filed under 'take care of this later' in my brain. As I was walking home from IDC one night it struck me... I had completely forgotten to add the holes for power and USB connection. I would need to drill. Due to the curvature of the part, it was a bit challenging to fasten. I used one clamp to tighten it down from the tripod connection point, and two other clamps were tightened on either side to prevent it from rotating. Also, at the suggestion of James, I used an end mill instead of a drill (hoping that the end mill would not try and pull the layers of the 3D print apart as much). It actually worked incredibly well.
[Image_Missing]
Another mistake and learning opportunity, always check what you are doing. When making the first round of laser cut parts, I had forgotten to add holes to some of them. When I re-cut the pieces, the acrylic I was using didn't have the backing paper on one side, and it created these blotches on the acrylic surface. Since these parts wouldn't be seen in normal operation, I was not too concerned. Also, after having tested making the structure with the first set of parts, I realized just how angled the cut surfaces were, and therefore sanded down the new set a little bit before assembly.
[Image_Missing]
A laser cut ø shape. The piece is turned 90° clockwise to lock in place the female 3/8" thread that attaches to the tripod mount.
[Image_Missing]
The most frustrating last-minute 'oh-no' sensation of the whole build. In my CAD model, I had used the effective pitch diameter when modeling the worm and gear, so that I would get the placement of the axels correct. However, the gear extends beyond the pitch diameter, which meant that the structure I had created for the gears and motor were too tight, and nothing fit. Dremel to the rescue!
[Image_Missing]
This is what sleep deprivation + frustration + dremel looks like. It's not pretty, but it works.
[Image_Missing]
Although the 3D printed parts seemed dry, every now and then when I'd be handling them and turning them around water would seep out from somewhere. I am assuming the structure is porous enough that water has gotten into the honeycomb structure, and seeps out when the part is in a certain position for long enough.
[Image_Missing]
This is what the final prototype looks like on a tripod, with the camera base attach to the top. Apart from the connection flanges, and the protruding motor at the back, I think the size and shape are not bad. This version uses the same board as what I made in week 10 for the output devices. Future improvements will include a jack to connect to the camera, LED's to indicate current mode, and up-dated code.
[Image_Missing]
...and another picture with the camera mounted to it. Unfortunately I finished the project so late, that I did not have time to include a rotating time lapse video.