It’s official, I have brought a child into the world. Multiple children, actually. Their names are PMD Enclosures 1 through 4. They are the first items I have designed (other than a 3D printed resin toy car) that have been taken from conception to finished product. It’s a really cool feeling to see hours of CAD work turned into a tangible piece, but it’s also been a rather brutal educational experience.
When I first started as a research assistant with the SES project, my first job was to help paint a room that would become one of the staging areas for the project. My second job was to design a camera enclosure that would be splash-proof and impervious to humidity. The only reasonable solution I saw was to encase the entire camera in a housing. The old “design” had been to simply put the camera in a turtle-shell like enclosure, that left the less-than-waterproof lens exposed to the elements. Gaps were filled with caulk and silicone.
I went over the basics of my design in a previous post, so I’ll just cut to the chase. My design was submitted for fabrication about 2 weeks before the end of the Winter semester, and was completed on Saturday (May 26th). It was a little behind schedule (a lot behind schedule…) but in the end it worked out. Unfortunately, because the job was finished late, there was no time to make any modifications based on the first prototype, so all tweaks had to be made in-house or returned to the company and rushed back out.
And I will admit, there were some mistakes on my part. I mis-measured the spacing on a bolt hole pattern, I incorrectly specified the through-hole diameters for some metric fasteners, and I was ambiguous in specifying feature on a drawing view that led to a hole being drilled on the wrong side of a bar. It happens. But you learn from it. And you become a better engineer. I also underestimated what was an appropriate amount of “wiggle room” for some of the parts. I had a couple of machined slots and bolt holes through multiple layers that were drilled for quarter-inch hardware at 0.266″, but would have been better at 0.277 (loose fit). As much as one would think that parts from a CNC waterjet or mill would be perfect within a few thousandths of an inch, the truth is that no material is perfect, and nothing with human assembly will be perfect, due to inconsistent techniques or fasteners. Until we can 3D print everything, non-trivial tolerances will be something we have to deal with.
The physical realization of a CAD model will always be imperfect in ways you can never predict. Of course, you also can’t predict when the machine shop will screw up your part. In the case of my boxes, the first prototype was welded together poorly, resulting in thermal expansion/contraction deformation of the support arms holding the camera plate (see above picture: note the angle between the main rectangular plate and the arms on either side, about 5 degrees). The resultant workpiece was essentially impossible to assemble without Hulk-like intervention, played by a C-clamp. When we received the piece, we immediately called up the fab shop and demanded the remaining pieces be properly fixed in a jig, squared, and tacked before being fully welded. I should have made a note in my drawings to ensure that the brackets were parallel, but it was an intention that should have been readily understandable from even the most cursory of inspections of the drawing package. Had they even conducted a fit check, they would have realized their failures. In the next few days, we learned that it had been a rush job essentially, and the piece was shipped out before the welder had time to learn from his mistakes. Of course, if they had started our work order early and delivered on time, this would not have been such a big deal.
One of the cool/bad things I learned about parts fabricated through welding is that the intense application of heat to a workpiece causes chemical changes to the surface, resulting in discoloration. The bands of color, from gold to purple, are strongly dependent on temperature and a specific pattern or colors is a sign of good welding (which was achieved on my pieces, thankfully). The discoloration is also a weak point, however. On stainless steel, the heat-treated sections of metal have different corrosion properties than the original metal. Exposed to water they will rust more readily, so therefore the coloration must be removed with scotch-brite and elbow grease (note the before/after on the weld seams in the above picture).
This project also gave me some first-hand experience integrating waterproofing features like gaskets and O-rings. Although I had designed the enclosure to be as easy as possible to seal, one could not simply slap a plate on a flange and call it watertight. On the main flanges a 1/16″ gasket was used. To seal the glass viewports embedded in a CNC milled piece of delrin, an O-ring was used. Designing for an O-ring is somewhat interesting. In compression-type applications, on O-ring is typically set in a groove that doesn’t allow the rubbery ring to sit flush with the face crushing it. It sticks out about 20% of the diameter of the O-ring rubber. This allows the necessary pressures to be generated for sealing the face.
I would say the experience of taking a waterproof camera enclosure design from start to finish has given me a great appreciation for mechanical design, and also taught me a lot of little things that you don’t learn in a classroom. In the coming weeks, I’m going to watch my creations get abused during testing at the LCC. This will be in addition to setting up a ton of instrumentation as the project gets underway.
Until next time…