Rev 0.01

Before the critters showed up I was already playing with some of this stuff, so I had a camera housing in white PLA created pretty quickly.

Although there were choices from the bins of filament I have, I started with the white PLA. It looks the most like a ‘security camera’, and I have had items printed in black PLA soften and warp when exposed to bright sunlight.

By now I had already made some design decisions.

One goal is to use 12v power. Extension cords and power-strips really won’t do outdoors very well. My plan is to use a low-voltage lighting transformer to supply 12v to multiple RPi cameras.

You can see the 12v to 5v converter, and the 12v receptacle on the cowl and wired to the converter.

Some other design points to note:

The cam-wall sits in a groove on the base and the base has to slide in to the cowl, so I created a test jig to make sure things would fit together.

You can also see the tabs at the top of the cowl that act as a backstop to the cam-wall and prevent it from being pushed back past 90 degrees. A last thing to note is the built-up stand-offs for both the RPi and the camera.

Making it was not simple. There is going to be a learning curve.

There are a lot of variables when 3d printing, and I was still learning to use the Shapr3D software on my iPad. Prints would fail, and tolerances would be off, rendering prints useless. Below is everything it took to make the first camera.

A couple of these were failed prints. Sometimes the filament will bind on the real because it is wound badly, and the printer will keep trying to go, but without a supply of plastic.  Other times I print with filament that is 1.75mm diameter, when my printer is made to use 2.85mm.  Some of the time using the smaller filament is not a problem, but it can bunch up in front of the hot-end and stop extruding, which is the same problem caused differently.

The other prints were design issues. The first one was too small for the power supply, the next two had bad tolerances; the first would not fit in to the cowl, and after making modifications narrowing it, the battery box needed adjustment, and the part of the base that slides in to the cowl needed to be thinned in the final version that I used.

One thing I did not want to have to do was print out the cowl again. That is by far the largest print, taking the most time and using the most filament. One design aspect of this housing was to try and keep it small; as narrow and short as possible. With the element of the cowl back-stopping the camera-wall I did not realize that the Raspberry Pi, once mounted, would not pass in to the cowl.

To make the Raspberry Pi slide in to the cowl I had to file down the wedges that back up the camera wall.

Finally, once I had a base that would fit in to the cowl, and a cowl that let it slide together with the Pi attached, I learned that the camera mounting piece was just slightly too tall to fit. As it turns out I did not like the camera cable having to flip around So I flipped the camera mounting standoffs around. This made it so the camera cable would come out the top of the camera instead of having to wrap around and come up the back.

As it turns out I don’t like this design much. The component choice for the buck-converter is more than the design needs. It is too big, and it was too expensive. In this design it is held down with double-sided tape in a box that was built up during printing. You can also see that the cowl is tethered to the buck-converter, where it is attached by screw terminals. This, it turns out, makes it cumbersome to work with.

It doesn’t look bad deployed, though.

Sitting there. On the table. Or on a different part of the table.

It turns out that having a good way to mount the camera is important. Being limited to tables and shelves or sitting on top of other stuff makes it a challenging design to work with. Additionally, something you cannot see well in the pictures, is that it is just kind-of squeezed together and only friction holds the cam-wall in place. When working with the model it will often flop apart in to pieces when assembling/disassembling it.

And, it is too small to accommodate the Raspberry Pi HQ lens or a v1/v2 lens with attached illuminators. Working toward one design that can accommodate different cameras begins the driving force behind the next iteration’s design.

It is not a bad start, though, and a number of the design elements will carry forward all the way to the first run producing a batch of cameras.