So we have these metal “lockers” we got from IKEA that we use as media storage. When we moved recently, the movers lost one of the little clips in one of the units and that meant we couldn’t use the shelf inside. As I was thinking about small things to fix around the house, I remembered we were missing this clip and figured it would be a good experiment since A) it was smaller and thinner than anything else I’ve printed, and B) it had a challenging shape — it was curved and had a narrower section that would have to be an overhang.

I took a picture of the profile of the clip, imported it into SketchUp and proceeded to trace it using curves. When I was done I exported as STL and loaded into ReplicatorG (the main software for printing to the Thing-O-Matic) to create the gcode file, which I then loaded into Pleasant 3D (a cool gcode visualizer). What I saw was that there was zero fill to the shape, despite me specifying 100% filled print. This meant that the walls of the piece were going to be too thin for the printer to work with.

I went back to SketchUp and recreated the piece using rectangular segments, each 2mm thick instead of the original piece’s thickness of 1.25mm. I also added a little bit of extra to the top joint, just for safe measure. Running this through the toolchain showed that it was going to print solid! Yay! Off to the studio.

Brief side-note: after I slammed the extruder into the build platform yesterday and recalibrated, the printer seems to be acting differently than before. It seems that the first few layers of plastic aren’t sticking together real well when they are long straight lines. This is either something odd with the extruder now or some settings that I need to change because I’m using a different plastic than the first few prints (natural color instead of safety orange, and if you recall my first print with natural had a bit of safety orange left in it). My gut is that it is the ABS difference, and I need to get the settings right for this plastic. Fingers crossed as I talk to my bot’s assembler this coming week and get his advice.

Anyway, I created a couple of variations of the item to print in different configurations — sideways with support structures and top flat on platform with no raft or support. I started with the original traced version, just to confirm that I was going to get a thin-walled piece that wasn’t going to work very well. That was, in fact, the case. I then moved on to the top flat piece and it worked exactly as I expected, except for the oddity in the first few layers.

I started printing a second copy and midway through the printing, I thought “I wonder if I printed it rotated 45-degrees if it would be better as I’d avoid the long, straight lines. So I rotated the item in ReplicatorG and saved it as a new version, and then generated the gcode. NOTE TO SELF: DON’T DO THIS WHILE PRINTING! It turns out that ReplicatorG doesn’t make a distinction between the currently printing gcode and whatever gcode you may be generating; that is, the current print data isn’t in a cache or buffer — it’s only in the text field in the app! All of a sudden the printer changed to the rotated 45-degrees version, ruining the second print. I quickly cancelled and decided to call it a day.

Once home I put the clip in the little slot in-line with the other, metal clips, and discovered that due to the extra width of my piece it was a very, very tight fit. The good part is that unlike the metal clips, this clip isn’t coming out easily. If I were able to remove the clip, I would sand it a fraction to make the fit more loose.

I put the shelf in and called the wife in. We both smiled; another experiment complete, and with good results.