Printing PET-G test

Posted at 2016-07-11.

Tarlab has a 3d printer, as hackerspaces often do. It hasn't been there long and I never tried using it before. Now I decided to give it a try, mostly following the instructions and my previous experience with other printers.

The printer is a Prusa Mendel i3 with a glass topped heated bed and metal frame. Probably kit built and modified very little if any.

Choice of models and materials

I didn't want the added step of changing filaments, so I decided to try whatever was loaded, which was labeled as "PET-G" and looked like a common spool of clear plastic filament.

I had a small part (prism nut for Unimat) that I wanted to print, but I thought I should also pick up some small test/reference part.


I sliced the models with Cura (which I was most familiar with and had with me on a laptop) and printed them off an SD card I brought along. I used a small brim and tried 20% and 10% fills on the nut, IIRC.

Printing was totally uneventful, parts came out fine and came off the board with fairly little effort and a carpet knife blade. We should probably get some sort of spatula for lifting objects. Surface was treated with glue stick and ABS juice.

I sliced the brim off with a knife. A brim is probably not necessary, but I like the idea that some "waste" plastic is extruded before the actual print starts.

The hollow cube bridged the top beautifully and looks fairly even all around. Probably not the finest print I've seen but far from many worse ones, even considering only succesful ones.

Testing the parts

The parts came out square and with the dimensions that were in the original OpenSCAD models as far as I can measure these things. A bolt hole and captive nut recess on the prism nut were totally useless, though. The hole is easily drilled out, but naturally having a captive nut recess working out of the box was one reason I wanted to test printing this part. A hole edge facing the bed surface was also considerably worse than a hole coming out the top. The bottom layer(s) gets squished some.

The parts feel very solid, even at low fills. The cube was tough and I couldn't crush it with little effort in a light vise. I didn't want to try harder. I know it would break eventually, but now I know it doesn't just fall apart easy at this size and shape. The nuts are very tough despite being almost hollow. 10% fill means a few little "wings" radiate out from the bolt hole. 20% also placed some extra wall farther out.

I tried fine wet sanding to polish the cube. It didn't work too well. It does flatten the surface some, but I think the plastic is too "bouncy" to cut easily.

Flame polishing caused the cube to bulge. It might be the air chamber inside, thermal expansion, or something else. Having a more controlled temperature might work. Too low will probably cause the part to sag as it approaches the melting point and too high will cause the warp I saw. It didn't immediately char, though.

I'm more interested in dimensionally good solid parts than smooth outsides anyway. There should always be the option of paint or other finish anyway.

The cube sides drill very easy and clean, leaving neat holes. Tapered reamers and stepped drills cut very well as well. Carving with a knife works, although the plastic is pretty hard. A hand nibbler also cuts cleanly leaving square edges and doesn't seem to cause layers to part or anything.

Using a bolt to attach the piece by one wall makes a pretty solid feeling joint which does flex with enough force and the bolt eventually leaves a mark on the plastic. Printed side to plywood had very high friction. I wouldn't expect mountings to fail easily.

A self tapping screw self taps to a pilot hole much like it would in wood, if not better. The hole dimension might be more critical in a thick part as the plastic won't compress like fiber. It also hold as well if not better. Of course the plastic is massively tougher considering the wall thickness.


It's very nice. This totally unquantified subjective test makes me think I can use this material and printer to make parts if I want. I won't worry about little bridging, but I will worry about bottom layer and hole dimensions.

I should definitely test other slicers, parameters, and plastics sometime. Fablab also has printers that might be available to try.

I don't have much in the way of applications at the moment though. For one, I probably want some custom case sides for one of the raspi boards that has a display on top. Not that I have an application in mind for that either.