3D printed pumpkins, ProPLA, and filament testing!


I love Halloween. I love the chill in the air, the lighthearted spookiness, and of course, and the decorations. There's something causally morbid about the season which appeals to me deeply. In previous years, I had a set of Trick-or-treat pumpkin baskets that I filled with sand and left on my doorsteps with some repurposed lawn-lights. I had a plastic skeleton posed in my porch.

Then 2020 happened. COVID. George Floyd. The Uprising. Somewhere along the way I put away the skeleton and threw out the plastic baskets where were now bleached from UV exposure.

This year, I wanted to get back into that a bit, but leverage my interest in electronics and 3D printing. So, I thought, Why not print some pumpkins and put some lights in them?

Almost all my 3D prints to date have been in Hatchbox filament. Due to my recent jamming issues, I had become unhappy with the choice and decided to try something else. While I could have ordered any number of popular filament brands on Amazon, I wanted to move away from that too and try to order something from a local supplier.

I had ordered parts previously from TH3D and Fargo 3D Printing. The latter of which is very much the closer of the two to me geographically. While they don't make filament themselves per se, they are an official supplier for 3DFuel, a popular local filament producer. One of their key products is "ProPLA" which, after annealing, has more impact strength than ABS.

While strength wasn't a key consideration for me in this decorative project, I wanted to give them a try anyways. So, I ordered a roll and went to Thingiverse to find some pumpkins. Unsurprisingly, there were several available, but I liked the design of this one the most, as it was hollow, had some cute faces with it already, and it seemed easy enough to design my own if I wanted.

Once the roll arrived, I did what I usually do, shoved it into the printer and printed a Calibration Cat. I like the calibration cat as it's a quicker print than the benchy -- a small boat commonly used as a printing test -- and it's a fun little trinket to give away. Unlike standard PLA, this ProPLA required printing at much higher temperatures, almost like ABS. The first calibration cat I printed, however, turned out as good as could be expected for an unknown filament.

Still, I wasn't yet ready to accept that the filament was fine and move on to printing some pumpkins. I then turned to the tried-and-true benchy. It's a longer print, but it is definitely a more comprehensive test. Things looked promising at first, but as soon as the print got to the boat's cabin, things went wrong.

The extruder was struggling and clicking. All the thin sections on the cabin came out either blobby or weak. One was so weak I could snap off the top with my fingers and little effort. Wasn't this supposed to be better filament? Was I just doing something wrong? Was the printer not up to it? It was already mid-October and I didn't want to spend an entire month troubleshooting as I did before. I needed to get these decorations done at least a week before the 31st.

Disappointed with the print quality, I decided to try running the filament through a series of tests instead to get a feel for how best to print it. After some searching, I came upon Teaching Tech's 3D Printer calibration page. The information on the page itself is generally useful, but the key feature is a series of gcode generators which can be used for practical filament tests. Since I only had an afternoon to spend on this, I focused on the two aspects I felt I could control the most: temperature and retraction.

Temperature and retraction can have dramatic effects on the quality of your prints. Too much heat can cause the filament to burn, which can result in jamming. To little heat can result in high backpressure and also lead to jamming. Retraction is a less intuitive parameter. When the print head is moving while not printing, it is best to back off the filament in the extruder, creating a small void inside the nozzle. The key effect of retraction is to control print artifacts like stringing, where hairlike threads of plastic show where the extruder moved from place to place on a print when not laying down plastic. I was hoping that by running these tests, I'd get a better sense of how the ProPLA worked, allowing me to print with a higher rate of success when I moved on to the pumpkins.

Consider donating to us on Ko-fi or subscribing on Patreon! It helps us buy materials, pay for hosting, and continuing making projects and posts like this one.

Interestingly, the temperature test was generally good across the entire range of 210-250C. The results were generally the best around 230C, with minimal issues and the truest of lines across the unsupported middle of the print. The retraction test was much more demonstrative. Anything higher than 1mm of retraction produced good results, although the best results appeared to be at 3mm.

I assumed I had that sorted, and then reprinted the benchy. The results were even worse than before, including a completely jammed extruder. When I finally managed to get the filament out of it, I found after the toothed gear, the filament seemed to get noticably bigger in diameter. So big, in fact, that the extruder couldn't advance it any more.

Perhaps I should ask the Dark Lords Of Printing for guidance...

I decided to do more research, hoping that maybe, someone else was experiencing similar issues as I was. What I learned is that "ProPLA" isn't exactly PLA. It's mostly, PLA with additives and other plastics which can increase its strength. The feel of the plastic right off the spool is notably different than standard PLA, and it should be treated as a specialty filament.

One post suggested backing off on the extruder spring (the tensioner) to give the filament a bit more wiggle room against the toothed gear which advances the filament into the hotend. That made sense to me, and I took apart the extruder to try it when I discovered something. My extruder had no way to adjust the spring pressure. In fact, the spring was just captive by two screws, neither of which compressed the spring further to add more tension. Near as I can tell, the aftermarket metal extruder I have was missing a key part which allowed the tensioner to function. When I took apart an unused dual gear extruder, the missing part became apparent.

I was about to switch to the dual gear, when I realized that it wouldn't fit on my modified printer. A bolt was in the way. So instead, I decided to borrow the spring from the dual gear extruder, which had more pressure than the one I was currently using. This added more force to the tensioner. While this was the opposite of the advice from the post, it was working from the assumption of at least some tensioner pressure, rather than nearly none as I had.

The same post also suggested that ProPLA was susceptible to retractions, so I turned off all retractions and printed another benchy.

While the result was stringy as all heck, it was a complete benchy and showed none of the issues as before.

This process took an entire afternoon to sort out, and it didn't seem like either temperature or retraction was the key problem here. Instead, the problem was an incomplete extruder assembly resulting from insufficient pressure against the extruder gear from the tensioner.

Taking a guess with retraction at 1mm and the temperature around 240C, I started printing. Despite the earlier tests, the result was still very, very stringy, but it was serviceable. After a few days of starting the print in the morning and shutting it down after dinner, I had two very good pumpkins.

Once I had these two, I started thinking about how to set up lights inside them. The easiest choice would be to buy one of several pre-made LED lights used for decorations. You can get them a number of different shapes and shades, ranging from candles, to little waterproof disks. Many, however, seemed overcomplicated or simply had features I didn't want.

Ideally, I wanted them to:

  • Run on rechargeable AAs
  • Change colors, but not quickly
  • Have a very long runtime
  • Not have any extraneous features like a remote control

And none of them seemed to suit.

While I sat on that problem, I turned to printing the remaining two pumpkins of my planned four pumpkin set. The remaining faces included with the original Thingiverse design, however, weren't all that appealing. After the last year, I didn't want something too ghoulish. While the included faces were good, I wanted something cuter.

I thought about this while sitting with my cat one night and it hit me. I should make my own designs. The Thingiverse design did include a plain back which could be importing into modeling software. I drew up a simple cat face as an SVG in Inkscape, and then imported that as a file into Blender. Using a boolean operator, I cut out the SVG face from the model. Once saved, I could import that into a slicer and send that to the printer.

When that worked, another opportunity hit me. One of my favorite games is the Animal Crossing series. I've been a fan since New Leaf and found it a good way for me to relax and disconnect. I may be unusual as a player because I'm not terribly interested in the acquisition or design aspects of the game, to me it's more a therapy exercise. So, when Nintendo did provide a template to carve their in-game avatar of Halloween, Jack, I thought, I bet I could print that.

Instead of importing the graphic, I did the same process as I did with the cat face. I created the SVG whole cloth, only using the carving template as a reference. Then the same process could be used to create the necessary gcode for the printer. It worked amazingly well, and it's a regret that due to licensing issues, I cannot release the SVG or the STL files for others to print.

Now I had to solve the lighting issue. I still didn't like my options I found online, nor did I like any I found locally. The natural conclusion I was arriving at was to build them myself. I do have a quantity of LEDs, as well as the parts to construct several "joule thief" circuits capable of lighting them for days and days. But, none of the LEDs I had were multicolored. For that, I thought, I'd need a microcontroller and more money than I'm willing to spend on this project.

Then I found out that two wire, multicolor LEDs were a thing. These have the form factor of a single-color LED, but have internal circuitry to change colors only requiring power. I ordered a pack of them for $7 and hooked them up to my prototyped joule thief.

It...didn't work. After talking to some people online, the consensus was that the joule thief wasn't giving the LED enough power, and due to the noisy output, was "resetting" the internal microcontroller of the LED, resulting in them being stuck on the initial red color. I decided instead to just power the LEDs directly, but I'd need more than one AA for each.

Getting the batteries wasn't so much of a problem as was the lack of holders for the batteries. I only had ones for a single cell, and not enough to start with. To solve that problem, I turned to 3D printing once more. This battery holder design only needs some wire for contacts, and uses a clever bit of plastic to provide the spring action to press the contacts against the battery. Instead of soldering each circuit, I bought a set of four tiny, adhesive backed breadboards and set up the light for each. The result was excellent.

While the ProPLA was challenging, I don't think I'll be buying it again. The experience, however, suggested I try standard PLA 3Dfuel filaments instead. They were much more my speed. I bought a forest green spool and used it for the stems, as well as some shelving brackets on which to hang them in the porch. Furthermore, I also hung up the skeleton, with some additional lighting which clicks on in darkness.

The project wasn't all that complicated in the end, but it was a good learning experience and relied on multiple disciplines to succeed. It was a fun way to test my skills and enjoy the season just a little more.