One of the things that interests me about 3D Printing is using it to augment existing fabrication techniques. I’ve already done some experiments with 3D printed joints for wood structures, but I’ve wanted to try building something with structural integrity. Printed PLA isn’t the strongest material, but the idea was to utilize its compressive strength, and minimize bending forces. I also wanted to restrict myself to only making 90 degree cuts in the wood, and minimizing the number of cuts I made.
I started with the joint for where the seat would join the legs. The goal was to make the joint easy to attach to both the legs and the seat, with holes to allow for easy sinking of screws.
The star shape creates a straight edge on a side for any 45 degree rotation. There are 3 holes in the bottom which allow for screws to be placed that go directly from seat to leg (for strength).
After printing, everything went together very quickly.
The problem now was that the legs bowed out too much when weight was placed on the stool. To solve this I decided to attach more supports to each leg, to prevent them from bending outwards (which was threatening to break the joints).
At this point, the only thing left to do was to add some feet and do some light finishing.
I think I could have made the stool a bit more aesthetically pleasing, but for a first attempt I’m pretty happy. It feels solid and sturdy when you sit on it. I also managed to only make seven 90 degree cuts in the wood (4 for the dowels, 3 for the legs, 1 for the seat).
A wrote a genetic algorithm that generates a tree like structure. This structure can be turned into a mesh using convolution surfaces.
Given a list of control points, the GA tries to optimize 3 things:
- Minimize the distance from each point to the closest branch.
- Minimize the sum of the length of each branch segment.
- Minimize the angle between adjacent branches (avoid sharp turns).
The results are very interesting:
I also tried printing the result. It took a lot of support material, but I think the result is pretty good (I had not yet cleaned all of the support material when I took the photo).
Since I purchased the Conscendo S, I’ve played around with a number of different FPV setups. At first, I took the camera/transmitter from the Eflite FPV Vapor and mounted it on the top hatch. This worked alright, but I quickly grew tired of having a range of only a few hundred feet. I upgraded to a better camera and more powerful transmitter, which worked great, except for the hacked up mount made from foam and duct tape. I found a model for a mount for my camera, and with a few changes to increase the height off the fuselage and a quick print, I had a much cleaner setup for my plane.
Also visible in the above image is the Mobius mount. Here is a video from the first flight with the new setup:
I was unhappy with the angle of the mobius mount, so I went back to the drawing board and came up with this:
I used a dremel to remove the old mount, and after some more printing and gluing, I have a mount with an adjustable angle.
This structure is about 10 feet tall, has 60 3D-printed connectors, and 168 wooden dowels. The connectors took somewhere between 150 and 200 hours to print on my Replicator 2 (including failed prints). Currently the dowels just press fit into the connectors, but I will probably glue them at some point. This project was inspired by my desire to use my 3D printer to build a large-scale object.
The very first thing I printed on my Replicator 2 was one of the default models included on the SD card, a chain.
Default Model for Replicator 2
Chains are a pretty cool example of what you can easily create using a 3D printer. Trying to machine something like that would be a much more difficult task, and certainly couldn’t be done in 12 minutes.
Inspired, I made a couple of my own chains in Tinkercad.
First Chain Printed
Second Chain Printed
As cool as this is, there are readily available parametric chain generators written in OpenSCAD that are much more practical and powerful, so I lost interest pretty quickly.
But there is more than one type of chain, and they don’t always have to be connected during the print. Thus began my ongoing interest in snap-together 3D-printed parts.
I started with a fairly simple two-part chain built in OpenSCAD:
First snap-together parts
From there it quickly led to what I just created!
Chain Link in OpenSCAD
As a programmer, it didn’t take me very long to look into programmatic modeling tools after I got my 3D Printer. And it turns out there are some pretty good options available. Here’s what I have done so far:
OpenSCAD is a programmatic CAD tool. It’s essentially an editor that lets you code in a simple programming language, with a display window in which to view your model. The simplicity of OpenSCAD is very powerful, and within a few minutes I was creating new designs.
The first model I made was a simple spiral staricase. In 10 lines of code I was able to produce this:
Tiny Little Staircase
My apologies for the low quality image, but I hope the point comes across: programmatic modeling is awesome! Later, in quite a few more lines of code, I produced this:
Weird Basket Thing
OpenSCAD is fun, but its language (and editor) are a bit lacking, and I quickly began missing the advantage of a real programming language.
Getting more power out of OpenSCAD
OpenSCAD has a feature which monitors a .scad file, and re-renders the file every time it changes. This is great for using your preferred editor, and especially great if you want to programmatically generate .scad programs.
There are a few scad generation libraries around, but I went with SolidPython. Now that I was using a more powerful language, I had much more freedom with my creativity. My first piece:
Brownian Motion Vase
I’m very pleased with the result of this print. I’m continually amazed by the overhangs I can get away with.
And my most recent project, a procedurally generated tree (again, the overhangs never cease to amaze me):
Printing Static Models is fun, but I’ve quickly become infatuated with the idea of printing things that move. I’ve only scratched the surface, but here’s what I’ve done so far:
After a brainstorming session with my friend Zack, one of the ideas we came up with was printing louvers. The idea would be create a panel or a brick or something that could be built into a larger structure, and the object would contain hundreds of little louvers that could be activated to shut out or let in light and air.
Eventually we want to be able to print a solid piece and have it be functional without assembly (possibly using a dual extrusion printer with dissolvable material). For now I’ve just prototyped a 2 piece assembled louver panel. I designed it using tinkercad, which is super easy to use.
Louver Design in Tinkercad
I made the hole for the axle 5mm in diameter, and the axle 4mm. This worked pretty well, but the axle is a little loose in the socket.
Last week I bought some assorted nuts and bolts, with no clear idea of what I would use them for, except that I thought they would be useful in assembling 3D prints. I designed this build explicitly to experiment with multi-part builds and moving parts. Again, I designed the model using tinkercad.
Design of Car
Partially Assembled Car
One Side of the Car
The other side of the Car
This one took a couple of iterations to figure out how everything would fit together. At first I made the axle too small to fit through the wheel. PLA doesn’t have very good traction, so next time I’ll probably buy little rubber wheels.
I recently acquired a Makerbot Replicator 2 3D printer, and I’ve been playing with it a ton. Here are a few things I’ve learned along the way.
Filament sticks to the Build Surface really well.
I’ve found a few things to keep it from sticking:
- Oil from the Skin on my fingers – Works alright for the main print, but is very useful for the initial strip of filament it lays down on the front edge. I rub my finger along that edge before every build.
- Hand Cream – Works better than Skin Oil, but still not great. I Probably should experiment with other lubricants, but that’s all I could find in my apartment.
- Painter’s Tape – Makerbot recommends using painter’s tape and laying it on the build surface. They include a few large sheets of tape as well. This works well, but the tape tears easily, requiring frequent replacement. Originally I would use a spatula or a flat head screwdriver to wedge under the print to get it off, but doing this quickly destroyed the layer of tape. I’ve found that lightly twisting the print using pliers works better and does less damage to the tape.
- Printing with a Raft – The raft seems to be easier to pry off the surface in some situations, and it peels off of the print surprisingly well. The MakerWare software makes adding a raft really easy.
The edges of the acrylic plate are sharp.
I gave myself a nasty cut when my finger hit the plate while trying to dislodge a print. I’m very careful now when I’m prying off a print, and I’ll often remove the build plate if a print is stuck too well.
Always watch the first layer of a print.
Almost all of my failed prints failed in the first minute of printing. During the warm-up stage, some material oozes out, and sometimes this can get stuck on the extruder head. I’ve always caught this early, but there are horror stories about the extruder head getting covered in filament.
Overhangs work amazingly well.
I’m continually impressed at the overhang angles that I can print. On sharp overhangs you can get some drip artifacts, but I’ve yet to have a build fail due to overhang.
Printing from the SD card is prone to errors.
This was true for me, but I’ve read that this can be an issue for some people more than others. I would get SD cards errors approximately once every 45 minutes of printing, which was terrible for long builds. I switched to using USB from my computer, and it’s worked great ever since.
Overall, I’m incredibly pleased with my printer, and can’t wait to learn more.