Speed Racer Testbots

It’s hard to find a better test print than the Ultimaker robot; it has insets, embossings, overhangs, bridges, posts, and fine features, all wrapped up in a model that’s less than an inch and a half tall. Plus, it’s cute. Here’s marrit‘s remix that includes removable supports for the hands, based on Siert Wijnia‘s original design from YouMagine:

Screen Shot 2017-01-16 at 1.27.39 PM

This robot is small and prints quickly, but… on an Ultimaker with standard Cura settings, not quickly enough! Theoretically, Ultimakers should be among the fastest desktop 3D printers, because with their Bowden tube construction they have the heavy feeder motor mounted on the back of the machine instead of weighting down the print gantry. Ultimaker is a Dutch company, and from what I can tell the Dutch are willing to slow down to make things nice, so their Cura settings are optimized for mind-bending quality but not so much for speed. On the other side of the pond, I’m an ugly American and I want speed like Racer X. Time to put theory into practice and speed things up.

How fast is fast?

On my old friend the Replicator 2, the default “High” quality settings print at 0.1mm layer height, “Standard” at 0.2mm, and “Low” at 0.3mm. Back in the MakerHome blog days, nearly everything I printed was at the 0.3/Low setting, because that was fast, sturdy, and good enough for me. In contrast, Cura’s “Fast” default setting for the Ultimaker 2+ prints twice as nicely and twice as slowly at half of that layer height, 0.15mm. Ultimaker’s “Normal” is 0.1mm layer height (same as the Replicator’s “High”), and Ultimaker’s “High” is 0.06mm (what?!). Because of these high-accuracy defaults, people sometimes think that Ultimakers print too slowly, at least compared with other printers they mght be used to. For example, the Ultimaker 2+ prints the test robot in 29 minutes on its lowest quality default Cura setting, or 96 minutes (!) on its highest. In contrast, the Replicator 2 can print the test robot in just 16 minutes at its lowest default setting, or 49 minutes on its highest:

FullSizeRender (43)

Testbot bases

The bases under the robots in the picture above were made in OpenSCAD and allow you to enter three lines of text on four different sides of the base, with or without a robot printed on the top. For my testbots I printed the robots separately from the bases and then glued them on.  The times listed on the bases are just for printing the robot, not the base, and the settings listed around the base describe the slicer settings, nozzle size, machine model, filament type, or whatever you want.

Screen Shot 2017-01-16 at 12.11.32 PM

The OpenSCAD code for these bases was created using the same method described in the Customizable Beefy Trophy + Blender Bake Tutorial on Thingiverse. This means that if you want to make your own testbot bases, then you can design them right in the Thingiverse Customizer instead of having to deal with the OpenSCAD code. Follow the link below to get started.

Screen Shot 2017-01-17 at 12.37.42 AM

Speed it up

Our goal is to identify simple Cura settings that allow our Ultimaker 2+ to print as fast as the old Replicator 2.  The process is: Play with slicer settings to see how fast we can speed things up, print a lot of test robots, figure out where that quality line is that we don’t want to go under, and then create and print a Speed Racer Testbot base with the corresponding settings and glue that robot to the top of it. Here’s an army of test robots we printed as we experimented to make Cura faster and faster. Can you print an 8-minute robot on an Ultimaker 2+ with an 0.4mm nozzle? Yes, but as you can see, it isn’t pretty.

FullSizeRender (44)

We decided we were most happy with the 17-minute robot for its balance of quality and speediness, printed at 0.2 layer height. That’s just one minute longer than the same robot printed at the lower-resolution default 0.3mm/Low setting on the Replicator 2. Even the 14-minute robot we tested looked pretty good, but I guess I have some Dutch in me after all, because I wanted that tiny bit of extra quailty and chose the 17. :)

So, what settings did we end up using? There are a LOT of settings in Cura, and in the newest version you can customize your interface to include whichever of those settings you like. (Pro Tip for Cura 2.3.1, or what I couldn’t figure out until I actually asked someone where all the settings were hiding: Click on the “gears” that appear when you hover over the setting submenus, then check boxes to indicate the settings that you want to add to each submenu.) There are some pretty fancy things you can do to speed up print times, but I wanted to find something that would be easy for even a beginner to do quickly. In the screenshots below you can see the settings I decided to focus on: Layer Height, Wall Count, Top/Bottom Counts, Infill Density, Print Speed, and Minimum Layer Time. The leftmost screenshot shows the default Cura slicing profile for 0.1mm/Normal on an Ultimaker 2+, with our focused slicer settings made visible. At this setting, Cura estimates that it will take 58 minutes to print the test robot. The middle screenshot shows Cura’s default for 0.15mm/Fast (29 minutes), and the rightmost screenshot shows our custom settings at 0.2mm layer height (17 minutes!).


Here’s how the robot prints under each of those profiles. The 58 minute robot is amazing, and the 29 minute robot is really great, but if you’re iterating a design or want to print quickly, then the 17 minute robot is the way to go.

FullSizeRender (41)

These printed Speed Racer Testbots will help us share best practices and slicer information with students and faculty who use the 3D printers at JMU 3SPACE.


Let’s collect all the data in one place for easy reference. The yellow columns of the table below show the values of our focused silcer settings for the three yellow robot Ultimaker 2+ prints shown above. The green columns show the corresponding values for the default High and Low Replicator 2 slicing profiles.

Rep2 High Rep2 Low UM2+ Normal UM2+ Low UM2+ Custom UM2+ 0.8mm
Layer Height 0.1 mm 0.3 mm 0.1 mm 0.15 mm 0.2 mm 0.3 mm
Wall Count 2 shells 2 shells 3 shells 2 shells 2 shells 1 shell
Roof Count 10 layers 3 layers 8 layers 5 layers 4 layers 4 layers
Infill Density 15% 10% 20% 18% 5% 5%
Print Speed 40 mm/s 40 mm/s 50 mm/s 60 mm/s 80 mm/s 50 mm/s
Min Layer Time 5 sec 5 sec 5 sec 5 sec 1 sec 1 sec
Print Time 49 min 16 min 58 min 29 min 17 min 9 min

The setting that makes the most difference will almost always be Layer Height. For example, consider that doubling the Layer Height will cut in half the number of layers that have to be printed, which in turn could cut the total printing time in half. Although it is possible to print at 0.3mm layer height on the Ultimaker, the printer just doesn’t seem to be optimized for it. I had trouble finding reliable settings to go with 0.3mm layer height, which is why we used 0.2mm in our UM2+ Custom settings.

In general, I’ve found that you don’t want to set Wall Count below 2 shells, unless you are using a larger nozzle (as in the last column of the table). The “Roof Count” shown in the table is my shorthand for the Top and Bottom Layer Counts. I’ve found that Infill Density can be lowered a lot, especially if your model doesn’t have a lot of large, flat roof areas or thin, fragile columns. If your model is wireframe or spindly then I’d actually recommend bumping up the Infill Density, even if you’re trying to speed print; it won’t cost a lot of time but it will make your model a lot less breakable.

Print Speed is something you can experiment with; when your Layer Height is lower then you can increase Print Speed even more, because smaller Layer Heights require less filament to be extruded. My tests showed that for 0.2mm Layer Height, 80 mm/s is the highest you can go without sacrificing quality.

Finally, the “Min Layer Time” setting can also make a big difference if your model is small; it measures the minimum number of seconds that the printer will spend on each layer. A layer that prints very quickly might not have time to cool, so the printer will actually wait until the minimum number of seconds have elapsed before it moves to the next layer. By reducing this number we can try shaving off those extra seconds, which can add up if many layers of your model are small.

That part in Mythbusters where they take it a little too far

Can we go even faster? YES; if we bounce up our nozzle size to 0.8mm (hooray Olsson block!), then we can print a superfast rough-draft Speed Racer robot in just 9 minutes.

FullSizeRender (42)

We can actually print one in *6* minutes, but quality starts to suffer. In the photo below you can see a comparison of the 9-minute and 6-minute prints from the back and the front. Remember that these are printed with an 0.8mm nozzle and these robots are just over an inch tall, so we don’t expect a whole lot of quality in this test; considering the large nozzle size, the quality of the 9-minute robot is actually pretty good.


What about the other Cura slicer settings? There are lots, and you can probably do better than I did above, especially on a per-model basis when you could optimize for the particular features of the design you are printing. Or, you could do worse, as I did when I accidentally unchecked “Enable Print Cooling” in one of the tests:



The short version of the long story above is that if you want to quickly set up Cura to slice quicker prints for Ultimakers, use the type of slicer settings in the UM2+ Custom column of the table above. Specifically:

  • Bump up Layer Height to 0.2mm.
  • Bring down Wall and Roof Counts as much as possible.
  • Reduce the Infill Density unless your model has a lot of thin columns or wires.
  • Experiment with increasing the print speed, keeping in mind that you can get away with faster speeds at lower Layer Heights.
  • Reduce the Minimum Layer Time, if your model has a lot of small layers and if cooling does not seem to be a problem.
  • Bonus speed tip: Switch platform adhesion from Brim or Raft to Skirt (the one that just makes a ring around the model). Or, if you have to use a Brim or Raft, decrease the Brim Width or Raft Extra Margin setting.

Good luck and print faster!

P.S. to Steve Albini / Big Black fans: I managed to get two Hammer Party references in earlier so here’s a third one: Cables!

Customizable Snowflake Cookie Cutters

For the past three years we’ve made a snowflake design for the holidays: In 2013 it was Snowflake Ornaments, a set of simple designs created by extruding an SVG image of snowflakes. In 2014 it was the Snowflake Cutter, which leveraged the image sketching functionality of the Customizer to mimic the way snowflakes are cut out of folded paper. In 2015 it was the Snowflake Machine, which provided a way to procedurally generate over a billion unique snowflakes in different styles. 

snowflakeornaments_preview_featured  photo_preview_featured  large_flake_3_preview_featured

We really weren’t sure where we could possibly go from there, until @matschi suggested the great idea of turning the Snowflake Machine into a Snowflake Cookie Cutter Machine!

56c65c935c82a0164efe81f348d9978f_preview_featured (1)

With these Customizable Snowflake Cookie Cutters the design of the Snowflake Machine is restricted to cookie-friendly patterns, and the model generates with one sharp side and one flat side (for pressing down on). From the Thingiverse link, click the “Open in Customizer” button, then click anywhere in the slider to make a snowflake cookie cutter. If you like it, click the “Create Thing” button, then come back and make some more. If you want a different design, just click somewhere else in the slider. Repeat over one billion times, or until you have enough snowflake cookie cutters for your holiday plans.

How to make cookies

Before you do anything, watch the sugar cookie video clip from the Cookie Clause episode of Good Eats, so Alton Brown can show you the basics of making sugar cookies. Trust me, whatever sugar cookie recipe you use, this video will help you do it successfully. Because your Snowflake Cookie Cutters will be somewhat detailed, it’s best to use a very firm sugar cookie dough recipe. I tested my cutters with The Best Cut-Out Sugar Cookies from The Kitchn.

c9dc6828b156878348ab45ece0263c7b_preview_featured (1)

Because I am lazy and the cookies were already beautiful and detailed on their own, I didn’t bother to decorate them; I just mixed some confectioners sugar and milk to make a glaze and poured it over the cookies.

7cdce3ebcb5426e33a4648b4fafc5db4_preview_featured 655806fdbfed886eb296effe0ec297ab_preview_featured

As you can see from the pictures, I might have burned the cookies a little. They were delicious anyway.


Cube Puzzle Quartet

This year at Maker Faire Bay Area we hung out at the Ultimaker booth and offered a challenge: Solve one of these 3D-printed Cube Puzzles and you get to keep it!


Of course, we took the pieces out of the boxes first:


The 3D-printed pieces of each Cube Puzzle can be printed without support and all four puzzles and the box container can be downloaded from the Cube Puzzle Quartet model on Thingiverse, or from YouMagine. All four of them are classic, known puzzles that you can read about in Stewart Coffin’s excellent book Geometric Puzzle Design:

Green = Easy:  The Soma Cube has 240 different solutions, so it’s easier to solve than the other puzzles. This puzzle was very popular in the 1960’s, and people enjoy assembling the pieces into many other pleasing shapes like stairs or pyramids.

Orange = Medium:  The Hoffmann Cube is the earliest known 3x3x3 Cube Puzzle, created in 1893. This puzzle has only six types of solutions, up to symmetry. One interesting thing about the Hoffmann Cube is that all of its pieces are flat!

Silver = Hard:  The Coffin Cube has only one solution and it is also known as the Half Hour Puzzle, created by Stewart Coffin. All four of the puzzle patterns used in our Cube Puzzle Challenge are from Stewart Coffin’s excellent book Geometric Puzzle Design.

Gold = Diabolical:  The Nob Cube also has only one solution and it is notoriously difficult to solve. Some puzzle makers craft this puzzle out of wood and it is said that the puzzle is so difficult that solvers sometimes resort to examining the grain of the wood to figure out how the pieces fit together. Are you up to the challenge?

Project: Design Your Own Cube Puzzle!

Here are some ideas for projects where students use their favorite 3D design software to create new Cube Puzzles. This project can be adapted for many different educational subject areas and at a variety of levels:

Beginner:  Use Tinkercad or Morphi to place boxes together to construct puzzle pieces that fit together into a 3x3x3 cube. How do you know they will all fit together into the cube? Print out your puzzle and have fellow students try it out to rate its difficulty.

Intermediate:  Use Fusion 360 to create puzzle pieces, and use fillets and offsets so that the pieces are pleasant to hold and fit together nicely. Create a container box that will hold the assembled Cube Puzzle. Who can make the most difficult puzzle?

Fundraising:  Print out students’ Cube Puzzles and create a solving event where parents and faculty try to solve the puzzles. Raise money for your school or classroom by charging per solving minute, having solving championships, or selling the Cube Puzzles themselves. You can inscribe one of the puzzle pieces in each Cube with text to personalize it for your school or event.

Designing with Code:  Use OpenSCAD or BlocksCAD to create pieces for a Cube Puzzle by translating and arranging cubes. Add code that creates clearance for where pieces will touch each other. Make this clearance customizable so that it can be easily changed for printing on different 3D printers or with different settings.

Exploring Mathematics:  Which other 3D shapes could be used in this type of puzzle? What types of polyhedra can fill space, and which of those can be used to create a puzzle whose pieces can slide together? As a starting point, think about 2D space-filling shapes and puzzles by exploring tessellations and tangrams.

Advanced:  Use the software BurrTools to construct a Cube Puzzle, analyze it for constructability, and test how many solutions it has. Can you find a new 3x3x3 Cube puzzles that has only one solution? You can also try your hand at more advanced puzzles like interlocking Burr Puzzles or puzzles with unit shapes other than cubes.