ALEX HATTORI

View Original

3D-Printed Swerve Drive Pt 1

Because I don’t have many tools at home, but do have a bit of free time, I decided to get myself a cheap SLA printer from Amazon. It’s amazing how cheap they’ve gotten. For reference, a Formlabs SLA printer costs ~$2000-$3000, and the printer I got costs around $250. The huge cost reduction has to do with the curing method. While the Form 2 that I’m familiar with uses a laser that rasters over the resin bed using mirrors to position the beam, the cheap Chinese SLA printers use an LCD screen with no hard backing to mask the UV light coming from an array of UV leds below the screen. With only one moving part on the printer (z-axis), the printer can get quite cheap. These new generation of printers are sometimes known as DLP printer instead of SLA. Downsides include the fact that the UV light degrades the LCD screen over time. Also bells and whistles of the Form printers like an automatic squeegee for the FEP screen, or large build volume are obviously not included in the cheap printers.

I chose this particular printer, the Phrozen Sonic Mini, because it uses a monochrome LCD screen which supposedly is more resistant to UV light degradation and so it can cure with more light intensity and thus prints faster.

I decided to make an entirely 3D-printed swerve drive using leftover parts I had lying around and came up with the CAD above. It currently uses a 35:1 25mm brushed motor for the pivot, and an 1806 motor with ~10:1 of reduction for the wheel. The swerve uses a coaxial gear with the pivot, as well as bevel gears to get a continuous rotation swerve (like most good FRC swerves). Basically everything is printed (including shafts) because I don’t have tools. For module position feedback, I made a breakout board for a magnetic encoder. The modules are not handed, so 4 (or maybe 3?) identical modules bolt together for the robot.

Here’s a slightly more detailed look at the individual module CAD. The wheel motor goes from belt to spur, to spur, to bevel gear, and the pivot motor uses one belt. Because I don’t have a wheel position sensor and the robot is teleoperated anyway, the backlash in the wheel gears is acceptable. However, I chose to use a belt for position to minimize backlash.

And some pics of the first module that was printed and will be assembled once I get my McMaster plastite screws in.