Skip to content

Week 12 - Machine week

I had little time during machine week, but I still wanted to continue and get going with some crucial components for the new final project.

Drive considerations

There are three options that came up into my mind for gear that have a high ratio, high accuracy and are self locking:

  • worm gear
  • harmonic drive
  • cycloidal drive

The worm gear is a simple gear where the ratio can be adjusted over a very wide range, by selecting the right number of teeth and pitch angle og the worm gear. Usually it has some backlash, but I think it might be tolerable. I could purchase components like a metal worm and use a 3d-printed worm gear or get a motor with attached worm gearbox. One disadvantage is the 90° angle of the motor axis towards the worm gear axis. This requires more space outside the worm gear.

Looking at harmonic and cyloidal drives - the obvious advantage is the high gear ratio, low backlash and overall great properties for high accuracy positioning applications. However, those gearboxes are very expensive and consist of many parts. Purchasing isn't an option, but making my own gearbox might be an interesting (but time-consuming) task.

The harmonic drive migth be a bit easier to design, but definatly requires some effort. The big downside I see is the flexing gear (wave generator), that needs to deform constantly and may fail at some point, especially when 3d printing it. There are however some people, who have used a timing belt for the wave generator, which sounds like a reasonable approach. Generally it will be easier to achieve a higher gear ratio with the harmonic drive then with the cycloidal drive. If found this very detailed tutorial on a cycloidal gearbox for a NEMA17 stepper motor. Because I not only want a working protype for some hours, but a machine that can operate at least for some time, I need the gearboxes to be reliable.

Let's make some quick calculation on the speed.

The azimuth will rotate \(360°/24h\) (one revolution per day) or in other words \(0,25°/min\).

\(n=6,944*10^{-4}rpm\)

This indicates, that the motors will most likely run intermittedly.

Positioning

There are two ways of positioning the axes.

  • absolute encoder on the axes
    • requires no homing
    • no accumulating error
    • expensive
  • encoder on motor (closed loop)
    • cheaper solution, requires less accuracy
    • requires additional home position sensor

I would rather go for a good absolute encoder with high resolution, but they are either too expensive (industrial ones cost several hundred dollars) or don't have a thru hole for the cables.

Endless rotation

One of many challenges in this project is to think of how to solve the routing for the cables.

The shading object needs to be moved up an down, while rotating, so the second (horizontal) axis needs to rotate around the first (vertical) axis. For that reason cables need to be routed through the center of the first axis drive, hence limiting the available options.

The motor needs to be placed either excentric or have a hollow shaft.

It would be ideal, if the platform for the instrument wouldn't rotate to simplify the cable routing.

After doing some research I Realized, that all the commercial sun trackers don't have endless rotation. Instead they must return once every day. Even though I don't like that design (both because of the loose cables being tossed around and gap in the measurement when returning during summer), it seems to be acceptable and at least for the first spiral I should consider going the easier way.