Last winter I decided to challenge myself to design a 3D printed clock from scratch using Blender and the Precision Gears addon (https://makertales.gumroad.com/l/PrecisionGears). From the onset I had a few goals in mind: Most of the gears should be planetary gears (because the look cool) It should have a 3 Hz pendlum (to avoid needing to print a split pendlum) Powered by USB (I have a USB power station on my bedside table) While my original goal was to have a spring power clock, I quickly realized that the extra friction introduced by planetry gears meant a "1 week wind" style clock would need a spring that would rip the 3D printer clock apart when wound, so I shifted to an induciton circut dirctly driving the pendulum. After serval prototypes and re-designs this is what I ended up with coming out of Blender (each part is assigned a random color to make it easier to see the parts when designing). For the drive circut I original wanted to use a simple one off of Amazon (as recommend for Steve's EM clock build), but the ones that are availbe in the UK are just made up of a single transistor hooked up to and inductor. This means it only provides push to the pendulum in one dirction, this is fine when you have a longer pendulum (e.g. 2 Hz), but for a shorter one it did not provide enough ampliduded to the swing. So instead I used a pi pico microcontroluer, hooked it's ADC pins up to the inductor dirctly, and wrote about 20 lines of python code to detect the maget on the pendulum each swing and give a push of the correct time to match the freqency. This gave much better control over the amplitude of the pendulum. This could be acheaved just fine with a analogue circut, but the pi pico was cheaper than the raw parts that would of taken. With the drive circut figured out the last hurdel was turning the swing into circular motion. For a 2 Hz clock you can using a simple ratchet that pushes the second wheel every other swing (i.e. 1 tick per second), but for a 3 Hz clock that would look odd. Eventually I ran across the 2-way to 1-way system that is typically used for the automatic winding mechnism in watches. This sytems alows the pendulum to puse the second hand forward regardless of what direction it is swinging. One down side to this kind of drive mechnisusm is the clock is tied to a very small range of pendulum amplitudes to keep accurate time (i.e. one ratche click per swing). But it turns out this ratchet design has a bit more wiggle room than I expected. By adjusting the white gears thta have the ratchet paw attached to them further clockwise I can reduce the friction the paw has against the ratchet wheels, in turn increasing the pendulum amplitude. The trick is fining what angle to set those two wheels at to tune that friction to give one click per tick. So far it has been running great for about 5 months. Every so ofter the screw holding the top ratchet wheel to the pendulum rod comes loose and I need to tighten it. Also now that summer is approching I had to adjust the the ratchet postion to keep accurate time again (I assum the thermal expansion of all the gears has added more friction to the system). Thoughout this process I have learned a lot about clock design and clock drive systems. I though this would a fun 2-3 week project, but with all the failed prototypes and re-designs it ended up taking closer to 3 months. But in the end I am qutie happy with the results and just wanted to share.

I finished the deadbeat clock about six months back and though I would share some of my feedback and observations. First I would like to say that it was very easy to build and keeps amazing accuracy. As has be pointed out in some other post, I need to move the magnet about half way up the flipper for it to work correctly, when placed too low the switch always stays closed. I also need to add some foam to the end of the flipper to hold the magnet a bit farther away from the switch when it is closed, otherwise it just sticks to the switch and does not fall back. (I think the magnets that came with the switches are a bit on the "too strong" side). I noticed that the clock was running through its batteries every three months, far more quickly than expected. But what I found more concerning was that also after three months the weight being put on motor's gear train was enough to cause it's gears to start skipping and the motor was no longer able to rewind the weight (I think the last gear arbor in the motor became slightly skewed to the gear teeth no longer engaged fully). Thinking this was just a one-off bad motor I replaced it (along with the batteries) and the clock ran well for another three months. At that time the exact same thing happened again, the batteries started to run dead and the gears on the motor started skipping. For these first two motors I was using a 20 RPM motor. So after a bit of thinking and tinkering around I decided to make some modifications to the weight. First I replaced the batteries with a USB port so I can just plug it into an outlet. Second I used the now empty space in the weight to add a ratchet system to (I hope) take some of the back pressure off the motor's gears and make the motor last longer this time. I also switched to a 10 RPM motor in case the extra torque helps with the back pressure. (I briefly tried a 120 RPM motor and it was rewinding way to fast and the flipper would not always reset. Also for a low torque motor the gears spin freely in *both* directions when no power is applied and the ratchet becomes needed for the clock to work). Even with all of this said, this is my favorite clock I have printed so far (I have also made the 32 days easy-build, and the electromagnetic clocks).