SP7 inspired me to return to a clock I made a while back, based on William Strutt's epicyclic gear clock. The original (from the 1820s) was spring driven. In my version I used an electromagnetic pendulum. It was driven by an Arduino and I was never very happy with it, as it couldn't run off batteries for very long. I used the same circuit as the SP7, and also modified my original design to reduce friction and increase stability. Here is some video. I still regard this as a prototype and I cut a few corners to make for faster printing. When I have more time, I intend to finish it off properly and publish the design. Hope you like it! https://youtu.be/zYvOex5JYME

Here is my build of SP7: https://www.youtube.com/watch?v=pgtvyV1DrsI. Another excellent design. A few notes: I am using the active pawl from Hoppy's mods and the original static pawl. I found Hoppy's pawl more reliable, but his static pawl put too much weight on the ratchet wheel and cause it to miss some ticks. I still have work to do on the static pawl selection. I made the pendulum shims thicker. I think the issue here is that I printed the base and frame with a 0.6mm nozzle and 0.4mm layers to reduce print time, and consequently the feet sit fractionally higher. The pendulum mechanism had the number 5668 on it, ordered from ebay. I believe it is the same circuit with SMD transistors. I also tried one labelled 12888. It did not work well, and it seems to have less power. The circuit is quite different: it uses only a single transistor. The battery cover won't sit quite flush with the 5668, so I added some stick-on feet to make the base sit better. The minutes hand was too loose and so I tweaked the STL. Not yet tuned! I take inspiration from a video by the curator of clocks of the British Museum, who talked about clocks that you bring out to show off on special occasions rather than using them as regular timepieces.

I've just posted my SP8 build at https://youtu.be/K6esLtZtmhQ. I can't really call it "the coup perdu" clock as I went with the deadbeat design. This was an easy one to get going. Apologies for the poor quality video and pictures - my phone wouldn't focus very well. A few notes: - the arbors were tight in the frame until I drilled them out with a 1.7mm drill. It wouldn't work reliably until I did this. - the motor has the opposite connections to the one described in the instructions. It might be useful to add a note saying that it should rotate clockwise when the shaft is viewed end on, so you can check this before soldering it all up. - I used a plastic encapsulated reed switch. This is a little larger than the glass ones: 3mm square cross section, instead of 2mm circular. I modified the switch cover to allow for this. Without doing this, the switch stood proud of the surface it is mounted on, and the magnet could never get far enough away to turn it off. Also I used a 6mm magnet, as I had some of these already, and opened up the hole in the flipper to fit it. - it might be nice to have levelling screws in the base. It doesn't make much difference to the time keeping, but it is easier to adjust using the Cuckoo Calibration App when it is level.

Hi, I am prototyping a clock of my own, and making many (hopefully instructive) mistakes along the way. I wanted to ask about something you mention briefly in the PDF that accompanies SP1. You say that the arbor spacing was increased for the gears to mesh properly (top of page 8). Do you have a rule of thumb for doing this? I am using the standard approach that for two gears with X and Y teeth and module m, they should be (X+Y)m/2 apart, or something similar if you use DP. In some places I can see that this makes the gears too tight, whether due to printer tolerance or because the load on the gears puts a bit of sideways pressure on them. I'm experimenting with this, and if you have any guidance it would help the process along.

I was wondering if you had looked at the Arachne slicing algorithm which is in Cura 5 and the alpha version of Prusa Slicer 2.5.0. It works much better on gear teeth than the old slicing algorithm. You get a lot fewer of the little tiny lines in the middle of teeth which you worked round by using fancy gears.

This is my latest clock: https://www.youtube.com/watch?v=dR3y-6C9Yx8. Based on William Strutt's epicyclic gear clock from the 1830s, with a guest appearance from Ferguson's Mechanical Paradox.

I thought you might like to hear more about my experiences with the stopping problem. I built the 10 day clock in August, and had it run reliably for around 5 weeks, after which it began occasionally stopping. Eventually it got to the point where it was stopping twice or more per day. We discussed this a bit at myminifactory and on discord, and I decided to reprint everything using regular hatchbox PLA instead of silk PLA. When I rebuilt the clock, I also re-cleaned the bearings and lubricated them with dry PTFE, and picked the best ones by spinning them on a 3mm shaft and counting how long it was until they stopped. After reassembling the clock it ran with less weight than before (down from about 7.5 pounds to 5.5 pounds). The pendulum swing was +/- 3 degrees most of the time, occasionally a little less. It ran reliably for 14 days. This morning I noticed the swing was only +/- 2 degrees. From when I set the beat originally, I know this is about the minimum that will work. This evening, the clock stopped. I noticed that if I restarted it with the pendulum swinging ~5 degrees, it decayed back down to 2 within a minute or so. I've just tried removing the pendulum+anchor and its arbor and reseating the bearings. Spinning them on a 3mm shaft as before showed they still run for as long, so I don't think they need further cleaning or lubricating. I also noticed the dial had a little (tiny amount) play due to some unevenness in the top of the base, so I sanded the top of the frame to smooth this out. Too early to tell if I've fixed the problem, but after an hour the pendulum is still showing a healthy 3 degree swing without quickly decaying to 2 degrees. Not sure what to make of all this. It eliminates silk PLA as being the problem for my clock. I've noticed that bearings are a loose fit in their hole (likewise the arbors), and this is something which can vary from one printer to another. We'll see what happens over the next few days (weeks? hours?). I did rewind the clock just 2 days ago, and it makes me wonder if this might shake the bearings out of alignment causing extra friction. It would only take a little extra to cause problems. The first half of this video (https://www.youtube.com/watch?v=FKpRrlZuuZo&t=88s&ab_channel=woodentimes.com) has some observations about the effect of misaligned bearings. If it stops again, I think I might try putting a little shim round the bearings to hold them in place better. I've done this on other projects by sticking a tiny piece of tape round the rim. Maybe polishing the arbor more would also help. I'm interested in any thoughts about all this.

What do you think is the smallest workable size for printed gears? I am starting work on a design and due to various constraints, it would require a pinion with 8 teeth and a module of 1.2, giving the gear a pitch diameter of 9.6mm. I know I can accurately print it that small, but would it be too fragile or have other problems? I read something which suggested 9 teeth was about the limit (at a 25 degree pressure angle)