Groups Feed

Great job documenting the pendulum swing data. I usually run my pendulum swing tests all the way down to a negligible amplitude, possibly 0.1 degrees. I really should make it a more scientific test, such as 4 degrees down to 1 degree. Your bearings are great if you are still getting nearly 1 degree after 16 minutes.

As you observed, a free swinging pendulum loses energy quickly if there is any rubbing. The escapement only needs to restore a miniscule amount of energy in a properly tuned clock. If there is excess friction in the gear train, then the clock can quickly stall.

I notice much less blobs and stringing in the gears compared to a few years ago. Several factors likely contribute to this. The "fancy gear" and "perfect print gear" profiles have significantly fewer retractions than a traditional involute profile. Filament has probably gotten better over time. Also, the slicer has evolved. Printing multiple gears at the same time can sometimes be a tiny issue because the head retracts away from the gear teeth, compared to retracting inwards when only a single gear is printed. The slicer may be able to compensate for this.

btw: the metal shields can sometimes be removed by using a sharp pin to pry out the tiny metal retaining clip near the rim. I use a safety pin because the back end makes a good handle. It is tedious, but possible to remove the clips. Sealed bearings are significantly easier to remove the seal by poking the pin into the rubber. Open bearings are harder to find.

That is interesting. I have purchased those screws in the past and don't remember the heads being oversized so much. There is a lot of variability between brands though. An 82 degree countersink is a quick solution. This trick should get added to the assembly guides.

I order screws for the parts kits from McMaster-Carr, product number 90048A151. They are specified to have a head diameter of 0.262" and usually measure close to 0.25". I keep using them because they are extremely consistent across many orders.

The parts kit can be found at https://www.etsy.com/listing/1680180262/sp4bsp5b-easy-build-clock-parts-kit

Thanks for the kind words. A lot of effort goes into optimizing a clock for 3D printing.

The horizontal form factor of this clock stacks gears on a single arbor that can be a bit sensitive, as you found out. Once it is adjusted, the clock should run great.

Regarding pendulum length, many old clocks have a 2 second period with 1 second in each direction. This often used two gear pairs with 60:8 and 64:8 ratios followed by a 30 tooth escapement and a 39" pendulum. I believe that a 39" pendulum is disproportionately long in a small clock.

All of my clocks have shorter pendulums. This requires different gear tooth counts. For example, gear ratios of 36:8, 36:8, 36:8 and a 25 tooth escapement has a 24.5" pendulum beating 4556 times per hour. In the days of hand cut gears, the 39" example has 170 teeth to cut, while the 24.5" pendulum only has 157 total teeth and an additional arbor. With a 3D printer, the total number of teeth is an insignificant factor. I select gear ratios that allow a good overall balance. There is no difference in accuracy of a 39" pendulum compared to a 24.5" pendulum.

That is the first time hearing anything about the click screws rubbing. Flat head screws should sit nearly flush and have enough clearance. You have a solution that also works.

The pendulum support bearings are easy to check by running the free swing test. Anything greater than 10 minutes should be OK.

The escapement characteristics after the clock stops can tell you a lot about what to debug next. Does the escapement still spin? If not, then look for defects on the gear teeth where they are meshing. Also check for end shake on all the arbors. If the escapement is sluggish, then look for friction. Make sure every gear spins freely on its arbor.

You need to hold the clock level when running without the pulley. It is a simple way to double the effective weight, but the balance is off so the frame wants to tilt and the clock will be out of beat unless the clock is held vertical.

Your clock should be very close to running perfectly.

I use TurboCAD Platinum 2018 with the faceting quality set to the highest level.

There are probably cheaper alternatives available today that would do an equivalent or better job.

Looks great.

The moon phase operation is only two gears and a friction clutch. Make sure that both gears can move freely. Touch the gears and they should move slightly. Test the friction clutch by manually rotating the moon dial by reaching behind the main dial. There should be a small amount of resistance.

Another possibility is that you need to wait longer to see the moon phase rotate. The moon dial makes one complete rotation every 59 days. This gives a full moon cycle of 29.5 days.

The video is very helpful for debugging. It shows friction in the gear train but cannot show where it is. That is something you will have to debug.

A good start is to test that every gear spins absolutely freely on its arbor and inside the frame. Test them individually, then work up to multiple gears running together in the frame.

This clock has several gears that need to be tested inside the frame. The central arbor has the most gears stacked on a single location. If the stack is too tall, then the frame can pinch on them. Gear 9 may also pass through the front frame. Another critical location is the gear behind the escapement. It is locked onto the arbor with a set screw and the arbor needs to be able to rotate in the frame or else gear 2 will not rotate easily.