This is a prototype of an escapement that I want to put in a watch movement.
It is equivalent to an anchor escapement, except the pallets are made of pins instead of having special shapes. It is also equivalent to a verge escapement, except the verge is in the same axis as the balance wheel instead of perpendicular.
I haven't seen an escapement like this before, but I am sure it has been tried as it is so obvious. Probably it is not used because the amplitude (rotation of the balance wheel) is so low, here only about 8.6 degrees at the start and 8.2 degrees at the end. A typical modern watch movement has amplitude around 300 degrees!
Usually a watch uses a "Swiss lever" escapement, in which the balance wheel is left free to spin on its own for almost the entire cycle, only interacting with the escape wheel through the pallet fork, which gets locked to the banking by the "draw" of the escape wheel while it's not interacting with the balance wheel.
The advantage of the Swiss lever escapement is that the balance wheel can swing at its natural rate, and isn't impeded so much by how much torque is being applied to it. The disadvantage is that the pallet fork is a tiny part that you have to make very precisely. I'd rather have a less accurate escapement that is easier to make.
...
https://www.youtube.com/watch?v=RApdxL61nmA
Prototype of a friction clutch for a clock. The idea is that the power train is coupled to the motion works through this clutch. In normal operation it grips and the hands are turned by the clock, but when you go to change the time, the motion works can slip on the clutch so that you can move it without having to drive the whole against the escapement.
...
https://www.youtube.com/watch?v=McyNvgNNA2A
Not sure what is wrong. I left it busy doing disk io and mandelbrot rendering for about 6 hours today, then it crashed and now it is totally unstable. What's changed?
...
https://www.youtube.com/watch?v=jp-ag3TcwiQ
I made a bit of a blunder on part 1. My programs ran much more slowly than I expected, the input was only 2000 lines so I really don't know what the problem was. Maybe it's just that parsing 2000 integers literally takes that long? Before tomorrow I'm going to profile it and find out - there may be some good low-hanging optimisation fruit that will help with later days.
...
https://www.youtube.com/watch?v=ZLF8bWWw9FA
Testing my laser level with only 1 axis free to move.
It uses 2x laser line modules in a 3d-printed housing, with tiny bearings allowing it to pivot freely. The laser line module angles are adjusted with grub screws in the housing, and the centre of mass is adjusted (in both axes) by a pair of larger & heavier grub screws.
I probably want to soak the bearings in some solvent to get the oil out of them as they currently stick ever so slightly. Looks promising though.
Still to do:
- second axis
- battery power
- convenient mount with magnets and a tripod adapter
...
https://www.youtube.com/watch?v=8tzOYczSHx8