Last modified: 2024-11-11 23:59:06
< 2024-11-10 2024-11-12 >I printed the second bending die in PC.
It did work, but:
So I will do a second iteration of the middle part. I think basically the same as the first die for the counterweight, just push down from the top. And I'll put an undercut on the male part.
But first, something to bend up the stand. Also we might want to make the score line in the stand wider/deeper.
This was not a very good success.
The stand got stuck inside the female part and I had to drill a hole in the bottom to punch it out.
And the fold lines weren't very well aligned with the outside shape, so one edge is taller than the other, which makes the scales not sit level.
My current thinking is that it'll be a weight-driven clock with a weight-driven remontoire as close to the escape wheel as possible. It will have a verge-style escapement, but with rounded backs to the pallets like on H4. As much mechanism exposed as possible, perhaps the face is off to one side so as not to obscure the works, powered by a chain or belt, with some sort of friction clutch near the hands to allow setting the time by just moving the hands. We could have anti-friction rollers on the escape wheel. And the balance wheel either in conical pivots or suspended by string.
The remontoire could be of the continuous-chain type? I don't really like the idea of using chain, but I suppose with any weight-driven remontoire you're going to have a chain?
The other idea is a lever that only turns through about 45 degrees, sits about horizontally, and bears against some spokes on the escape wheel with a ratchet. And periodically it gets pushed back up, but then there's no maintaining power while it's rewinding, the continuous-chain thing solves that easily.
With the lever method, you could:
This mechanism has a weight-on-a-lever method: https://www.youtube.com/watch?v=5yIRSKr28JY - pretty elegant.
Apparently this is a "Korfhage remontoire" - http://www.my-time-machines.net/korfhage_remontore.htm
So you could put the bevel gears directly on the back of the escape wheel, and still have the escape wheel ride on anti-friction rollers. And the rest of the train would have relatively ordinary pivots.
As long as the remontoire weight is enough to drive the escapement through the bevel gears, and the main drive weight is enough to wind up the remontoire, everything works.
It winds up every time the lever falls far enough to disengage from the arm that stops the driving train, and keeps winding until that arm has done 1 full revolution and stopped itself again. So you need to make sure they are geared that the lever is going to move about 45 deg. when that shaft does 1 revolution.
So I'm keen for the clock to have:
And if we say we're going to tolerate 3d-printed parts where appropriate, it becomes a lot less work.
I did some calculations in 2024-04-29.
I reckoned a ratio of 1:96 from the drum to the minute hand, and then if the escape wheel goes around once in 4 mnutes, then a further 1:15 from the minute hand to the escape wheel.
If we want the lever to fall by 45-90 degrees between rewindings, then we need the shaft with the stopper arm on it to be a factor of 4x to 8x away from the shaft that does the winding. Actually we maybe don't care so much about that, because there can be multiple arms coming off that shaft, and I think we in fact want the gearing to be going in the opposite direction?
We can definitely make gears down to module 1, with either the CNC machine with a 1mm cutter, or proper gear cutters, or on the 3d printer. Smaller is probably possible but maybe not appropriate.
The largest module 1 gear that will fit on the CNC table is about 195 teeth. And the largest on the 3d printer is about 250 teeth.
We can get 1:96 from 10:100 and then 10:96, which gives us 1 rotation per hour.
Then take the minute hand off there. Then a 1:5 and 1:3 to get down to one rotation every 4 minutes? Or 1:7.5 and 1:2.
You could get 1:2 in the Korfhage remontoire by making the bevel gears different sizes.
So let's say 10:75 after the minute hand shaft, and then some 1:2 ratio on the remontoire.
If the Korfhage remontoire is in the middle of a shaft then the arm that stops/starts the winding is going to be on the shaft that turns once per hour. And if we want it to rewind every 45 degrees, then that is an eighth of an escape wheel turn, or 30 seconds. So we'd need the once-per-hour shaft to have 120 arms that catch the remontoire??
That's not going to work (I mean, it might, if we said the "arms" were actually just a gear with 120 teeth, but it's not really what I'm after).
OK, next idea is we forget about having one of the wheels turning once per hour. If the hands are driven off a belt or chain then we introduce the required extra gearing through that.
What then? We just need 1:1440 from the drive shaft up to the escape wheel. And if we put 1:2 in the remontoire gearing, and we want the remontoire rewound every eighth of an escape wheel turn, and we want let's say 8 arms on the shaft adjacent to the remontoire shaft. Then for catching 1/8 of the adjacent shaft to cause 1/4 of a turn on the remontoire thing then we only want 1:2 from the adjacent shaft to the remontoire shaft?
Maybe let's say we'll rewind the remontoire every 1/4 of a turn, so it moves through 90 deg., then we can allow 1:4.
So escape wheel goes around once in 4 mins. Remontoire drive goes around once in 8 mins. Adjacent shaft goes around once in 32 mins.
OK, update: I don't think you can have asymmetric gears on the remontoire because then the gear in between them won't mesh properly with them both.
So the remontoire shaft goes around at the same speed as the escape wheel (once in 4 mins), and we want to rewind it once per minute, which means we need to catch an arm of the adjacent shaft once per minute, which means if the adjacent shaft has 8 arms then it can drive the remontoire with 1:2 ratio.
So that leaves 1:720 from the drive shaft to the adjacent shaft. Actually, let's have more than that. This ratio is if the drive shaft goes around once every 4 days, which means ~100 times per year, and to do that it needs to be 6.4mm diameter and 30cm long. But hanging a heavy weight off the middle of such a long and thin bar seems like it would make it bend.
If we made it half as long then I think we get to double the diameter for free. A 150mm long 13mm diameter rod seems much more believable, and gives us the same 2 metre drop, but turns half as fast, so we need 2x more gear ratio. So we still need 1:1440.
You get there in 3 steps of about 1:11.5.
OK so for gearing we have:
And for bearings:
I decided against suspending the balance wheel from string because it will need bearings to constrain it radially anyway, may as well take the load with them, and then we avoid any issues with the string twisting up or lifting the wheel as it winds up.
Oops, just worked out that if the adjacent shaft is spinning slower than the remontoire then it will have more torque, which means that by pressing on the lever it will have some influence on the lever. You really want the adjacent shaft spinning very fast, with very little torque, so that it doesn't influence the lever very much. So let's lose the requirement that the adjacent shaft be part of the going train, add the adjacent shaft as a separate part, and calculate a new train.
And actually we will probably need to find a new train once the escapement is built, it might not actually be suitable for 4 mins/rev at the escape wheel.
I recall previously I have written ad hoc perl scripts for calculating gear trains. This time I got Cursor to write me a go program.
And actually this might be better as a web page...
Cursor is so good. It got the Go program almost right on the first try (I just had to prompt it to use rationals instead of floats, to allow finding perfect ratios), and also turned the Go program into a working web page on the first try at that.
https://incoherency.co.uk/clock-gear-train-calculator/
< 2024-11-10 2024-11-12 >