TBH I'm not sure.  It's proving pretty difficult to work out how the gravity escapement is going to fit in.  Also, as you say there's a lot to re-design whereas I already have a lot of the parts made for a standard escapement.  I might be able to get the deadbeat escapement working in appearance if I drive the pendulum separately.  Rather like a slave clock.

Maybe I should have another go at the deadbeat escapement.

Decided to have another go at the deadbeat escapement and see how it goes.

I've printed the deadbeat anchor and the escape wheel and now reassembling the clock, having stripped it down to try the gravity escapement.

The anchor didn't fit so redesigned it.

On 18/01/2021 at 14:58, Gina said:

I've worked out that I should be able to get away without an extra gear pair but with higher ratios - though I have to test that out. 

6:1 from chain wheel to centre wheel with the chain wheel going faster (can do this because the auto-winding can keep up with the faster chain drive).  Then two stages of 10:1 give the overall ratio from centre wheel (60m) to escape wheel (6s) of 600:1.  Or do you think I'm being too ambitious?

The critical thing is how small (tooth count) a pinion can you use? You might consider using 1 mod gears?

ATM I'm trying the deadbeat escapement as that need far less work.

On 20/01/2021 at 14:59, Stub Mandrel said:

The critical thing is how small (tooth count) a pinion can you use? You might consider using 1 mod gears?

Considering I'm printing GT2 timing pulleys very nicely I see no problem with mod 1 spur gears.

Decided on timing belts instead of clock chain and sprockets.

On 22/01/2021 at 12:09, Gina said:

Considering I'm printing GT2 timing pulleys very nicely I see no problem with mod 1 spur gears.

That would let you double the pinion tooth counts and get a better form.

On 23/01/2021 at 23:31, Stub Mandrel said:

That would let you double the pinion tooth counts and get a better form.

Do you think that would produce better gears with lower friction?  I'm just wondering if printing new gears with lower modulus would make my clock work better.

This is a photo of the clock "going train".  The white wheel with the timing belt hanging over it is a 90t timing pulley (3D printed).  Under that and attached to it is a 40t spur gear.  Centre wheel is 80t.  Next is 10t pinion on a 75t spur gear which drives a 10t pinion on the escape wheel.  All spur gears are mod 2.25.  The going train has ratios of 8:1 and 7.5:1 (8x7.5 = 60).

Now to the auto-winding system.

With the current system the timing pulley rotates 4x an hour (15m) and I plan a 20t timing pulley on a stepper motor. 

Some calculations :-
    Timing pulley ("great wheel" in clock parlance) takes 15m per revolution = 15x60s = 900s
    Timing belt and pulleys have a ratio of 90:20 so motor pulley takes 900x20/90 seconds per rev = 200.
    NEMA steppers have a stride angle of 1.8° degrees or 200 full steps per 360° (revolution)
    Thus stepping the motor once per second gives the required rate for auto-winding.
    Q.E.D.

An alternative to printing new gears might be to drive the intermediate wheel rather than the centre wheel giving 8x more "leverage" to the drive.  To cater for this change the stepper motor for auto-winding could be driven at 8 steps per second.

I'm going to have to reprint the escape wheel, as in spite of extreme care, the axle is not in the centre of the points circle!!  Or indeed of the pinion.  How this has happened I don't know!!!

On 25/01/2021 at 11:43, Gina said:

Do you think that would produce better gears with lower friction?  I'm just wondering if printing new gears with lower modulus would make my clock work better.

It's highly controversial but as a general principle involute gears of any size should be better than traditional clock gears. Actual size really chiefly affects power transfer capacity, friction should not be a big issue with well-formed involute gears  - although you could test this.

I agree, involute gears are designed for lowest friction I gather and is what I'm using as I see no need to stick with tradition in my clocks.

Just printed a new escape wheel with fresh filament to see if it is any better.  Presently cooling.

It's printed well and seems fine - vast improvement!  Didn't realise stale filament could be so troublesome!!!

Have the clock assembled including the double weight drive system and stepper motor.  The drive is working and the clock partly runs but there is still the problem that the escape wheel is not exactly even in the position of the teeth relative to the anchor.  I can adjust the anchor height for optimum performance but when the clock runs over part of the escape wheel it skips teeth on the opposite side.  The problem is that the tolerance is just a few tenths of a mm.  OTOH I wonder if the shape of the pallets is ideal. 

Here are photos of the test rig and a closeup of the escapement.

Analysing that video it seems the RH pallet on the anchor is not working correctly.  I need to redesign the anchor.  Maybe there's too much "drive" angle.

This is a CAD screenshot.  Think I'll try reducing the "drive" projection into the tooth circle.

New anchor design with reduced drive part angle.

New escapement anchor finished printing with good result and installed in clock.  I'll leave testing until tomorrow but it looks promising.

Set up the testing for the clock but the new anchor is wrong.  I don't understand why though - the pallet separation (width of anchor) seems wrong in practice and different from the CAD design.