Gina

Another difference of this 30 legged version is that the pendulum support of well above the escape wheel.  This would suit my clock.

Been doing some testing on my clock.  With the pendulum on its own without the crutch and anchor system attached it will continue swinging for something like 10m before the swing is down to half.  There is far too much friction in the anchor bearings even though these are ball bearings.  With the anchor system attached it halves in about 10s!!   The gravity escapement has a free swinging pendulum over most of its swing rather than being driven all the time.

Been doing a lot of testing trying to find out where this error is coming from and I thing it's poorly defined centre hole.  I'm trying a 5mm hole rather than 3mm to see if this helps.  A disc printed with a 5mm hole measures up as correct!

Failing that I have come across a different form of gravity escapement with 30 "legs" instead of 3 :-
https://mb.nawcc.org/threads/30-legged-gravity-escape-clock.148968/

This uses the gravity escape principle without the extra 10:1 gearing needed for the 3 legged version.  I quite like the idea of a sweep seconds hand of his design but whether I would go for that I don't know.

Yes, I think that could be one problem solved but I have another - the escape wheel.

Measuring the new escape wheel.  This shows a clear mm difference in the radius of the teeth!!!  I'm at a loss at to how this can happen.

That is only one aspect of it!!  How to arrange the parts for a gravity escapement is causing considerable headaches!!

I'm thinking of giving the deadbeat escapement yet another go.  The anchor could be reinforced with a backing plate.  Preferably a different colour so that the pallets show nicely.  I'm also considering different shaped teeth on the escape wheel with a flat top rather than a point. I don't think that would affect how it works but may make for a more accurate print.

Modified escape wheel.  Pointed teeth widened with flat ends.

Test anchor design.  Reinforcing plate built in just to see it the principle will work.

Think I'll do some calculations :-

• escape "wheel" turns in 6s
• stepper motor - 200 steps per rev
• rotation time 200s, 100s, 50s, 25s etc. with various powers of 2
• escape pinion with 12t - gear 50t gives 25s
• or maybe 12:75 giving 37.5s
• timing pulleys of 90t and 60t gives 3:2 ratio which would give 25s at the motor - 8 steps/sec

Now to drive the centre wheel :-

• rotation time of first gear is 37.5s
• rotation time of centre wheel is 3600s giving an overall ratio of 96
• first gear is turning clockwise so a 2 stage spur gear reduction could be used
• or a single stage epicyclic gear set with 95 and 96 teeth.

The Gravity Escapement is beginning to look more attractive!!

I'm beginning to think I am fighting a lost cause in trying to 3D print a pendulum clock!!   Or at least with this sort of escapement.

More trouble - this time a failure of printed plastic.  Latest pic of escapement.  The anchor has suffered from "plastic creep"  and straightened out a bit.  The RH pallet is now catching again!!!  OK I can redesign the anchor with wider arms but not sure that will cure it.

I had another thought about how a pendulum clock could be driven, as opposed to a weight and chain or belt.  The idea is to allow for the intermittent motion produced by the escapement, and that's a spring.  OK been done before but I don't just mean a mainspring wound by hand but with auto-winding.  As with the weight system and auto-winding the spring could be shortened right down.  The escape wheel could have another wheel on the same axis and drive the escape wheel with a spring.  This wheel would be driven by a stepper motor.  From here the rest of the clock could be driven by spur gears where the pinion drives the larger gear and very high ratios are possible or another arrangement like epicyclic gears or even worm drive.

One thing I've noticed while testing is that the pendulum swing decays very rapidly as if there's a lot of friction.  It seems to be worse than it used to be and I can't think why though I did shorten the suspension spring.  Maybe a different sort of suspension would be better.  eg. knife edge or even tiny ball bearing as in the anchor bearings.

Of course it could be the anchor axle bearings or something else associated with anchor of crutch.  Maybe I'll try the pendulum on its own without the crutch etc.

I seem to spend all my time stripping and reassembling my clock, interspersed with thinking, designing and printing new parts.  All the time it's the escapement that's the trouble!!  I guess that's telling me to change the type of escapement...  I dunno...

1. Printed new 25t gear and attached to 90-t timing pulley. 
2. New drive system installed in clock and clock installed in case ready for testing.
3. Have been testing clock and still no joy - still same problem of running and skipping.

Practical considerations showed 50t gear would be too big so going for 25t and stepping at 2 per second.

Drive calculations :-

• If I drive the red gear, it's the same as driving the pinion on the escape wheel (except for the direction). 
• The escape wheel rotates once a minute and the pinion has 10 teeth.
• A 50t gear on the large timing pulley would rotate with a period on 5x60=300s
• Stepper motor takes 200 steps per rev.
• Stepper motor takes 200s at pulse rate of 1Hz
• A timing pulley ratio of 3:1 gives 100s
• 90t and 60t would give 200s for motor rotation and one step per second.  Ideal!

Maybe I could move the drive (white timing pulley and spur gear) from the blue centre wheel to the next in the train (red) but though the gear ratio between these is 8:1 the red gear is 75t rather than 80t so not actually 8x.  This means a re-calculation of the motor speed and ratios needed to obtain a sensible stepping rate.

Redesigned the anchor, moving the RH pallet a couple of mm inwards and printed.

Swapped anchors on my clock and the new one is much better.  The clock actually ran for a minute or so.  I think it just needs a bit more drive force now.  Heavier weight for instance.

Looking closely at the anchor, it still isn't right. 

LH Pallet shows the tooth just coming off the locking side of the pallet. 

RH Pallet.  The locking surface on the left should be half way between teeth and the RHS should clear the teeth.  This pallet is wrong.

Pallet should be here - shown in black.

After fiddling with the anchor position and the clock angle I got it running in fits and starts.  ie. as before tick tock tick tock where the teeth were a bit higher and skipping where lower.

Been testing and no joy!  The anchor seem about right but now it's as if there's too much friction in the gearing!!!  Yet I haven;t changed anything except the escapement.

Put the clock back together ready for testing again.  Meanwhile, I've been looking on the web for deadbeat escapement design and there seem to be many variations.  Don't suppose there's anyone here who knows anything about this is there???

Clock parts printed.  Made up a test rig with a 3mm hole in a piece of wood and a mark at the escape wheel tooth points.  Then turned the wheel to see any errors in the teeth.  Found a difference of just under a mm so got a file and and filed the high ones down to near the low ones.  Reckon the teeth are now within 2 or 3 tenths of a mm.

Designing the deadbeat anchor.  Green circle defines the locking surfaces.  Black is the tooth point circle.

Changing that was easy.  Replaced the 32 with 30 and got this new CAD model.  The anchor will be slightly less easy.