Graham Escapement Desk Model, Fully Printed

Fri, 06 Nov 2026 00:00:00 +0000

Background/Motivation

I’ve always been fascinated with mechanical clocks, and the escapement in particular is kind of addictive to watch. One tooth of the escape wheel just gets let go by the anchor, and the other side is just in time to catch the next one. You always wonder, will it miss or not? Right? It’s kind of hypnotizing to watch how it works. So I always wanted to make one for myself to look at, basically.

And I wanted it printed completely: no screws, no bearings, no metal and no glue. So this is my take on it, a Graham dead-beat escapement, the kind used in good regulator clocks, driven by a printed PETG mainspring.

This post is the story of how it went from “it doesn’t turn” to “it ticks”. For the files and the step-by-step assembly, see the MakerWorld page.

Model Design/Modeling

The look is a skeleton clock: a black portal frame on a base, so you can see the whole train, with the gold wheels as the focus and the pendulum hanging in front.

On paper everything worked. The first full build told a different story.

The first dry assembly: five problems at once

When I put the first full set together, I found five things wrong:

the click (the ratchet that holds the spring) didn’t hold, the pawl just slipped out
the wheels were stiff
the barrel lid was loose and rattled
the pendulum bob touched the front plate when the clock tipped back a bit
one frame leg lifted during the print, the first layer didn’t stick well

Some were easy. The bob got a window in the front plate behind it, and the frame got a brim. The other ones took a few rounds.

Stiff wheels: the test coupon lied

The wheels were stiff even with the front plate off, so it was the back pin and the bore itself, not the plate alignment. The funny part is that I tested this fit before with a small coupon, and it was fine.

The problem is that the coupon was a thin ring, and the real wheel hub runs much longer over the pin, inside 6 walls. So I printed a new coupon with exact copies of the real hub at three bore sizes. The smallest one was as stiff as the real wheel, and the two bigger ones turned freely. Lesson learned: a fit coupon has to copy the real part, same length and same walls, otherwise it doesn’t tell you much.

It needed a full wind to start, then stopped

Even with free wheels, the train needed a full wind to get going and then stopped. After some reading, the reason was the gear teeth. Normal (involute) gear teeth start pushing before the line between the two centres, and in that part of the mesh the teeth rub a lot. In a car gearbox that doesn’t matter, but a clock train runs on very little force.

Real clocks use a different tooth shape (cycloidal), where the contact happens after the line of centres. So I changed all the teeth.

The pawl that kept sliding out

The click pawl is printed in place and held in its socket by a small bead. It kept sliding out, and it took a while to see why: a groove I cut later in the model removed most of that bead.

So the order matters. The bead has to be added after the cut, not before. After that fix, the pawl stays put and the click holds the spring.

If you ask me which part of the clock I like the most, after the escapement of course, it’s this one: the pawl and the ratchet. I really like that the pawl is printed in place, and then when you wind it, you hear the click. It’s so satisfying, and it’s just so smooth. There’s nothing else to put it together, no pin, no threading, just the pawl printed in place, and it works perfectly. I hope you guys enjoy the clicking sound of the ratchet as well.

Your browser doesn't support HTML5 video. Here is a link to the video instead.

Setting the beat

The pendulum first sat on a friction clamp so you could turn it to set the beat (an even tick-tock). It turned way too easily. The clamp turns along the layer lines, and with a 20 cm pendulum, a light touch is a lot of force. In the end I made it a threaded boss with a thumb nut, plus a small printed spanner to hold it while you tighten. Loosen, set the beat, tighten.

Printing Notes

A few things in the slicer made a real difference on this one:

Slits inside the gear teeth. With the default Arachne wall generator and 6 walls, I saw thin slits between the wall loops along the middle of the teeth. A few Arachne settings changed (minimum bead width, minimum feature size, transition settings) and the slits are gone, for about 6 extra minutes of print time. Everything is set in the 3mf.
Stringing. There were some mild strings across the pendulum hole and between the ratchet teeth. Turning on Avoid crossing walls fixed it. It’s on in every plate now.
Colours. I printed the frame in PLA Basic Black, the wheels and barrel in PLA Silk+ Gold, the anchor and pendulum in PLA Silk+ Titan Gray, and the spring in PETG. The silk wheels worked fine with the fits, which I wasn’t sure about at first.

The two frame plates are about 31 cm long, so it needs an H2-size bed for now.

Model Usage/Resources

Wind it with the key from the back, set the beat, and watch it tick. On my build it runs about 20 minutes on a full wind, and honestly, I can stare at the escape wheel for those whole 20 minutes. It’s a movement to watch, not a timekeeper, there are no hands. It also gets smoother after the first few windings, and a little grease on the pins and pallets really helps.

Your browser doesn't support HTML5 video. Here is a link to the video instead.

All the files, the 3mf with every setting and the step-by-step assembly guide are on MakerWorld.

If you’d like a version that fits a 256 mm bed, leave a comment there or below. If enough people ask, I’ll split the frame.

I hope everyone enjoys it as much as I do.

Seahorse Square Tail

Fri, 30 Oct 2026 00:00:00 +0000

Background/Motivation

When I saw the MakerWorld Biomimicry Design Contest, the seahorse tail came to mind right away. A seahorse tail isn’t round, it’s square, made of small bony plates. Researchers found (Porter et al., Science, 2015) that the square shape grips better and doesn’t twist as easily as a round one. I wanted a model where you can actually see that tail curl.

At first the idea was a clamp, a seahorse that grabs things with its tail. Along the way it turned into a model that shows how the tail curls, and honestly that is the better version of it.

Model Design/Modeling

The tail is 14 square plates, print-in-place, getting smaller toward the tip like the real one. The joints only bend toward the belly, so the tail only curls one way. One cord runs along the belly like the muscle, and the black eye is the winding knob: turn it and the tail rolls up. The fin above the eye works as the release.

The joints were the easy part

I was most worried about 14 print-in-place joints in a row. I printed a few clearance coupons first, and a 0.2 mm gap in the design turned out right on my printer. The first full tail came off the bed with all joints free, and they hold well even if you try to pull a joint out sideways. That part just worked.

The ratchet that never locked

The first eye knob turned, but the teeth and the pawl never locked, in either direction. When I checked the print, the small sharp teeth came out rounded, so the real overlap between tooth and pawl was much smaller than in the design. On top of that the knob sat on a split post that wobbled. Basically, there was nothing left to hold.

The fix was bigger, solid hooked teeth, a matched pawl, and the knob printed in place in two bearings instead of a wobbly post. That one worked well on the next print.

If you ask me what I like most about this one, it’s the print in place again, especially the knob. The whole knob is printed in place, so it comes off the bed and you can just turn it, without any assembly. Yeah, that’s the part I like.

Small things you only find with your hands

The second print worked, but using it showed a few more things:

the fin was in the way when turning the eye with a thumb and finger, so the grip on the knob moved up above the fin
the holes in the tail were tight for the 0.6 mm nylon cord, so they became tall slots
the cord channel at the tail tip had a sharp turn, now it’s a smooth curve, and the exit under the eye gets bigger toward the end so the cord never meets a smaller hole
the knob had a bit of play, so it got more teeth and a tighter bearing

One more came from the slicer preview: the wall on top of the head was too thin over the release lever pocket. That got thicker before the final print.

Printing Notes

It prints in one piece, no supports, a bit under 4 hours, in white and black PLA Basic on my H2C with 0.2 mm layers. The joints and the eye knob come off the bed free, nothing to break loose.

Model Usage/Resources

Wind the eye and it clicks, the tail curls and stays curled. To let it go, turn the eye a tiny bit more, push the eyebrow fin, and let it unwind slowly.

Your browser doesn't support HTML5 video. Here is a link to the video instead.

It’s a model to show how the tail curls, not a strong clamp. The files, the cord stoppers and the threading steps are on MakerWorld.

Hope you find it useful, and if you know an easier way to thread the cord, please leave a comment below.

Print in Place on Plicatim

Graham Escapement Desk Model, Fully Printed

Background/Motivation

Model Design/Modeling

The first dry assembly: five problems at once

Stiff wheels: the test coupon lied

It needed a full wind to start, then stopped

The pawl that kept sliding out

Setting the beat

Printing Notes

Model Usage/Resources

Seahorse Square Tail

Background/Motivation

Model Design/Modeling

The joints were the easy part

The ratchet that never locked

Small things you only find with your hands

Printing Notes

Model Usage/Resources