Clockwork Factory by Eric Whitman

Clockwork Factory is a planned clockwork automaton featuring three figures (factory workers) working in a factory. Presented here is the CAD model generated in SolidWorks 2010 and rendered using PhotoView 360. The CAD files are available here. (It takes a pretty powerful computer to do much with the full assembly.) Please note that for the people, only the moving parts (one or both arms) are shown here. The rest is intended to be sculpted from clay, and I have not attempted to depict that here.

General Description

It's built on a scale of 1"=1' and has overall dimensions of 21.5" long by 16" wide (including the crank) by 11.5" tall. It's mostly aluminum with some parts made from steel, brass, or bronze. The wooden dowels are made of mahogany, but I haven't yet settled on a species of hardwood for the floor. The people (not shown) will be painted/enameled ceramic.

A cycle (one turn of the crank) is intended to be about 4 seconds. It is intended to eventually replace the crank with a motor, but I have no idea how much torque will be required, so selecting a motor and designing a gear box will have to be Phase II.

The main belt moves forward incrementally. It moves forward one section in 1/4 cycle, then remains still for the remaining 3/4 cycle. As it travels, the belt passes by three stations. At the first station, Worker number 1 uses a press to make a hole. At the second station, Worker number 2 puts an item (a rivet?) into the hole. At the third station, Worker number 3 hammers the item down.

How it works

Pictures are below their associated text. Some parts are hidden (notably the floor) for visibility. Gears are rendered without teeth, some of the materials are wrong, and I have no idea how to fix the lighting.

A geneva drive drives the belt forward in discrete increments.

At the first station, a scotch yoke drives the press up and down.

A rack and pinion turns the press crank, which is linked to Worker number one's arm (the only part of Worker number one that won't be ceramic).

At the same time, a bar rotates the paddle wheel 1/4 turn, causing the cartridge in the middle of the belt segment to turn over, showing a hole rather than a flat surface.

At the second station, a cam is used to rotate Worker number 2 90 degrees back and forth around a vertical axis.

Her right arm (again the only part actually shown) therefore moves back and forth from a crate of "rivets" (not shown) to the belt. A pin is pushed back and forth so that a rivet is shown in her hand when moving towards the belt, but hidden when moving away. At the same time, the cartridge is flipped over again, showing a rivet rather than an empty hole.

Worker number three has the distinction of having two metal arms. A 4-bar-like mechanism gives her a human-like hammering motion, which is controlled and timed by a cam. As she hammers, the cartridge is flipped over yet again, to show a more rounded version of the rivet.

On the bottom of the belt, the cartridge is flipped over a 4th time, returning it to the flat side.


I've finally gotten my basement shop upgraded enough to begin this project. I should be able to do most of the machining in my basement. A friend has graciously offered to help me CNC mill some of the more complicated parts on her lab's mill. Here's a picture of the wall containing most of my hand tools.

This is my workbench with my benchtop tools, a grinder, sander, vice arbor press, and the centerpiece of my shop, a mini-mill.

Here's a closer picture of the mini-mill. So far, I like it a lot (for the money). It's not as rigid and has a smaller workspace (about 9" x 4" x 8") than a full-sized mill, but it can still get decent precision (a few thousandth's of an inch) and a good surface finish if you're careful. I'm working on adding digital read-outs for all three axes so I don't have to count turns or worry about backlash quite so much. I've so far used a digital caliper to add a readout for the y-axis. This little upgrade cost me about $6 for a cheap 6" digital caliper and another $7 for a tap and drill bit I broke while attaching it to the mill. Unfortunately, the other 2 axes will require a more expensive measurement tool because they travel farther than 6".


As Of 9/4/2011

I've gotten all the stock in from,, and I priced it all out at $810.20 a few months ago ( A detailed bill of materials is given here.), but prices have gone up 5-10 percent since then, and I got a bit extra to give myself more flexibility in case some modifications are necessary.

Here's a picture of the first part, complete with pressed in bushings.

As Of 1/16/2012

I've gotten a good start on the construction. All of the major structure is done. The middle mechanism (the placer) and most of the final mechanism (the hammerer) are complete except for the parts that need to be CNC'd.
Here are two old pictures from a few months ago that I never posted showing the front and back.

Here are two pictures from today showing my current progress.

I have also made (and partially made) a number of other parts that have not yet been attached.
A friend of mine recently suggested we make a stirling engine. I think that would be the perfect power plant for this project, so I am scrapping my original plan of eventually replacing the hand crank with an electric motor. Instead, I hope to create a suitable stirling engine. I want it to run at 15 RPM. I don't have any good way of estimating how much torque I need, so I'm guessing about 10 Nm, which I hope is conservative. That means I'll need about 15 watts of power output from the engine, which should be doable with some work.
The whole stirling engine idea has gotten me motivated to get back to work on this project if only so that I can figure out the torque requirement for the engine. Unfortunately, I'm still held up on the parts that need to be CNC'd and probably will be for at least another month.

As Of 2/12/2012

I finally finished the arm linkages for the hammering person. I started making these pieces ages ago, but ended up working on other stuff instead. Unfortunately, I'm still waiting for the cams and the cam-following driver linkage to get CNC'd before I can power the arms.

After taking this picture, I loosened the black retaining ring a bit so that the arms hang down loosely. Everything moves pretty smoothly, so I doubt this mechanism will use up much of my torque budget.

Comments to

Last updated on 2/12/2011.