Mystery Clock




Introduction: Mystery Clock

The Mystery Clock is so named because even though the entire clock movement is visible, what keeps it going can be somewhat of a mystery to many. The secret is a magnet embedded in the bottom of the pendulum, and an electrical coil hidden in the base. But you don’t have to tell your guests that – make then guess! I've actually had people believe that this is a perpetual motion machine!

Wood gear clocks can be finicky to get and keep running accurately, but this one is quite robust. A microcontroller drive circuit makes it very accurate.

All of the wood parts can be made with a CNC machine. I use a Carvewright machine, and used Carvewright Designer software to design the clock. I have since replicated the design using LibreCAD and InkScape to produce DXF files. There are some small differences in the replicated design, so if you use the DXF files you will see some differences from the photos in this Instructable.

There are also a few STL files for carved parts.

The Instructable covers not just the build but also the design. The intent is that you don't simply replicate my design, but perhaps also design your own clock.


  • 1/2" and 3/4" thick hardwoods - I used oak, maple, hickory, and pine
  • 1/4" Baltic Birch plywood
  • Hardware and electronics (see list)
  • Brass tubing (see list)

Step 1: Hardware

In addition to the wood parts that you CNC machine, you'll need some hardware. The Mystery Clock uses brass tubing for arbors, shafts, and bearings. It also uses ball bearings, screws, threaded rod, fasteners, and electronics. Hardware and tubing lists are provided.

A complete hardware kit including tubing, fasteners, magnet, coil, and electronics is available at

Step 2: Basic Design

This clock uses an electromagnetic pendulum. In most pendulum clocks the pendulum regulates the timekeeping, but is driven by other means such as a spring or weights. In an electromagnetic pendulum clock the pendulum both regulates and drives the movement. A magnet in the pendulum swings past a hidden coil. This generates a current in the coil, which is detected by a circuit. The circuit then passes current into the coil, making it an electromagnet to repel the pendulum. This little push keeps the pendulum moving.

Due to variations in temperature and humidity, the period of a pendulum can also vary slightly, resulting in inaccurate timekeeping. I wanted an accurate clock. Is there a way to compensate for environmental variation?

The period of a pendulum is primarily determined by its effective length, but the swing angle also plays a role. The larger the swing angle, the longer the period of the pendulum. If we vary the swing angle from about 15 degrees to about 25 degrees, we can adjust the period about 1/2%. The Mystery clock is designed to allow the swing angle to be variable and thus is able to compensate for environmental variation. The electronics adjusts the swing angle to keep timekeeping accurate.

For this clock I used a pendulum period of one second. That means that the effective length of the pendulum is 9.78". The actual length is about 12", but the effective length is determined by the location of the round bob, which can be adjusted up and down on the pendulum. We do need to adjust it to get the period in the ballpark of 1 second; after that, the electronics vary the swing angle to keep it accurate.

Step 3: Electronics and Software

The electronics and software for this clock is the same as is used in one of my other clocks, but just packaged a bit differently. The software and electronics are described in detail here:, so I am not repeating it in this Instructable. If you don't' want to build and program your own electronics, I have it available at

Step 4: Ratchet Design

To convert the back-and-forth movement of the pendulum to rotary movement, we're using a ratchet and pawl mechanism. A moving pawl operated by the pendulum pulls the ratchet wheel by a tooth, and a locking pawl prevents the ratchet wheel from moving backwards.

I chose a 60-tooth ratchet wheel for this clock, so with a 1-second period pendulum the ratchet wheel will turn at exactly one RPM.

The pawl must move exactly one tooth per pendulum swing. But, since we are varying the swing angle of the pendulum, how can we guarantee this?

Step 5: Cam Design

To convert the varying back-and-forth movement of the pendulum to a consistent back-and forth movement of the moving pawl, we're using a cam at the pendulum pivot. The cam has a profile with two different radii smoothly connected. An arm with a wheel rests against the cam, and the weight of the arm keeps the wheel in contact with the cam. As the cam moves, the arm swings back and forth. Notice that when the pendulum swings left, the arm (with pawl attached) moves right. But because of the shape of the cam, when the pendulum moves far left the arm moves no farther. This ensures that our pawl will move exactly one tooth; no more, no less.

Step 6: Gear Design

With our ratchet wheel moving at exactly 1 RPM, to get our minute hand to move at 1 revolution per hour we obviously need to reduce the rotation by a factor of 60. For this clock I used two sets of gears, one with a pinion of 12 teeth and a wheel of 72 teeth (72/12 = 6) and a pinion of 8 teeth and a wheel of 80 teeth (80/8 = 10) to get the 60:1 reduction.

The hour hand requires a reduction of 12:1. I used a pinion of 8 teeth and a wheel of 32 teeth (32/8 = 4) and a pinion of 10 teeth and wheel of 30 teeth (30/10 = 3) to get the 12:1 reduction.

There are several gear designer applications available, but I used the Radial Vector Generator program developed by one of the Carvewright users and can be found on the Carvewright User's Forum. The program can create files in the proprietary Carvewright file format, but also can output in SVG format. I did that, then used Inkscape to convert SVG to DXF. Then I used LibreCAD to properly scale the DXF files. You can use other software to generate the proper gears.

The Radial Vector Generator was also used to design the ratchet wheel.

Step 7: Frame Design

This clock uses a pretty simple frame to hold the gear sets and numerals ring. The wheels and pinions of course need to be staggered, so this needs to be taken into account with the frame design. I have the pendulum up front, with the ratchet at the front of the frame behind the pendulum. The minute hand gear sets are behind the frame, and the hour hand gear sets up front.

You can get quite creative in arranging your own gear sets.

Step 8: Clutch

To permit movement of the hands to set the clock, a clutch is needed. I designed a circular one with ratcheting teeth to do this. The driving clutch is attached to the 80T wheel, and the driven clutch is attached to the minute hand. The clutch has 60 teeth. It is spring loaded to keep the teeth in contact when the clock is operating. To set the time, grasp and pull the clutch back to disengage it from the movement, then rotate the clutch to move the minute and hour hands.

Step 9: Machine the Parts

As stated earlier, I designed all of the wood parts with Carvewright Designer software. I have since redesigned to the same specifications using LibreCAD to generate DXF files. I have imported these into Carvewright Designer to compare with the originals, but I have not actually used the DXF files to build a clock. If you find problems with the files, let me know and I will see what I can do.

My CNC uses only a few sizes of cutting bits and no drill bits. Holes larger than the bits are cut by moving the bit in a circular fashion. These holes are not precise enough for this clock. Many parts such as pinions and wheels must be press fit tightly onto brass tubing. Precise drill bits are needed for this. Consequently, I used my CNC to drill 1/16" pilot holes for most holes, then used a drill press to drill the exact size hole. The instructions detail this for all of the holes that need to be drilled in this fashion.

I have also converted the carvings for the hands and clutch surface into STL files. You'll need to lay the clutch pattern onto the clutch wheels.

I used a 1/8" cutting bit to cut all of the parts, except the frame parts were cut with a 3/8" cutting bit. The DXF files define the outside edge of the parts.

Machine the wheels out of 1/4" Baltic Birch plywood. Baltic Birch has more plies than typical plywood - 1/4" has 5 plies. Also there are no voids in the plies. So it is very stable and suitable for clock wheels. 1/2" and 3/4" hardwoods are used for the other parts. You can use various species as suits your taste.

Added 9/26/2022

Some CNC students built this clock using Vectric VCarve software. They kindly provided the files that they created:

Step 10: Numerals Ring

Start assembly with the numerals ring. You'll need to lay out numerals of your choice on the ring segments - I used V-carved block numerals. I used 4 segments, fastened together with mini biscuits and glue.

Step 11: Hands

Drill a ¼” hole through the hour hand. Drill a 7/32” hole 3/8” deep from the back of the minute hand, but don’t drill all the way through.

Step 12: 72T Straight Arm Gear Assembly

Drill a 7/32” hole in the center of the 72-tooth straight arm gear, using the pilot hole as a guide. Drill a 7/32” hole in the center of the smaller 8-tooth pinions.

Insert a 7/32” by 1” long tube into the pinion, with 1/8” of the tube protruding. You may need a hammer to drive it home. If so, protect the end of the tube with scrap wood.

Drill a 3/16” hole 3/8” deep, but not through, the ¾” diameter 72T gear cap.

Apply glue to the surface of the pinion. Align the tube and the hole in the gear and press them together. The tube should protrude from the back of the 72-tooth gear as well as from the pinion. Clamp the pinion to the gear and wipe off any excess glue. Allow the glue to set.

Step 13: 80T 5-arm Spiral Gear Assembly

Using the pilot hole made by your machine as a guide, drill a ¼” hole in the center of the 80-tooth 5-arm spiral gear. Do the same in the center of the 80T (1” diameter x ¾” thick) hub.

Insert a ¼” brass tube 1¼” long into the hub, and into the 80T gear, gluing the hub to the gear as you did for the spiral gear. (Note the direction of the spiral arms.) When finished, the tube should protrude 1/16” from the 80T gear.

Next, drill a ¼” hole through the center of the driving clutch. Press this onto the brass tube arbor protruding from the hub of the 80T assembly, and glue it to the hub. The clutch teeth face outward. Check that all these parts are square.

Step 14: Hour Hand Assembly

Drill a ¼” hole in the center of the 32-t0oth 3-arm star gear. Do the same in the center of the small spacer, and through the hour hand. Insert a ¼” diameter brass tube 1¼” long into the gear, and then glue the spacer to the gear. Likewise, glue the hour hand to the spacer. Align the hour hand and the arms of the star gear in a manner that is pleasing to you. Clamp and allow the glue to set.

The brass tube should protrude about 1/16” from the back of the gear and likewise from the top of the hour hand.

Step 15: 30T 3-arm Spiral Gear Assembly

Drill a 7/32” hole in the center of the 3-arm spiral gear. Insert a piece of 7/32” diameter brass tube ½” long into the 3-arm spiral gear, securing it with a drop or two of superglue. Drill a 7/32” hole into the larger 8-tooth pinion, as deep as possible without drilling all the way through.

Step 16: 10- and 12-tooth Pinions

Drill a 7/32” hole completely through the center of the 10-tooth pinion.

Drill a 3/16” hole 3/8” deep in the center of the 12-tooth pinion, but don’t drill it all the way through.

Step 17: Ratchet Wheel

Drill a 3/16” hole through the ratchet wheel. Drill a 3/16” hole 3/8” deep, but not all the way through, in the ¾” diameter ratchet wheel hub. Press a 3/16” diameter by 1-7/8” long tube through the center of the ratchet wheel from the back, allowing it to protrude from the front. Press the hub on from the front. Glue and clamp the hub onto the ratchet wheel, making sure that the hub, wheel, and shaft are squarely aligned.

Step 18: Main Bearing

Locate the protrusion for the main bearing on the back of the frame as shown.

With a ruler and a pencil, mark the center. Using a drill press, drill a ¼” hole through the frame.

Insert a piece of ¼” diameter brass tube 1-3/8” in length into the hole in the frame. Place a piece of scrap on top of the tube, and gently hammer the scrap to drive the tube home. Or, press it in with a C-clamp. If the tube protrudes a bit from the frame, you may file it flush.

Step 19: Jigs

To ensure smooth gear operation, we’re going to make and use jigs to drill critical holes in the frame.

To make Jig 1, drill a ¼” hole in one end of the jig. Drill a 3/16” hole at the other end 2-1/2" apart. Partially insert a length of ¼” diameter brass tube into the hole in the jig. Don’t insert it fully, as you’ll want to remove it later. Likewise, partially insert a 3/16” tube in the 3/16” hole.

Drill a 7/32” hole through the center of the 10-tooth pinion gear. Temporarily slide this gear onto a short piece of 7/32” tube, so that tube protrudes on both sides of the gear. Insert the tube and gear into the ¼” bearing in Jig 1.

Take the 3-arm spiral gear and slide it onto the 3/16” tube, meshing it with the pinion. Manually rotate the pinion and check the gear mesh. The mesh must not be too loose or too tight; but loose is better than tight. There can be no binding. You may need to sand some teeth for a good fit. Rotate the pinion by hand for several revolutions until you are satisfied with a good fit. The gears should mesh smoothly and not catch. At the same time, there should not be excessive play or slop in the gears (some play is normal).

If the gears mesh too loosely or too tightly, remove the 3/16” brass tubing from the jig, and repeat the process, this time drilling a hole to one side of the bad hole, but closer or farther from the ¼” bearing as needed. Insert the 3/16” tubing and test the fit.

When satisfied with the fit, slide the 3-arm star gear assembly onto the shaft. Place the 8-tooth pinion onto the protruding tube of the 3-arm spiral gear, but don’t glue the pinion. Test the fit of these newly-added gears by rotating the 10-tooth pinion. Make corrections as required.

When satisfied with the fit of both sets of gears, remove the gears and the 3/16” brass tube. Insert a piece of 7/32” tubing into the frame bushing, and slip Jig 1 over that tube. Center the 3/16” hole over the frame, and clamp the jig in place. Using the scrap as a guide, drill a 3/16” hole into the frame through the jig hole. Drill only 3/8” deep into the frame.

Temporarily insert a 3/16” diameter brass tube 1” long into the hole, part way so you can remove it again. Test fit the gears in the frame just like you did in the scrap. Remove the tube.

With a 3/32” drill, make a new concentric hole inside the 3/16” hole that you just drilled into the frame. Drill at least ¼” deeper than before, but take care not to drill through the back of the frame. This will be a pilot hole for a screw when the clock is assembled later on.

Step 20: 72T Gear Assembly Shaft Hole

To make Jig 2, drill a ¼” hole and partially insert a section of ¼” tubing in one end. Drill a 3/16” hole 5-1/2" away and partially insert a piece of 3/16” tubing in the other end. This time, slide the 5-arm spiral gear assembly onto the ¼” tubing, and the 12-tooth pinion assembly into the center of the 5-arm spiral assembly. Put the straight arm gear assembly on other end.

Check the fit of the gears.

After you have created the jig to locate the rear gear post hole (far right in photo), drill this 3/16” hole only 3/8” deep into the frame. Insert 3/16” diameter tube 2-1/8” long into the hole. Test fit the gears in the frame. Keep the tube in the frame for the next step (after which you may wish to remove it for finishing).

Step 21: Ratchet Wheel Bearing Holes

To make Jig 3, drill a 3/16” hole in one end. Insert a 3/16” diameter tube into the first hole. Slide the the straight arm gear assembly onto the tubing. Because the mating 12-tooth pinion is not drilled through, we need to use a bit different procedure to locate a 1/16” pilot hole. By now you should have a feel for good gear mesh – not too loose, and not too tight. Position the 12-tooth pinion to mesh with the straight arm gear, and check the distance to the center of the pinion. In my case, I measured a distance of exactly one inch as shown. Whn satsified, drill a 1/16" hole in the jig.

Remove the 3/16” tubing from the jig, and slide that end of the jig onto the straight arm gear assembly post on the frame. Locate the 1/16” pilot hole on the center of the frame. Using the jig, drill a 1/16” pilot hole through the frame.

A 3/8” Forstner bit is recommended for this next step, but if you don’t have one, a regular bit or a brad-point bit may also work. Using the pilot hole as a guide, drill a 3/8” hole 1/8” deep. Do this on both sides of the frame.

Next, drill a ¼” hole through the center of the holes that you just drilled.

Step 22: Pivot and Lever Mount

Get the pivot and the lever mount. Drill a 1/16” pilot hole through each part. Then drill 3/8” holes 1/8” deep front and back in each part, and drill a ¼ ” hole through each, like you did for the frame.

Mark the center of the bottom of the pivot. Drill a 3/16” hole ½” deep into the bottom of the pivot.

Step 23: Ring Mounting Holes

Mark the center of the lower ring mounting arm on the frame from the front. Mark the upper arm ¼” from the bottom of the face of the arm. Drill a 5/32” hole completely through the frame at each mark.

From the back of the frame, countersink the holes using a 5/16” drill bit. Drill the countersunk holes to a depth of 1½” above the front surface of the frame. (When you insert a #6 x 2” wood screw, it should protrude about ½” from the front of the frame.)

Step 24: Pendulum Pivot and Lever Mounts

Mark the center of the frame ¾” below the protrusion shown in the photo (11¾” from the bottom of the frame). The mark must be at least ¾” from the protrusion for adequate clearance for the pivot. Drill a 3/16” hole ½” deep. Insert a 3/16” diameter by 1¾” long brass tube into the hole.

Glue the lever mount onto the frame, aligning the back of the mount flush with the back of the frame, and the top flush with the top edge of the frame. The hole in the mount must be exactly ½” higher up on the frame than the pivot shaft that you just installed. This is a critical dimension.

Step 25: Lever

Get the lever, and locate the three pilot holes that have been machined into it (the first photo is the back side of the lever).

Drill a 3/16” hole 3/8” deep, but not through, the lever at the knee using the pilot hole.

Drill a 1/16” hole completely through the lever at the next pilot hole.

Drill a 3/32” hole 3/8” deep, not through, the lever at its lower end.

Flip the lever over. Below the knee, drill a 3/16” hole 3/8” deep, but not through, the lever using the 1/16” pilot hole that you previously drilled through from the reverse side.

Glue the round lever weight onto the end of the lever.

Drill a 3/16” hole 3/8” deep, but not through, the ½” diameter lever shaft cap.

Step 26: Clutch Hub

Drill a 7/32” hole 3/8” deep into the center of the clutch hub, using the pilot hole already machined. Insert a 7/32” diameter by 5½” brass tube into the clutch hub. Glue if necessary.

Step 27: Pawls and Support

Using pilot holes where available, drill a 1/8” hole through the driving and locking pawls. Drill a 9/64” hole through the bottom of the locking pawl support. From the back of the support, drill a 3/32” pilot hole into the top of the catch support, but not all the way through to the front.

Drill a pilot hole in the center of the base of the front of frame, ½” from the bottom of the frame, for the locking pawl support.

Step 28: Frame Support

Clamp the rear support to the back of the cross support. Drill two 1/8” pilot holes through the cross support into the rear support. One hole should be just below center, so that it does not poke through the rear support. The other should be below that, close to the bottom.

Countersink the holes in the cross support. Fasten the two support pieces together using #6 brass 1 ¼” flat-head wood screws, making certain that the heads are at or below the surface of the cross member.

Step 29: Finish

This would be a good time to apply any stain, varnish, or other finish to the parts of your clock.

Many wood gear clock builders recommend applying no finish to the surface of the teeth of gears, as this can add friction. The Mystery Clock uses a fairly robust pendulum drive, so it is OK to apply finish to the tooth surfaces.

Step 30: Frame to Base

Mark the underside of the rear frame support in the center and 3/8” from the end of the support. Drill a pilot hole. Fasten the frame to the base using a # 6 1” wood screw from the underside of the base top and into this frame pilot hole, tightening sufficiently to hold the fame in place.

From the top of the base, align the frame so that it is square. Clamp in place if needed. Drill two additional pilot holes from the underside of the base top into the frame. Secure with two additional wood screws.

Step 31: Base

Invert the base and place the battery holder in position. The battery holder is secured by screws that run through the thin top surface of the base and into the frame. The preexisting battery holder mounting holes cannot be used. Use the two side screws that secured the frame to the base as a guide for the frame location. Drill two pilot holes in line with these screws, through the battery holder and into the base and frame. Countersink the battery holder holes. Secure the battery holder with two #4 x ½” screws.

The electronics can optionally be powered with an AC adapter. Drill a 5/16" hole in the side of the base at the socket recess to accommodate the AC adapter socket.

Put the circuit board, LED, and coil in place. The LED should protrude slightly through the top of the base. Bend the LED leads.

Place the coil, LED, and battery wires into their respective slots. Secure them with tape if needed. Place a small piece of foam on top of the coil, to hold it up tight against the top of the base. The coil must not be able to move after the base is assembled.

Drill pilot holes using the four shallow 1/8” holes that have been machined into the upper base as guides. Take care not to drill through the top of the base.

Place the lower base onto the upper base, and secure with four #6 x 1” wood screws.

Attach the self-adhesive feet.

Step 32: Bearings

Install 2 3/16” ID x 3/8” OD x 1/8” thick ball bearings into the lower frame, the mid frame lever mount, and the pendulum pivot.

Step 33: Ratchet Wheel Assembly

Slip an M5 washer, or a short piece of 7/16” tubing, onto the ratchet wheel shaft. (Washers are provided in the hardware kit.) Slip the shaft through the bearings on the lower part of the frame.

I’ve had mixed luck with the bearings and shafts. Sometimes the shafts slip right into the bearings, and other times the shafts need a little work with some emery cloth or fine sandpaper to get them to fit.

Slip another M5 washer or tubing spacer on the back, and then press the 12-tooth pinion on the back. Glue the pinion if necessary.

Spin the ratchet and make certain that it turns freely.

Step 34: Lever

Fasten the driving pawl to the lever using a #4 5/8” screw, with a washer between the pawl and lever. Tighten the screw sufficiently so that the pawl does not wobble, but still allows the pawl to pivot freely.

Insert a 3/16” diameter by 1-5/8” brass tube into the back of the lever as shown. Secure with a drop of super glue if needed.

Press a 3/16” diameter by 1¼” long brass tube into the front of the lever. Secure with superglue if needed.

Use a pair of M5 washers, or a couple of very narrow spacers out of 7/32” brass tubing, to fit onto the 3/16” tubing. Slip one of the washers or spacers onto the tubing, and then slip the tubing through the two bearings on the frame. Slip another washer or spacer on the back, and then push a wooden cap onto the tubing. Secure the cap with super glue if needed. Make certain that the lever pivots freely.

Slide a 7/32” diameter by 3/8” long brass tube onto the lever tubing. Next, slide on a bearing. Finally, secure the bearing by sliding on a small piece of ¼” OD x .170” ID plastic tubing.

Step 35: Locking Pawl

Using a #4 x ½” wood screw, fasten the locking pawl to the support, placing a small washer between the two parts. Tighten the screw sufficiently so that the locking pawl does not wobble, yet moves freely. Fasten the catch support to the bottom of the frame with a #6 x ¾” brass round-head wood screw. Make certain that this screw is solid brass and not brass-plated steel – test it using the magnet. Tighten the screw just enough to hold the support for now.

Step 36: Pendulum

CAUTION - Proper pendulum operation is affected by the weight of its parts. While attempting to accommodate variations in wood density and weight in the design, it’s possible that a pendulum built to these dimensions with very light or very heavy wood may not be properly adjustable, and the size or weight of the bob or coupler may need to be changed. Be sure to test the clock for accuracy before permanently fastening the pendulum parts that capture the bob.

Glue the two halves of the coupler together, making sure that the parts are aligned. A drill bit temporarily slipped through the center hole may assist with this.

Carefully expand the pilot hole in one end of the coupler using a 9/64” drill to a depth half way into the coupler. Also drill the magnet holder to 9/64”. Drill the other end of the coupler with a 3/16” drill bit, ½” deep.

The photo illustrates one method for sanding round parts like this one. I temporarily inserted a section of brass

Cut threads into the magnet holder and the one end of the coupler using a 10-32 tap. (If you don’t have one, you can drill the holes to 3/16” instead, and secure the threaded rod with epoxy. But be mindful that all the parts must be assembled, and the overall length of the pendulum must be adjusted, before you permanently glue.)

I machined a slot in the back of each half of the bob to act as a pilot hole. Glue and clamp the two halves of the bob together, making sure that the half hole slots in the back of each half are aligned. After the glue has dried, clean up the hole by running a 3/16” drill bit through the hole, top to bottom.

Insert the magnet into the magnet holder. Don't glue it yet. Hold it in place with tape if needed.

Screw a 10-32 x 5½” threaded brass rod into the coupler to at least ½”. Thread a brass knurled nut onto the rod, followed by the weight, another knurled nut, and the magnet holder.

(Make sure that all of these parts are solid brass, and not brass-plated steel. The magnet will tell you!)

Press a 3/16” diameter by 4½” long brass tube into the pivot, and then press the other end into the coupler.

Slip either an M5 washer or a very short section of 7/32” brass tubing onto the pendulum shaft. Next, slip the pendulum pivot bearings onto the shaft. Follow up with another washer or spacer.

Ensure that the pendulum pivot remains clear of the frame behind and the numbers ring (to be installed later) in front.

Also, check the range of pendulum travel. It should be free to move the magnet holder a few inches past the edge of the base. If the top of the pivot contacts the frame protrusion at extreme travel, it will probably not be possible to get the clock running properly. If this happens, do not alter the pivot. Instead, bevel/trim the bottom corners of the frame protrusion slightly until the pivot has adequate clearance.

Adjust the length of the pendulum by screwing the magnet holder up or down so that the magnet holder just clears the base when the pendulum swings – about 1/16”. The magnet should be close to the base while ensuring no contact even when the pendulum expands in length slightly due to temperature changes.

The circuit in the hardware kit has a feature to help with magnet and coil polarities. Upon inserting batteries, the LED will flash red and then green, and then the coil will be briefly turned on. If coil and magnet polarity are correct, the magnet will be repelled.

Because of the lever, the pendulum naturally rests off center to the right a bit when viewed straight on. If polarities are correct, the pendulum will be repelled to the right. If incorrect, the pendulum will be attracted to the left (center). Flip either the magnet or coil to make a correction if necessary.

If you did not tap the magnet holder and disc, use tape or other means to temporarily fasten the magnet holder to the threaded rod. Eventually, fasten the threaded rod to each part using epoxy glue. Fasten the rod to the coupler first, then to the magnet holder. Make a shim to elevate the magnet holder above the base and hold it in place while the epoxy sets. Be sure to have the length correct! Also, the magnet will be very hard to remove once it is installed, so make sure it is installed correctly. (If you thread the magnet holder, you can screw it up the rod to force the magnet out.)

Check the swing of the pendulum, making sure that the magnet holder and bob have sufficient clearance at the base and frame. If not, gently bend the pendulum’s brass tubing to ensure good clearance. But don’t allow it to be too far from the frame either. The magnet must pass directly over the coil for good operation. This is critical – even a fraction of an inch off center of the coil can result in unacceptable side force on the pendulum and failed operation.

Secure the magnet holder onto the brass threaded rod with a drop of super glue only after you have verified magnet polarity, pendulum clearance, and timekeeping accuracy.

Step 37: Locking Pawl Adjustment

Manually swing and hold the pendulum to the left. The driving pawl should rotate the ratchet wheel clockwise. Look at the locking pawl and ratchet wheel interface. The locking pawl should clear a tooth on the ratchet wheel and land about halfway onto the next tooth with the pendulum swung well to the left. Adjust the locking pawl support position until this is the case.

Next, swing the pendulum and hold to the right. The ratchet wheel should freeze movement, and the driving pawl should skip onto the next tooth. The driving pawl should land about halfway onto the tooth as shown.

Move the pendulum back and forth several times, and check both the driving and locking pawl for proper action. When satisfied, tighten the screw securing the locking pawl support to the frame.

Step 38: Ring Assembly

Mark the back of the numbers ring opposite the number 6 and ¼” from the lower edge of the ring. Drill a 3/16” hole about ¼” deep.

Align this hole with the pendulum shaft, slipping the ring over the shaft. Hold or clamp the ring into position. From the back of the frame, drill two pilot holes into the ring.

Fasten the ring with two #6 x 2” wood screws.

Step 39: Large Straight Arm Gear Assembly

Slide the large straight arm gear assembly onto the shaft at the back of the frame. Secure the assembly with a ¾” diameter round cap pressed onto the shaft. Secure with super glue if needed.

Step 40: Small Spiral Arm Gear Train Assembly

Slip the 3-arm spiral gear onto the shaft at the top of the frame. Secure it with a #2 x 1” machine screw.

Press the 8-tooth pinion onto the shaft. Secure with super glue if needed, taking care not to get glue on the frame shaft.

Step 41: Main Shaft Assembly

Place a large-diameter spring onto the clutch hub. Slip the driven clutch wheel onto the hub, compressing the spring. Position the 5-arm spiral gear assembly up to the frame, straddling the 5-arm star gear as shown. Slip the hub shaft into the spiral gear assembly and through the frame.

Slip an M6 washer or a very short spacer onto the front of the shaft, and then press the 10-tooth pinion onto the shaft.

Press the pinion on sufficiently to leave only 3/8” between the driven clutch wheel and the hub at the rear of the clock.

There should be sufficient clearance to pull the driven clutch wheel back to disengage the clutch and allow the hands to be moved.

Check for proper alignment and clearance of the gears on the back of the clock. The large gears especially may have some wobble and touch if holes were not drilled carefully. The large gears must not touch each other, the frame, or the clutch. If they touch, odds are the clock may run for a while but will not keep running.

Step 42: Hour and Minute Hands

Slip the hour hand gear assembly onto the shaft. Align the assembly so that the hour hand points straight up to 12 o’clock, using the clutch as needed.

Press the minute hand on in the 12 o’clock position.

Step 43: Setup and Adjustment

Insert 4 D cells into the battery holder. The indicator LED flashes red, then green for a second to indicate proper operation. Also, as mentioned, the pendulum should move a bit as the coil is energized once. There is no on-off switch to the clock. As long as the pendulum is not swinging, the circuit draws only minimal current. The swinging pendulum triggers the circuit.

Adjust the bob to the center of its travel, and lock it in place with the knurled nuts. Give the pendulum a gentle swing. Set the time by grabbing the clutch from the rear, pulling towards the back of the clock, and turning in either direction.

Check the LED after a minute or so. If it is consistently blinking red, the pendulum is swinging too slowly. Raise the bob to speed up the pendulum by loosening the top knurled nut and then tightening the bottom one, and then try again. Conversely, if the LED is consistently blinking green, the pendulum is too fast. Lower the bob a bit. Keep the pendulum stopped for at least 3 seconds between adjustments to allow the electronics to reset itself.

When the LED no longer consistently lights, the pendulum is initially set using short-term measurement. (It’s OK if the LED occasionally flashes red or green, as long as it is off most of the time.) The LED is disabled after 5 minutes. However, the pendulum speed is then measured with a more dependable long-term measurement. The LED may start blinking again after a period of time. If so, adjust the pendulum slightly, using as little as ¼ turn of the knurled nuts. Keep the pendulum stopped for at least three seconds after adjusting, and restart it. It’s probably best to ignore the short-term measurement blinking of the first 5 minutes after this adjustment. But watch for additional long-term warnings and then adjust again if needed.

Once the LED stays off for an extended period, the electronics will generally compensate, and no further adjustment will be needed.

When testing the clock before all pendulum parts are glued, make sure that your temporary fastening is secure. Any looseness or wobble in parts fit will affect the pendulum, and may make adjustment for accurate timekeeping impossible.

Due to variations in wood density and weight, you may be unable get the clock to run slow or fast enough with the bob at the extent of its travel. If the bob is all the way down and the clock runs too fast, try a bigger bob. If the bob is all the way up and the clock runs too slowly, try a bigger coupler (expand the diameter). It may be helpful to temporarily attach a weight to the coupler or bob to determine what is required. (Avoid magnetic weight near the pendulum magnet.)

Once you’re confident that the clock is keeping time, and there is still a bit of bob travel in reserve for future adjustment, then you may glue the magnet holder to the pendulum (it can become loose eventfully if not glued).

Step 44: ​Troubleshooting

If you can’t get your clock to run or keep running, check the following possible causes:

  • Improper magnet or coil polarity
  • Weak batteries
  • Pendulum magnet not centered directly over coil
  • Pendulum contacting frame or base
  • Pendulum pivot contacting frame at large angles
  • Gears not meshing properly
  • Rough or misshaped gear teeth
  • Faces of large gears rubbing on another gear, the frame, or the clutch
  • Bearings or shafts binding

(How do I know what to troubleshoot? I’ve built several of these clocks, and I’ve had each of these problems at one time or another!)

You may find it helpful to temporarily disassemble parts of the clock to isolate problems. For example, remove the driving pawl to see if the pendulum will operate on its own. If it does, the gear train is probably binding somewhere.

If your clock runs but then stops, carefully examine the clock at the stopping point. It may be helpful to release the pawls and move the ratchet wheel or gears back and forth to locate the source of binding, rubbing, etc.

If you find the clock ticking to be too loud, you can quiet it a bit by carefully applying cork pads to each of the two pawls. This can introduce some extra friction, so it’s best to get the clock running without these first.

CNC Contest 2020

Second Prize in the
CNC Contest 2020

1 Person Made This Project!


  • Lamps Challenge

    Lamps Challenge
  • Puzzles Challenge

    Puzzles Challenge
  • Rice & Grains Challenge

    Rice & Grains Challenge



Question 13 days ago on Introduction

Hi Dick
Are the top and bottom of the base 3/4" thick? Didn't find that on my DXF files. Also was wondering how you cut the 3D grooves on the clutch parts. I used 1/32" end mill and that seemed to work pretty well. Thanks - Mike G


Answer 12 days ago

Yes, 3/4". For the clutch I used a 1/16" tapered ball nose bit in raster carving mode (Carvewright CNC).


Reply 12 days ago

Thank you for the quick reply


Question 6 months ago

Hi Dick
Looks like a fun project. I think I will give it a try. Can you help me find the file for the numerical
ring? Thanks - Mike G


Answer 6 months ago

The ring is made in 4 sections. The "three quarter inch parts" file has one section. I did not include numerals; you'll need to add those.


Reply 6 months ago

Thanks Dick.
Not sure how I missed that! - Mike G


Question 7 months ago on Step 3

I recently stumbled across your mystery clock and think its amazing. I do have a question about the electronics though. Is the "hardware kit" mentioned in Step 1 different than the "program and electronics" mentioned in Step 3?


Answer 7 months ago

The hardware kit includes all of the non wood parts. Brass tubing, fasteners, magnet, and electronics.


Tip 10 months ago

Update on my progress with Dick's clock replication. Yesterday was my first successful test of all the functions Dick has illustrated in his main.c file from his instructables entry titled "A Wood Gear Clock With a Unique Drive Mechanism". It's taken me too long to get through the code, build the clock, and stitch both of those things together into a working model. As mentioned in an earlier post, my clock code drops reference's to Dick's use of the GRACE formally available in the CCS tool kit apparently. I hope including my version here will save you the effort I had to go through to figure that out.

The modified code includes several debug sections to either dynamically manipulate variables while the clock is running or see what variables may be in real time without pausing the code in the CCS environment and therefore stopping the clock. Those areas would be commented out during a production version use of it. Dick has suggested using an array var to capture things over time and investigate it's values during a CCS pause. That should work, but I've not incorporated it here.

I wouldn't expect you to directly use the code, as I wasn't able to directly use Dick's work line for line. But, it be a powerful tool to start from.

Also removed the short section of code where the pendulum was being pulsed at program start. My experience indicated either hook-up method of the coil wires would produce a visible 'motion' at that point. A better method for me was to run a fixed pulse width and observe the pendulum max swing excursion with both connections tested. I used the hook-up that produced the highest swing amplitude.

Still running with breadboard set up, but have gotten a perf board built with components connected to embed in the base very soon.

The included code is heavily commented, so won't go into further details here.

One loose end for me yet is use of a continuously incrementing pendulum and crystal sec count in the code. Both are currently defined as unsigned int's. I suspect the upper limit occurs at 65535. What happens at the next sec? I need to develop a test rather than waiting for 65535 secs to pass and see what happens.


1 year ago

Progress report; May 2021

I'm still 'tweaking' on my build of this. Would like to pass along some more comments.

Once again, this has been a unique solution for a time piece. I continue to like the idea and believe it's underpinnings are sound, but... Choice of components and fit of those to work together is sensitive. I've achieved running of the unit with full drive train, but it has to have maximum pulse length applied to do so. That effectively means I've currently got an electrically driven pendulum, but no margin to change it's 'speed' because the accelerator is on the floor {in car speak}. Primary load is in the cam to lever interface to drive the ratchet pawl. I've done everything I can think of to reduce that load and friction source.

Other things I've noticed in no particular order:

* I had to add some length to the upper frame to accommodate the hour hand gears.

* The lower ratchet pawl attachment point ended up higher for me than author proposed.

* I could not get through my head how the author was counter reacting the spring force used in the clutch. I did not want to side load any of the gears. Instead, I chose to use his clutch plates, but 3D printed a couple small C section clamps that were flexible enough to slide over the plates outer surfaces and hold them together. It doesn't take much effort to do that. No need for his machined grooves to add friction. Flat plates are fine.

* The extra counter weight glued to the lever that's driven from the pendulum cam was unnecessary. In the end I recreated this part with that whole counter weight section removed and replaced with a threaded #6 screw and weights {if needed} so I could vary it's effect while on the clock. You want just enough weight there to move/return the upper ratchet pawl along the teeth. In my case, the screw and a couple nuts were enough.

* The locking pawl on the lower side of the ratchet wheel also had much more counter weight than necessary. I drilled 2 large holes in that counter weight area to reduce the parts friction load on the ratchet, but remain capable of locking the tooth.

* There appears to be some 'black magic' in coil performance. Several different configurations were tried. I ended up with a form 10mm in inner dia. and 8.5mm in length. The form was 3D printed with a spindle on one end so I could rotate it with a drill during winding. Spindle was removed for use in the clock base. Used +- 100' as author suggested of #37 wire. Some tests were not producing results I expected. Believe now that some of those eradicate tests results were due to bread board integrity. i.e. resistance in connections or voltage drop. That should be taken care of when the components are eventually placed on a permanent board.

* There was a lot to absorb in getting my head around the TI micro chip and it's programming. I don't play with these things daily. The web series 'Scientific Instruments Using the TI MSP430' was extremely helpful.


Reply 12 months ago

Can you provide a dxf of the lengthened frame that you found necessary?


Reply 11 months ago

Hi Dick, good to see your out there.
I went back into my CAD files and found this DXF of the frame. It's derived from the 3D model built in Blender. The DXF conversion of it brought all the extra internal lines. Looks like it's using some info from it's .stl convert engine. Anyway, you can extract the outline from it. Just guessing that's what your interested in. FYI, it's built in mm units.

I printed it and laid it against my model just to make sure. Included a pic of my interpretation of your design and the frame printout.

Lots of activities have got in way of doing the final electronic feedback work in the electrical side of this. Once it cools down and limits my outside work, I should get back to it.
Regards; Gary Clark


Reply 1 year ago

Nice progress!

To set the time, I grab the outer clutch wheel, pull back on the spring, and rotate, then release the spring.

I experimented with different coil parameters myself. I have been using a 1-1/4" diameter form with 3/8" between the plates, wound to a diameter of 1". 32 gauge wire.

The 3.3 V voltage regulator that I used accepts only up to 6 V input. If you swap it for a different one, you can power the circuit from a higher voltage, say 12 V, and get more power to the coil.


Reply 1 year ago

Dick, I don't understand how you are winding to a dia of 1" when the starting dia was 1-1/4" for your coil form.

Research reading indicates more turns increase coil strength, so a smaller coil form seemed to theoretically produce more field. But, practical experience in the clock 'seems' to not show that. I'm starting to wonder if a flatter {larger dia} coil may play a role in exposure time to the pendulum magnet.

And a mistake on my part; I've been using #37 wire. I see you talk about 32. And I just realised your earlier instructable for electronics details also specs 32. Larger gauge should produce more current. I'll get some new wire ordered.

I'm getting about 70ohm out of 100' of #37. That seemed high to me considering you were getting 50ohm. Should have been a light bulb moment long before now. I'm slow.


Reply 1 year ago

I'm using 1-1/4" diameter 1/6" thick acrylic discs. I drill a 1/4" hole 1/32" into the discs and glue a 1/4" diameter 3/8" long acrylic tube to the recessed hole in the discs with acrylic glue. I drill through the discs and tube with a 1/8" drill so that I can pass a screw through. I secure the assembly with a nut, then wind the coil with the screw secured into my cordless drill. I wind until about 1" diameter, then secure the wire with hot melt glue.

Magnetic force is determined by current-turns. Larger-diameter wire will pass more current, but allow fewer turns. #37 wire yields more turns than #32 but less current. The smaller the diameter of the core (1/4" in my case) yields more turns at lesser resistance and more current. A flatter coil puts more turns closer to the magnet. This size coil mates well with the 3/4" diameter magnet. Thicker magnets yielded very little additional force.

Many use an iron core (a nail) for the core, but the magnet is attracted to the iron and did not work as well for me. I've had better luck with nonferrous cores.


Reply 1 year ago

Have been away from the clock development, but put a couple more hours into it recently.

After reading about your latest info regarding the coil form, I printed another 3D model to spec's you have suggested. Used #32 wire on a 1/4" OD spool of a bit under 3/8" length. Wound it to near 1" dia. Resistance measured out near 35 Ohm.

In short, this new coil has made a world of differance in the force applied to the pendulum. Yesterday's testing indicates 7-8mS fixed timing pulses will keep the clock running. Earlier coil's were needing 35mS to do this. Soo, think I've got enough horse power to respond to pendulum swing change needs now.

I also expect the feedback constants you were using for Kp and Ki are much too high with my 7.5mS nominal pulse length. You are suggesting I might be able to go +-0.5% with timing tick changes for speed up or slow down. @ 7.5mS pulse I'm using 246 ticks. 0.5% of 32768 is 164 ticks {5mS} change for speed up/slow down. I expect my clock may still run with 12.5mS pulse on the long side, but not 2.5mS on the short.

I have an o-scope available but no data logger. Can you suggest a reasonable method to come up with new Kp and Ki constants?

Have also experimented briefly with a nail in the coil to increase flux. As you found, it attracts the magnet. I removed it quickly.


Reply 1 year ago

As described, I picked the constants by trial and error. If your oscope has an exteremely slow timebase you may be able to use it as a datalogger. You can crate a data log in software by adding an array variable and store the error value periodically, say once a minute. Run the clock with the emulator for several minutes, then stop and examine the error values captured in the array. Look for the error value to close in slowly on near zero without overshoot.


1 year ago

This item has sparked my interest because of it's mechanical aesthetics and electrical integration. I have a little micro background but primarily simple pin state change and A2D conversions.

I've managed to figure out Dick's "Grace" use in the earlier release of TI's CCS to ones that are C code based only and run in the 10.1.1 version {current best of my knowledge today}. Have also had to work around Linux integration {I'm running Ubuntu 18.04} limits where sub-routines don't work in the CCS. Not that big of problem since Dick's original code used them sparingly and were short in length.

Have only built a test rig so far to pulse a pendulum like Dick's design. But, I did model and print a plastic shell for the pendulum adjustment weight. It's partly filled with lead shot to provide ballast. Bearings were purchased via Amazon. Took the seals out of them to reduce friction. The pendulum magnet was also purchased via Amazon but required some mechanical 'development' for shape optimization {came with a hanger that I removed}.

Most of the testing so far has been with the 2211 chip sitting in a TI MSP-EXP430G2ET development board. But I have on short occasions placed the chip on another bread board with needed components as Dick's circuit diagram proposes with an exterior crystal and ran satisfactory under fixed pulse lengths.

The fixed pulse length I believe is necessary now because I don't have the inertia and drag from the complete clock loading the pendulum yet. It runs fine with 15ms shots right now.

All of the electrical components I purchased via Mouser Electronics.

I've been modeling the gears and other parts from Dick's .dxf files using Blender and it's integrated CAM tool kit. I plan on cutting the gears using a home built CNC machine after plans from V1Engineering. Other clocks have been made with this CNC tool in the past but I've printed gears for those designs rather than machine them. Test tool paths created so far look reasonable. I did cut the pendulum head earlier with it.

Not familiar with how this 'sharing' works, but I'd be willing to pass along details of what I've found if desired.


Question 2 years ago

Ohhhhhh 44 steps ah ....!!Instructables typing may take 2 days to post it????


2 years ago

Wow...... Great instructables I have ever seen .. Brilliant one....