Triaxial Tourbillon Clock Build Notes
(Download the Build Notes PDF for this and more information.)
Here are the parts that I used. You can use hex socket or Philips head screws, your choice. I used hex socket head cap screws for the balance wheel weights because they were heavy. I paid about $1 a piece for all the ball bearings on ebay or Amazon.
Some of the “2mm” and “3mm” shafts that I’ve picked up on ebay or Amazon have been slightly oversized, just enough so the bearings won’t fit on to them. Look for a slip fit of the shaft in the bearing. DON’T force the bearings on the shaft. I tried and jammed the bearings, making them useless. My solution was to order from another vendor. Maybe you will get lucky the first time.
(2) 3x6x2.5 ball bearings
(4) 2x5x2.5mm ball bearings
(9) 8x22x7mm ball bearings
(3) 12x28x8mm (6001) ball bearings
(~200mm) 2mm shaft
(~100mm) 3mm shaft
(18) M2x8 flat head screw
(10) M2x8 cap screw
(2) M2x10 cap screw
(8) M2.5×8 cap screw
(2) M3x8 hex socket head cap screw
3x6x2.5 ball bearings
• 2 for the Balance Wheel
2x5x2.5mm ball bearings
• 2 for the Fork
• 2 for the Escape Wheel
8x22x7mm ball bearings
• 1 for the Moon
• 1 for the Earth
• 2 for the Main Frame-32Crowngear and the Forkframe-forkframeconnector-1
• 1 for the Main Frame-24tgearplusshaft and the Main Frame-carriertop
• 1 for the Clock Face-72tminute shaft halfs and the Main Frame-carriertop
• 1 for the Clock Face-72tminute shaft halfs and the Clock Face-Face Plate
• 2 for the Main Frame-carriertop and Mainframebottom-mainframetop a. Note that I used a A26’s bearingadapter.STL as spacer between them.
12x28x8mm (6001) ball bearings
• 1 for the lower 30tbevel gear
• 1 for the lower drum
• 1 for the upper 30tbevel gear
• 16mm long for the Tourbillon-15Tanchor2 (Fork)
• 13mm long for the Tourbillon-15Tesc (Balance Wheel)
• 13mm long (x3) to hold the three 10tbevel gears in the drum.
• 12mm long for the Forkframe-40Tgear 2
• 18mm long for the hourgear2
• 14mm long between the Clock Face-72tminute shaft halfs and the Main Frame-carrierbottom
• 13mm long between the Main Frame-24tgearplusshaft and the Main Frame-carrierbottom
• 10mm long between the Main Frame-Earth North and the Main Frame-carriertop
• 10mm long between the moon and the Main Frame-carriertop
• 46mm long between the Mainframebottom-Mainframebottom2 and the Mainframebottom-18tgearplusshaft 2
• 51mm long between Forkframe-forkframe back 2 and Forkframe-forkframe1
M2x8 flat head screw
• 3 to hold Tourbillon-tourbframeback2 to the Tourbillon-tourbframemiddle2.
• 3 to hold the Tourbillon-tourbframemiddle2b to Tourbillon-tourbframemiddle2.
• 4 to hold the Clock Face-Face Face to the Clock Face-Face Plate.
• 2 to hold the Clock Face-Face Bridge to the Clock Face-Face Plate.
• 2 to hold the Main Frame-carrierbottom to the Main Frame-carriertop.M2x10 flat head screw
• 2 to hold Tourbillon-Tourbframetop2 to Tourbillon-tourbframemiddle2.
• 2 to hold Forkframe-forkframe 2 2a to Forkframe-forkframe back 2.
M2x8 cap screw
• 4 to hold Forkframe-forkframe 2 2a to Forkframe-forkframeconnector-1.
• 4 to hold Forkframe-forkframe1 to Forkframe-forkframeconnector-1.
• 1 to hold the Main Frame-39tgear cap to the Forkframe-forkframeconnector-1.
• 1 to hold the Main Frame-39tgear cap to the Main Frame-carriertop.
M2x10 cap screw
• 2 for the Main Frame-32Crowngear to the Main Frame-carriertop
M2.5×8 cap screw
• 8 to the Mainframebottom-Mainframebottom2 to Mainframebottom-mainframetop c. Had to drill this out a little to get them to fit.
M3x8 hex socket head cap screw
• 2 for the Tourbillon-balancewheel to add weight
Some sources I’ve used
10pcs 2x5x2.5mm MR52-ZZ Precision Ball Bearings Chrome Steel, Metal Shield
3x6x2.5 mm Miniature Steel Bearings MR63ZZ L-630 673ZZ Deep Groove Ball bearing 10 PCS Skateboard Bearings
M4x8 hex socket head cap screw
The Astronomia was on my list of clockworks to model too, but since A26 did so much work on his design, I thought that I would start with it (https://www.thingiverse.com/thing:3061474). I’m very impressed with the effort that he put into it.
The assembly instructions were a little lean, and I haven’t found an easy way of documenting a complex design either. So I took an easy way, I imported all the STLs and created some assembly so that I could see how all the parts fit together. My posted ‘Assy’ STL files show the whole assembly and several sub-assemblies, which can be viewed with Meshmixer to get an idea of what goes where.
I made many modifications and additions to the design, hopefully to improve its accuracy and the reliability. And a couple of the changes were just to make it look a little more like the actual Astronomia. With these changes it runs a couple hours with about 5 feet of nylon 30 pound test fishing line. It is hard to see in the time lapse mpeg movie file.
- As A26 said, the clockface was much heavier than the tourbillon. I found that when mounted on a wall, it would run fine as the clockface dropped, but would stop on the upswing. So I redesigned the clock face to be lighter (and to look more like that of the original watch), and added a counter weight (packed with #4 lead shot) that clips onto the 32crowngear. The result doesn’t perfectly balance the assembly, but now it will reliably run through a full revolution.
- I changed many ‘bearings’ from plastic on steel shaft to ball bearings. The ball bearings are cheap, (typically, $1 each on ebay or Amazon), they made all the tolerances more accurate, and I figured that for all the time I was taking to put it together, it was small price to pay for a reliable design. I don’t know what they use to lubricate those cheap ball bearings, but many of them were pretty stiff out of the package. However I found that it was nothing a drop of Mystery oil couldn’t fix.
- I modified the escape wheel (15Tesc) and the fork (15anchor) to use 2x5x2.5mm ball bearings.
- I was showing a print of the original escape (15Tesc) wheel design to a friend, and he accidently snapped one off a tooth handling it. The teeth were only a couple fuses thick, so I beefed them up by adding chamfers to them.
- I modified all the components that rotated on the main tourbillon shaft (tourbframetop, tourbframemiddle, and the balancewheel), to use a 3mm shaft and 3x6x2.5 ball bearings.
- I split forkframe2 into 2 parts, the forkframe and the gear that engages the escape wheel. I did this for 2 reasons; now both parts to be printed without supports and glued together. And I made the standoffs for the gear 1mm longer to better align with the escape wheel gear that it engages with.
- Since I was tweaking the design a lot, I changed most of the parts that were originally to be glued together, to be screwed together instead. Now the tourbframetop, tourbframemiddle, and tourbframeback can be disassembled.
- I cut my main tourbillon a little short (which I have since fixed), but I was having problems with the tourbillon forkframes (1 and 2) spreading and the shaft popping out, so I beefed them up, making forkframe 1 a mm thicker, and forkframe 2 wider. And adding chamfers that rest against the forkframeconnector. I also changed the design of forkframe 2 and forkframe back, so that the forkframe back is now attached with screws, rather than glue.
- I was having problems with the teeth of the 40Tgear walking out of the teeth of the crowngear as the tourbillon rotated, causing the whole clock to jump in time. Instead of rotating in the plane that it was supposed to, the 40Tgear would bend/flex a little, and work its way up and out of the crowngear teeth. And if the gears didn’t separate so much that they would disengage, then you would hear pop (and the whole assembly would shudder) when the 40Tgear popped back into alignment. So I tapered the teeth on the crowngear and the 40Tgear, and made the 40Tgear teeth thicker so if it did walk up, there was still enough ‘tooth’ left to remain engaged.
- To allow the tourbillon to be removed from the crown gear, I printed a plastic washer, and modified the end of the forkframe connector to accept a 2mm screw, so the 12Tgear could be secured with a screw, rather than being glued on.
- The design worked fine when resting on a table, but when I hug it on the wall the carriertop and everything attached to it would pop out. This was because I had not glued the 36tgear to the carriertop shaft. To allow the thing to be disassembled, I printed a washer, and used a 2mm screw in the hole that was supposed to be used for a 2mm alignment shaft between the carriertop and mainframebottom, to secure the 39tgear to the carriertop shaft. This also meant that I had to shave 2mm off the top of the mainframebottom center shaft, to make clearance for the washer and screw.
- To make it easier to see what I was doing in the previous step easier, and to allow the mainframetop to be printed without supports, I split it into 3 pieces. After printing, the middle part of the mainframetop is glued to the top part, and the bottom part (that attaches to the mainframe bottom), is then screwed to the top/middle parts.
- The torque generated by the 5lb weight that I was using, caused the two 30Tbevel gears to spread apart and I was afraid that they would slip, because only the tips of the teeth were engaged with the three 10tbevel gears. Also the bearings were not a particularly tight press fit into the 30Tbevel gears and I was afraid that they might pop out from the stress, so I printed some 24x35x1.5 washers, and glued them to the 30Tbevel gears to keep the 30Tbevel gears from pushing themselves off the bearings. Using the 1.5mm thick washer, rather than the 2mm thick spaces, brought the two 30Tbevel gears closer together, requiring moving the two 2mm spaces so they were between the upper 30Tbevel gear and the bottom of the mainframetop.
- The mainframebottom has a chamfer in one leg. If I positioned that leg on the bottom when I hung it on the wall, gravity would cause the pawl to drop and not engage, so I added a spring to the pawl so it always positively engaged the 30Tbevel gear.
- The 8x22x7mm ball bearings that were used for the earth, the moon, the crowngear, the clockface and other places, were not particularly tight fits. Some bearings would slide right out. So I printed some 19x22x1 rings that I glued in place, to keep the bearings from moving.
- During assembly I bumped the clock, and when it fell off my desk, the main clockface shaft on the 72minute gear broke. To make the shaft a little stronger (and to shorten it a little to take out some of the slop), I split the 72minute gear from its shaft, then split the shaft into two halves, so it could be printed parallel to the print bed, and not use supports. My theory is that laying down a shaft for printing means that the plastic fuses are extruded the length of the shaft, vs radially if the shaft was printed standing up, making it stronger. I glued the two halves of the shaft together, stuck it in a drill and used a file to smooth it, and to take it down so that it was a tight fit in the 9mm hole that I had put in the 72minute gear, then glued them together.
- The posted STLs for the earth had different underlying grids for the North and South hemispheres, which looked a little goofy. So created a simple hemisphere, and used Meshmixer to merge A26’s north and south earth STLs with my simple hemisphere STLs, and hide the grids.
- I also added the center badge that the original Astronomia has, just for looks…
- I made some parts that can be used to make a weight for the clock. It weighs about 5 pounds when filled with #4 lead shot.