3D Printed Pendulum Clock

by StevePeterson Mar 29, 2019
Download All Files

Thing Apps Enabled

Please Login to Comment

Hi Steve!

This is AMAZING! I'm blown away by this marvel. I'm a clock guy, I have too many to count. I got the love of clocks from my grandfather. The reason I was drawn to your clock is my grandfather's brother made clocks by hand and I've always wanted to do the same, two in which I have in my house.

Before I ask my question I would like you to know I'm new to the 3D printing world and haven't learned everything yet. My question is, and if this as already been asked sorry, can all the files be scaled down in a slicing software to print on a smaller printer and not effect the outcome. I have a Qidi X-Smart with a build volume of 160x150x150mm. If I scale down each part by the same amount, for example going from 100% down to say 80%, will the clock still function properly? I know clocks are very sensitive and need to be perfect for them to work properly. Before I embark on this project and invest the amount of hours it will take to finish I want to make sure that at the end I'll have a smaller, but working clock.

Thank you for putting in the work and making this available to all. It truly is amazing and I can really see your love of clocks throughout this build.

Thanks Walter,

The frame and the weight shell are the only parts that would have trouble fitting that build volume at 100% scale. The split version of the back frame would probably fit. The front frame includes the 167mm diameter dial with no easy way to split it. The weight shell will be too tall, but it may be okay to shrink the height. A section would need to be taken out of the center. Simple Z scaling would result in clearance issues around the pulley. The pendulum shaft might also be too long, but it is easy to split into 3 pieces.

Back to the original question, the clock could be linearly scaled for everything except the pendulum length. Many parts are designed to fit specific sized bearings or bushings. There are 3 skateboard bearings and 2 smaller bearings that would need to be replaced with something smaller. All of the stainless steel bushings would also need to be replaced with something smaller, or possibly made to fit using a 1/8" drill. I suspect that the drive weight will probably need to be at least 4-5 pounds, so only scale the height. The lettering on the dial is designed to print perfectly using a 0.4mm nozzle. Some lines might disappear if scaled unless the nozzle size was also reduced. You can see the results in the slicer before committing to print.

Another option is to print all the components except the frame and weight shell. The largest gear is around 80mm. The frame could be made from wood. Or you might be able to get it printed at the local library or other print service.

I am looking into options to reduce the design size, but it would likely be many months away.


Hi Steve! Thank you so much for splitting both front and back plate to enable printing on smaller beds. I have 1 (stupid?) question:
Your instruction say "epoxy the parts together with 2" pieces of 6-32 threaded rod for strength." Since there are 2 'dowel' holes provided how can I screw the threaded rods on the second part as I can't rotate this part around the threaded rod. Or do you mean to have these threaded rods used as dowels, i.e. just push them in with the epoxy glue and rely on the threads to give more adhesion?

Hi WildGinger,

Yes, the threaded rods are acting as dowels. The threads give something for the epoxy to grab on to. The hole should naturally have enough surface texture for a good grip.

The forces on the pendulum rod are small because the bob only weighs a few ounces, so threaded rods are overkill. Any type of dowel or glue would probably be acceptable here.

The frame components support the weight shell pulling straight down. There are 4-5 pounds on the front frame and 2-3 pounds on the back frame because of the position of the weight shell. The entire clock could crash to the floor if it breaks loose, so a strong bond is needed. Threaded rods and epoxy are strongly recommended here.

Also note that the split frame might be untested. It should work the same, but I have only built the full sized frame. Do a dry fit and measure the hole to hole spacing. Lightly sand the parts if needed to keep the spacing the same across the break as it is everywhere else. Please post a make or a comment if the split frame is working.

Good luck with your build.

Hi Steve, from me too, thanks to this splendid project (not only the design, but also the excellent build notes and... you support for people with strange questions like me lol).
I am still in the phase of understanding your design and test-printed some parts, especially those that need some pre-assembling. I live in the "metric-world" so it is not easy to find metal parts that fit your design...
I am currently on the gear4 part (minute hand) and have several questions :

  1. If I do understand well, the arbor will rotate in the hole of the back frame, and the gear4_lower part and the gear4_18 are tight to the arbor, so the arbor will be driven by gear4_18. Gear4_54 can rotate around the arbor, and has the spring-construction, so it can in the final position be moved (how ?) about 0.1" to the front of the clock. How would this "...allow changing the time without stopping the pendulum..." ??? Should one manually move the gear4_54 to the front while changing time ?
  2. The spring presses gear4_54 towards the gear4_lower part. That creates quite some friction in my part: if I rotate the gear4_54 it stops after less that one round. In addtion, gear4_lower part has a chamfered end, so the contact surface under pressure is even bigger. I understand friction is a major issue. Did you consider replacing gear4_lower by a small 0.1" tight-fit spacer, and have a R2 bearing Inside gear4_54 ? What do you think?
    Many thanks!

Hi MArimont, thanks for noticing the build notes. It is a strange world where I have trouble finding metric size parts. Many parts should be interchangeable with the closest metric or imperial equivalent.

  1. Gear4_lower, gear4_18, and the minute hand are all locked to the arbor. The arbor floats inside gear4_54_18 and should be able to rotate in either direction with a slight resistance. Gear4_54_18 is the only component that is involved in the power train. It transfers power from the winding drum to the escapement. The arbor normally rotates along with gear4_54_18 unless the time is being set. There needs to be enough friction to hold the arbor in place during normal operation and enough slippage to allow changing the time. The margin between enough friction and enough slippage is quite large.

2a. Gear4_54_18 should be able to rotate an infinite number of times in either direction. Is the spring biting into the PLA and either expanding or closing around the arbor? I can't imagine any other reason for the two halves to lock. Does it lock in only one direction and still rotate smoothly in the other direction? That would still be functional with the restriction that time can only be changed in one direction. Another option is to remove gear4_upper and use a small washer to prevent the spring from biting into gear4_18. Gear4_upper is mostly cosmetic to hide the small portion of the spring that would otherwise be visible.

2b. The chamfer on gear4_lower is designed to match the inside of gear4_54_18 which requires a chamfer to print without supports. This is one place where a small amount of friction is deliberate. Gear4_lower should be able to spin freely inside gear4_54_18 before they are assembled on the arbor. If not, then sand gear4_lower until it fits easily inside gear4_54_18.

Hopefully this helps,

Hi Steve, thanks for your quick and extended answer!

Yes, the world would be easier with a common measurement system, but... maybe also with less fun ;-) ?

I understood from your explanation that the small friction on gear4_54_18 is wanted. I am not sur yet I understand why, but I will probably find out when progressing in my builds :-)
I think there was a misunderstanding : gear4_54_18 is indeed rotating freely (so there are no "locks"), but when I gave it a push, it only rotated for about one tour (due to the friction).
Gear4_lower also can spin freely within gear4_54_18, but due to the pressure from the spring, it generates the friction talked about above.

I also saw your discussion with "ragbagger" about the form and printing of the gears and theeth. I printed the gears with "3 perimeter" and "0.3mm elephant foot compensation" settings in prusaslicer, and they came out perfect ! No cleaning or sanding really necessary...

I redesigned the ratchet somewhat so it doesn't need springs anymore: everything is 3D printed. I'll share the results if it works perfectly ;-)

I'll keep you informed if I do (or need to do due to incompatible metric systems) other changes.

Hi MArimont,

If gear4_54_18 spins for a complete rotation before stopping, then it sounds like it may be too loose. The clock would still operate with power being provided through gear4_54_18 to the escapement, but the minute hand could slip and the clock would lose time. If my understanding is correct, then the solution is to stretch the spring to provide a bit more pressure. You could wait until the clock is fully assembled to see how it operates. As mentioned previously, the operating range between too loose and too tight is huge. You would still be able to see the clock ticking even if the minute hand is slipping. It should be OK to carry on with the rest of the build and debug this issue later if it needs to be fixed.

I would like to see any suggested changes if you want to post a remix.

Elephant foot compensation looks like a great improvement. I never noticed it before, but will definitely start using it. 0.3mm sounds a bit extreme considering that the nozzle is only 0.4mm, but you can't argue with perfect results. :)


Hi Steve, Thanks for posting this project. I have one question regarding the weight. I see your design uses about a 6lb weight and has a pulley on the weight. The string goes from the spool drum, thru the weight pulley and then back up to the clock frame. What would be the disadvantage of connecting the end of the string directly to the weight, and bypassing the pulley? Would this change allow the clock to function on less weight (like 3 lbs)? If so, would the only downside be that the clock would have to be wound more often, since the weight would lower more quickly without the pulley in place? Or is there some other reason for using the pulley system on the weight that I am missing?
Many Thanks!

Hi jasejtw,
Yes, the clock could run using 3 lbs with a single string. As you mention, the run time would be cut in half to 4 days. Alternatively, the gear ratios could be changed to bring it back to 8 days using 6 lbs of weight. This would shift the center of mass to one side and the clock will hang sideways from the single mounting screw. This could be overcome by offsetting the keyhole hanger or moving the location of the winding drum.
I did consider these options, but eventually selected the existing design primarily for cosmetic reasons. The pulley keeps the clock symmetrical. The loop of string performs two functions. The other end of the string is near the front of the clock, so the weight hangs slightly forward to provide clearance for the pendulum. It also helps to gently guide the string onto the middle of the winding drum. The cost is an additional bearing in the weight shell, a few extra feet of string, and around 30 turns of the crank every week.
My next clock might be a single string design.

Dear Steve, I am sorry, my English is not good. I can only speak Chinese.
Thank you very much for your design. I use it in teaching and teach it to children.
My foundation is not very good, so it took me about a month to basically understand the principle.
My teaching plan is in progress and I have already had some results to share with you. Below is my 13-year-old student's painting.
We made some changes. All gears, pallets and escape wheels are drawn in 123d, and the modulus and number of teeth use the same parameters as you. The 12-tooth gear of the escape wheel was changed to an iron.
The copper sleeve of the fixed iron was replaced by teflon (2*4).
Most of my energy is spent on drawing trays and escape wheels in 123d. Although it is not necessarily standard, it does work.
I will send it to you when I make a complete timepiece.
I tell you that these are mainly the feeling that your work will become very happy in the distant China.

xuyang, Tell your student that the design looks great. I would like to see the completed design.

Dear Steve, I think we will finish it.
I have some problems. But I am not sure if I am right or wrong. When I was drawing the tray, I found that according to the principle, what I painted was inconsistent with the generation of gearotic. The inconsistency is mainly the angle of the impact surface.
In principle, when the two lines indicated by the red arrow in the figure should be 45 degrees, the tray receives the most power. The angle generated by gearotic is less than 45 degrees. The gray color in the figure is generated by gearotic.
I am still thinking about the reasons. I don't believe that I am right, gearotic is wrong. But in principle, it is certain, so I am confused.
The figure is not a usable version, it is the ideal tray and escape wheel.
Over.stp is my design.

Hi xuyang,
You may be correct that a 45 degree impulse angle produces the highest efficiency. There is an explanation at http://www.nawcc-index.net/Articles/Headrick-EscMechanics.pdf with an efficiency chart on page 9. It was a while ago when I generated the escapement, but I recall adjusting a few parameters to widen the tooth for 3D printing. There does not appear to be a way to adjust the impulse angle. My design is around 26 degrees which would have a maximum efficiency of 40% compared to a 45 degree angle with a 50% efficiency. It looks like an experiment is needed.

Over.stp seems like a good design. One thing you need to be careful about is pointy tips on the escapement wheel. That shape might work great for brass clocks, but 3D printers cannot print all the way to the narrow tips. Chojimian.stp is probably not printable. The tips are only around half of the desired length.


I tried a new pallet with a 45 degree pressure angles. Here is what I found. This design has a 30 tooth escapement with a 7.5 tooth span, so the pallet touches the escapement at two places that are 90 degrees apart. With a 45 degree angle, every 1 degree of escapement rotation pushes the pallet by 1 degree. Each tick allows 6 degrees of escapement rotation consisting of 5 degrees active and 1 degree of drop. This implies that the pendulum must swing a minimum of 5 degrees or +/-2.5 degrees in each direction. The actual design had a small amount of overlap so it required around +/-3 degrees of motion just to get escapement to operate. There was enough energy to push the pendulum around +/-4.5 degrees, so the clock was functional, however the pendulum needs a very large swing for the clock to operate. A design with a crutch might allow a smaller pendulum swing as long as the crutch has a large swing. This would involve some sliding friction, although this may be small compared to the sliding friction of the escapement.

The original design with an approx. 26 degree angle has around 2 degrees of escapement rotation for every 1 degree of pendulum movement. This allows 5 degrees of escapement rotation with only 2.5 degrees of pendulum swing. A small amount of overlap allows the escapement to tick at around +/-1.5 degrees and the clock is functional with around +/-3 degrees of total pendulum swing.

Thanks for encouraging this experiment. I learned a few things. The first is that the escapement has a fair amount of tolerance and can operate across a wide range of pallet angles. The maximum efficiency comes at a cost of requiring a larger pendulum swing, so the extra air resistance may negate the additional efficiency. I prefer the smaller pallet tip angles that allow smaller pendulum swings.


I probably understand what you mean. I have been very busy recently and have not responded.
Today I brought the finished work. It is not perfect.
But it does work. . I like watches and I look forward to your new work. I will come to see your work from time to time.
I also want to try to make some new watches. But I need to have more spare time.
Below we basically complete the work.

Thanks for posting videos of the modified clock design. It's great to see them working with all the modifications and I like the bright colors.

Dear Steve, I also use this pdf as a theoretical basis. I think gearotic should not be wrong, maybe there is a reason why we don't know.
Whether it is right or wrong, when I have time, I want to do some experiments, even if the experiment does not necessarily find the cause. Also look for some information to see if you have ignored anything.
Before my students painted, I didn't have time to experiment to verify.
I will continue. If there is new progress, I will come here to tell you.



Might sound odd, but any chance the 4 day clock gearing is still posable? Or would it take way too much modding?
I assume it would be more work then changing sdomething about the weight train gearing..
The reason I ask is because, seems a good way to start braking it in, until you get it fully working.

regardless. I do plan on printing this clock. As it looks amazing.And I have a love for timekeaping machanics. Looking forward to anything else you might design. What would even come next? ballance wheels? I would assume that could be even harder to get good run time out of then this already seems.

The 4 day gear train is a relatively small change. Gears 8 and 7 change from a 54:18 teeth to 44:28. This changes the ratio from 3:1 to 1.57:1, so the winding drum diameter was also increased slightly. My initial clock had the 4 day gear train with simple 1/8" steel arbors. It would randomly stop every few days even with larger driving weights. The key breakthrough was to use 1/16" arbors inside steel bushings.

I printed an additional copy of the clock with some unrelated experiments. The initial trial had a pendulum swing of around +/- 2 degrees, but it only took a few hours to increase to nearly +/- 4 degrees. You can help the break-in period by adding additional weight to the side of the weight shell. Also, remove the pallet and let the clock run down over 5-10 minutes without the pendulum slowing it down.

I am not completely sure what comes next. There are several projects in the early stages.

Thanks for the compliments and good luck on your build.

it's interesting to hear that the changes needed are quite small. I did however think after a little bit, that the easiest way would just be changing the weight drum itself right? as a larger drum means it gives more force, and drops more per tick.. at least I think so.
regardless.. I will have to hunt around for compatable metric bushings and such being in the UK. wish me luck!

And that sounds quite nice, it is in the end, a very good design and that a 8 day clock works when mostly 3d printed amazes me.
good idea by the way, letting the clock free run a few times should bed in the gears nicely, then use the pendulum. maybe out of tune to let it run fast for a bit.

And no worrys. on my end, I want to one day work out how to make a 4 day Tourbillon clock, or at least, a clock that uses one, they may not be all that needed, but they make for a wonderful looking part.

Lastly thank you.. I might need it, right now printing the gears in ABS. the idea being I can smooth them carefully if needed to keep friction as low as I can. however, the frame.. I need to order some PLA for that, even with my MK3S I don't want to try that in ABS.

Ohh! one last thing for now, how long will it run from one turn of the winding handle? I always find info like that somewhat fun.

Increasing the diameter of the winding drum would increase power at the expense of run time. The weight shell can also be increased in size. The frame on this clock should be able to handle at least twice the weight. However, the best option is to tune the clock to get it to run on just slightly more weight than it needs to run reliably. Some books say to reduce the weight until the clock stops, then add 20%.

I have my printer inside my house, so ABS is not really an option for me. The internet says it could have up to 8% shrinkage as it cools. The gears might end up so small that they don't mesh at all, especially if the frame is in PLA with minimal shrinkage. You could upsize the gears before printing if you really want them to be in ABS. It would take some experimentation. I use PLA from Amazon or eBay and I rarely pay more than $15 per Kg. It seems to work just fine.

that is all true. And i'll be trying things out to see what works in the end. be it wraps of tape for testing the clock. or very heavy weights.
but you are right about that being the best choice. so long as I get it running i'll be happy, best working i've had so far is A26's one plate clock. Which I could get to work as a 1 day clock.. but it often stoped after a hour or so and could never work out why. I did everything I could to improve it.

wanted to say, if you ever do another 8 day clock, why not add in a day of week dial? Or moon phase?
Would be interesting to see more complications in printable clocks!

And I am aware of the shrinking and problems ABS poses.. I print in a very very well vented room. so lowered consern. However I will one day soon be enclosing it and filtering the air outside.

Also that is a good price, around what I pay for PLA.. Only time I pay more is for ASA/ABS+ Or other exotic filaments.

Anyway being in the UK, not sure where to get lead for weight, Might look for other highly dense metals I could use. Sadly can't think of many, as copper is costly. And tungsten is not cheap aswell.

It is surprising that the UK seems to have plenty of clock makers, yet lead is scarce. I bought 50 pounds of used lead shot from a firing range for under $2 per pound including shipping in a flat rate box. The going rate now appears to be around $3 per pound in the US. Scuba diving weights look to be around $4 per pound.

Different metals would be an option if the weight shell was larger. Lead is 1.44X heavier than steel, so the diameter only needs to increase by 1.2X to be close to the same weight. There is enough clearance in the design to support this larger size. The components could be scaled up in the slicer. The bottom plate screws might be loose, but it would be OK to glue them in since there is never any need to open the weight shell.

A day of week dial or moon phase are interesting options. The most common question I get when people see my clock is "Does it chime?". That would be a nice addition.

Yeah.. I think it's because it's a toxic metal to some people.. and honestly. to any kids. it makes it hard to get but I understand the reason.
I can and will keep looking around.
But if worst comes to it. Will just scale the weight up for it! so long as it works. I know of a few dense metals.. but, none compare to lead for cost in the US.

Really looking forward to seeing this thing ticking away happily! Will gladly post a make once it's started ticking!

I think they are, oddly, chimes are something I never care for.. I would keep the clock in my room, don't want the thing dinging when i'm trying to sleep. but I can see the reasons. It would make it feel more. finished?

Lead will probably get harder to find everywhere. Too bad it is so toxic. But anything denser is crazy expensive. Tungsten would be great with 1.7X the density of lead, but 50X the cost. What other metals are you considering?

This clock does have a fairly loud tick. I have it in my office and it is a soothing sound. Not sure if I would want it in my bedroom though. I may try to design a grasshopper escapement in the future. It would be almost silent.

At this point the best choice seems to be using a load of pure iron. from the cost point of view.
copper is more dence then steel/iron but also a lot more costly.
Sadly tungsten is too much more costly, and most other metals around as dence as lead are very costly.. or insanely bad. (mercury being the stand out there, costly, toxic, nightmare to handle.)

And that's understandable.. I tend to be a very heavy sleeper anyway, but we'll see on that. I might end up placing it in another room or something.

Because of metric units.. I'm gonna be using m3 brass tube, OD3 ID2. might need to faintly drill out parts.. but should all work, the idea for brass is to keep frition as low as I can, I remember brass being a very slippy metal.

Also from testing, the clock gears are just fine in ABS! getting them warp free is not the most easy, but a raft helps

OD3 brass tube should work great. Pick a hard alloy if it is available. The online metal suppliers in my area list 260, 270, 272, and 330 alloys. 260 is labeled as the most ductile, so it is probably very soft. 270 or 330 might be better choices. Select music wire that just fits inside the tubing.

Good to know about ABS gears. It will be interesting to find out how the pallet performs. This is the part of the clock with the most sliding friction, but also the smallest load. It could be sensitive to warping.

Indeed.. I might end up hand making parts as needed, but we'll have to wait and see on that. but regardless.. It's coming along slowlyh. got all the main gears printed. Will pick harder metals if I can, I assume to slow the wearing down as much as one can. in which case bearing bronze would likely be best, but not something i've any knowledge of. I've only really worked with copper and aluminium and even then only lightly.

And, I am aware of that, it's the main drive train part I'm making PLA.. the escapement wheel and pallets. Still, showing promise. And even if I have to redo gears in PLA it's no problem, the ABS I am using now, was something I was gonna throw out before it suddenly started printing really well!

Steve, this is a proper project very impressive! Years ago I made a pendulum clock out of Meccano and if I recall got about 15 minutes run time this one should be much better.

Thanks. Long run time was one of my primary goals when designing this clock.

Steve, this is an excellent design with the most comprehensive instructions I've seen here.
One minor issue: the BOM in the instructions calls for 12 #6x3/4" wood screws, when 13 are used (the one used to attach the knob to the crank handle doesn't appear in an illustration). I suppose I'll have to return to Home Depot again before I finish...
Thanks for the great project!

Sorry about that. Yes, you are correct, there does appear to be 13 #6x12 screws.

You should be able to wind the clock without the knob on the crank handle. It would be too embarrassing to make a trip to HD to buy 1 screw. :)

Steve, I have been reviewing your design and it appears that the involute gears you designed are using a 20 degree pressure angle. Reducing the pressure angle lowers the load on the pivot and thus reduces friction. I would like to try a remix of your clock using 14.5 degree pressure angle. The one problem I have so far is that I can't get a usable STP model of the frames from the STL files. Is there any chance to get the frame models in STP or IGES? The rest of the parts are fine. I am designing in Alibre and have a cool Excel macro that will generate all of the gears which includes the ability to adjust tooth thickness to provide backlash. It also adds root fillets and tip relief to add strength and better motion. I would like to get the frame to edit to add some more bearings instead of bushings.

I added STP files for the front and back frame. They are untested. The front frame has very little room for bearings, so you may need to add some material. This was deliberate to keep the gears visible.

I am curious what improvements you come up with. Please post a remix if you can.

The original gears have 20 degree pressure angles with 48% tooth width. 14.5 degree pressure angles have a bit more undercutting on the pinions even with a minimum of 12 teeth. It might still be an improvement. I had to increase the spacing from 1.65" to 1.66" to prevent binding on my test cases. In retrospect, it might have been better to reduce the tooth width much more, possibly as low as 40% to provide extra clearance. And the roots could be made slightly deeper.

Another idea I am considering for my next clock is to optimize the tooth profile for 3D printing. Standard gear teeth are great for general purpose use and they are easy to machine. A clock has different constraints. The gears always rotate in a single direction so the back side surface is unused. I am running some test cases with the tooth profiles shown below. The normal gears leave lots of tiny gap fill segments. Each time the print head retracts, it allows a possible string to be left behind.

The gears labeled crazy_gears have a tooth width just wide enough for one inner pass and one perimeter. There are very few retractions and the surfaces are extremely smooth compared to normal gears. They also print about 25% faster. One downside is that the gears are uni-directional so you need a different gear for clockwise and counter clockwise directions. I may try this on my next clock.

Thanks for the STP files, they work perfectly. The crazy gears you propose are interesting but I would expect the tooth bending strength to be a little less than the standard tooth due to the reduction of width at the base. For the tooth loading on most of the gears this may not matter but the bending strength goes with the square of the width at the base. I will probably edit the pivot points to the standard pitch diameters. I use a different nomenclature for tooth thickness where 100% is the standard tooth tooth thickness with no backlash. I have found that if I use 96% of full thickness I get gears that mesh fine printed on Prusa MK3S. I like to stick with standard PD because most of the gearing I have been doing is planetary sets. My latest project was a planetary shot glass holder that uses enlarged set screws as shot glasses. I attached a picture. I also printed all the nuts and bolts that put it together. I have found that I can print threaded fasteners for STP models downloaded from McMaster Carr. If I scale the female threads by 104% in the x and y plane (leave z alone so it doesn't change pitch) then they screw together fine with no clean up. I should probably post the shot glass project when I get a chance.

On adding material to the frames for bearings, I was thinking of increasing the gear hubs and putting the bearings in the gears so it wouldn't change the frames much, I would just increase the pin diameter a bit to fir the ID of the bearings. Thoughts on that?

The only gears with any significant tooth loading are the winding drum and the ratchet. These gears are slightly thicker to compensate. The idea to place bearings in the gears seems like a good idea. It probably works better if they are spaced as far apart as possible.

Your nomenclature for tooth width makes perfect sense. 96% of one tooth is the same as 48% each for a tooth/pinion pair. There are a few comments about gear teeth binding with some builds using different printer brands. A clock has the same accuracy with 2% backlash or 10% backlash. My next clock design will have a lot more backlash to avoid any risk of different printer tolerances.

The crazy gears seem quite strong. The rim flexes before the tooth starts to bend or break.

Thanks Steve. May I also complement you on your excellent build instructions. They are by, by far, the best I have seen. You must be some sort of technical writer. I have had hundreds of engineers work for me over the years and none of them could come up with as clear a document as you put together. Congrats. I will let you know how my edit on this comes out and if it results in anything as good as what you have.


Brad, Thank you so much for the kind words.

I am a recently retired computer engineer. Fine tuning the design and creating the build notes were my first projects after retirement. Hopefully, there will be a few more clocks in the future.

Good luck on your build. Please keep us posted if you have any suggested improvements.

I'll be posting a make soon - just finished it. My first test only ticked for 30 seconds or so. I found that the ends of the escapement teeth weren't as sharp as they could have been (there was a sort of bulge at each one) so I'd end up with places where both ends of the pallet were contacting the escapement at once. Not good. A bit of work with a file re-profiled each escapement tooth to a nice sharp point.

Also, my pendulum bearings weren't as good as I would have liked. I grabbed another pair from my order, cleaned them, and applied a lot less of the dry chain lube and got to where my pendulum would swing free for about 5 min - much better than before.

Now it's been ticking away for about 10 minutes - the pendulum swings about 2 degrees to either side. I'll wait a few days and then start adjusting the pendulum for proper timekeeping.

Many thanks for the design! This was a lot of fun to print and build.

Steve, this might be a dumb question but, would it be possible to print this out of wood PLA? (around about 10% wood to plastic)


It should work great for most parts. I am not sure about the gears or the escapement. It might add extra friction if it leaves a gritty surface.

I tried a cheap brand of wood filament and was not impressed. The PLA around the wood was very stringy and also more likely to clog. Try an experiment on a few parts and see if your brand works better.

It's amazing. But there is a small mistake - number 4, is IV correctly.
I have problems finding the components to buy (metric country). It's a challenge, I'm already printing.


The Roman numeral IIII is technically wrong, but I can only find two instances that use IV. One is the dial of the Great Clock of Westminster (Big Ben). The other is a battery powered clock in my house. Every other instance in my house and several books on antique clocks show a huge preference for the incorrect spelling of IIII.

What parts are the hardest to find? Most of the screws should be interchangeable with a close metric equivalent. The 1/8" stainless steel tube could possibly be replaced with 3mm tube with appropriate sized music wire to fit inside it tightly. Is the small bearing hard to find? I will try to design my next clock to be more metric friendly.

Paradoxically, the bearings were the easiest to get. I bought them in a shop in the street.
Problem was 5/16” brass rod. I use 8 mm brass tube and fill with steel rod. (but it was harder to drill it through), and lead shot for the weight.
Maybe I'll change design for weights.

8mm rod and 5/16" rod are very close in size. Either will work. Your solution is good. The function is to support the winding drum and also to provide a notch for the winding key. The material could be brass, steel, aluminum. As you found out, steel is harder to machine. The rod might be OK using a hardwood dowel (oak or maple, not poplar). I am also trying to design a 3d printed version for my next clock.

Thx. Great project - I made it and it works. I had to use more weight (5kg), due to friction.

This are the only two bearings ?

Yes, only those two sizes. Three large bearings to support the drive weight and two small bearings to support the pendulum.

There is very little force on the other arbors, so they only get bushings.

2x 1/8x3/8x5/32 R2 bearing ? Do you meen

2mm x 1mm /8mm x 3mm/ 8mm x 5mm ?

Here is a part drawing from Mc-Master-Carr.

Really nice clock. I want try to print. I dont understand the numbers of bearings witch do you discribe. 8mm x 5mm x 32mm ?

The 608 bearing is commonly used for skateboard wheels. It has OD=22mm, ID=8mm, and width=7mm. Two bearings support the winding drum and another one is in the weight shell pulley. The entire drive weight runs on these bearings so they are fairly large. They are also really easy to find.

The R2 bearing has imperial sizes ID=1/8" (3.175mm), OD=3/8" (9.525mm), width=5/32" (3.969mm). The pendulum swings on these bearings, so they are small to reduce friction. It is difficult to find this size bearing without seals. If you get sealed bearings, then the seals should be removed to lower the friction. Metal shields in this tiny size are very difficult to remove. Rubber seals are much easier to remove with a small pin. Clean out the grease with solvent after removing the seals.

Good luck with your build.

Well i've finished it, the 2RS bearing kept me waiting a few days as I live in a Metric country and I couldn't find any in stock at local places so ordered online. Did manage to snap the lower right pole that joins the frames together earlier when I was disassembling it to adjust clearances between gears. Substituted some of the bushes and arbours for things I could get locally that are a good fit. Will get some pictures up after I reprint the back frame, it is disassembled now. I was surprised at how freely the gears moved striaght off the printer, after some fine tuning I have to say I wasn't expecting a (mostly) plastic clock to be as good. Only had it running for around and hour before I felt the need to tinker and broke the frame (not a bad thing though as i dont like the two tone, going to do the back in red to so the gears stand out).. i've included a snapshop of it when it was mid-build, will put up a make etc when its back together and been running for a few days..

Congratulations. I like the bright colors. Have you considered two tone for the clock hands? Not sure if white on red or red on white would look better. They print so quickly that you could try both and decide.

Thanks for posting the picture.

It is interesting that the support post broke, considering how large it is. The only function is to define the spacing between the front and back frame, so it would be OK to just glue it back together. All of the weight hangs off the top support. However, if there was a partial clog that caused a weak spot, then there is a risk that the top support is also weak.

Thanks for the reply, a clean break level with the base, it was printed at 60% infill too I believe... strange indeed but maybe some poor bonding or just me flexing it too much.. i figured it would be fine to glue it so long as it didn't upset the spacing but to be honest I wanted to make the rear frame red to match the front after seeing it assembled as I think the gears will stand out more. about the colors.. I love old clocks and watches (i used to service them as a hobby) so for this I wanted to make something that has that modern 3d printer bright viberant feel so i chose a bright color.. actually I already printed a set of 2 tone clock hands the other day, I found that the plain white stood out more but I havent thought about solid white with a red border so will run a set of those off next and compare.. thank you for the suggestion :)

Thanks Steve for the split version of the front and rear frame

Great project. I have started to work on it. I have a few questions. There does not seem to be any bearings for gear 4, minute hand stack. As it goes through the front frame it will be plastic rubbing on plastic and the back will be metal rod on plastic this will have a lot of friction am I missing something.

Second when adjusting the gear fit there is no way in the current design to move the arbor points so how do you adjust the fit. Do you need to sand each tooth if the fit is too tight. I am thinking about redesigning the frame to have an eccentric insert that can be rotated to change the gear separation thus allowing me to fine tune the friction between gears. Do you have the the original design files for the frames so I can make this change without having to reverse engineer the stl files.

Thanks so much for sharing your project.

Yes, that is correct. Gear 4 does not have any bearings or bushings. Adding bushings would require two additional sizes of metal tubing, using only about 1/4" from each piece. This aspect bugged me for a while, but I can't think of a good solution. They are slow moving so I expect that they will last for several years. My clock has made it for several months without any noticeable wear.

I made an assumption about the gears that they will have similar tolerances as the frame if they are printed on the same printer. If the gear is too large, then the frame will also be large by the same amount. The gear teeth were designed with a 48% tooth width, allowing for 2% backlash (or over-extrusion). My initial tests still showed lower binding with increased spacing, so the pivot distances were increased from 1.65" to 1.66". The only sanding was to remove any elephant's foot where the bottom of the gear squishes out on the build plate. Printing the gears at 99.5% or 100.5% size seems easier than adding eccentric cams.

Thanks, I had already printed the gears and the back frame so hopefully I can avoid reprinting them at a different scale.

I saw when test fitting that the gears meshed and worked but with a lot of friction. and the pallet was too close to the escapement. Even though all were printed on the same printer with the same filament. So I traced the frame onto a piece of plywood and drilled out 3/8 inch holes at each pivot point. I then printed some eccentric inserts for the 3/8 holes and test fitted the gears into the wood frame. I then adjusted the inserts until all the gears spun feely. I had enough adjustment in the inserts to go from teeth fully meshed to no mesh at all. The final placements are very close to you placements so I may reduce the eccentricity to have less but finer adjustments.

The end result was I could start all the gears spinning and if I took pressure off the drive wheel the gears would continue spinning for a couple of seconds. However the eccentric does not work well for the pallet adjustment as it moves the horizontal center alignment between the two. So for this I plan to just print an insert with just a vertical adjustment and keep trying until I get the position correct. It only take about ten minutes to print an insert so it is faster than reprinting the whole escapement at different scales. I have made a few wooden clocks and I have not really seen a good way to fine tune the spacing. They just say sand until it works. So I thought this might be a novel approach.

One other benefit is that you do not have to disassemble the frame to adjust a gear. Just pop out the insert and the arbor can be pulled out of the insert hole as the hole is large and runs completely through the frame.

Hi Elmkom,

That is quite interesting and maybe a bit perplexing. Is it possible that your printer is over-extruding? That might make each tooth a bit wider than expected and use up the 2% of backlash. Smaller diameter music wire might allow the gears to push out of the way if they bind.

Cams for adjusting gear spacing does seem like an interesting idea. It could be a bit tricky since each adjustment affects two gear spacings.

Having the ability to take out the gears without disassembling the frame is a great idea. I may modify the design to allow the pallet to be removed to help with debug. On my clock, I was removing the pallet more than any other component. One experiment was to remove the pallet and see how much weight it takes to get the gear train moving. I put a piece of wire on the string (no pulley) and added washers until the escapement started turning. 8 ounces was not quite enough. 8.5 ounces was very reliable.

Keep us posted if you find a solution to the gear binding. Printing a new gear set is probably quicker than printing a new frame with adjustment cams. Hopefully, there is a simpler solution.

Thanks Steve,

I reprinted the frame at 100.7% and the gears spin freely at that setting. When I measured the spacing it was 1.66 before it was 1.65.

I found something interesting by accident that you might want to incorporate into a update. When printing the front frame I had a power outage and the frame was left half printed. But as it turns out the print stopped just before the arbor holes were closed off. So I used the frame to test the gears and I could simply push the arbors through the open holes. I could now just print the second half of the frame from where it stopped and just attach it onto the back frame support pins. So now if I need to do any adjustments I do not need to disassemble the frame just pop off the front faceplate section and pull out any of the now accessible arbors pins.

You may want to separate the front frame so that the assembly and maintenance is easier. You can see a little line between the two halves but you may be able to design a cover flange to hide any visible line.

Thanks for the idea elmkom. I will consider a split frame for the next spin of the clock. A seam in the back frame would be a lot less visible.

I am starting a few experiments for the next clock. One test was to print the gears at 80% scale (25 DP) with 45% tooth width. The gear teeth run very smoothly at the expected spacing. I will experiment with other parameters as well, possibly 40% tooth width and 25 degree pressure angles.


I was thinking the back would be less visible too but the pendulum blocks the back and you would would also need to add some more pegs to clip the too parts together. You would also need to dismount the clock from the wall to access the back. Whereas on the front you would only need to remove the clock hands to remove the face plate. Just thoughts you have already achieved an excellent design and I am sure whatever you decide will be equally as good.

Thank you very much for your time. It has been a great help. Now I can have the necessary parts. You are a genius.

I have problems finding the components to buy. I am Spanish and the translation of google does not give me good results, I can not identify the pieces. It is for me impossible to find the following pieces in local shops or internet: (click pen springs), (keyhole hanger). If you are so kind to be able to guide me, thank you.

Hola, estoy imprimiendo el reloj. ¿Puedes decirme donde has encontrado los ejes metálicos y los tubos? Buscando las referencias que pone Steve, solo los encuentro en webs de USA. Gracias.

The closest metric size tubing would have a 3mm outer diameter and 1.5mm inner diameter. It may be a bit loose, so it may need to be glued into the holes. The inner diameter can probably range from 1mm to 2mm if you select music wire shafts that just fit inside.

The keyhole hanger is often used for hanging picture frames. It allows the clock to be secure on the wall using a single screw. I added a component called keyhole_hanger.stl if you want to try a printed version using at least 50% density. Metal ones would be even stronger if you can find them.

The click pen springs are the springs from a standard ball point pen. I don't know what else to call them, so I added a picture if it helps. You should have several laying around the house.

This is amazing! This clock is beautiful and interesting and the attention to the detail that was put into it is mind-blowingly impressive. The PDF, by itself, is another work of true art! It is so incredibly well written and illustrated (..and charted too!?!) that it has given me the courage to dive head first into this project. I already have most of the parts purchased/scavanged/repurposed and will be prepping them and printing soon. I cannot imagine doing something of this magnitude and releasing it out into the world...for...free. I hope this nabs you the job of your dreams if you don't have it already.

Thanks. I certainly enjoyed designing the clock and also the positive feedback. I don't know if there is enough of a market selling clock plans to quit my day job, so I decided to just release it for everyone. There is an option to tip the designer if someone wants to. :)

Steve again I want to thank you for making this great project available to the public. I finally got my clock running well and it is keeping better time now that it has run for a while. Also I got your package, thanks. I broke a small black piece of the face at about 6:30 due to my rough handling during troubleshooting. I have the piece and plan to glue it back on.

Congratulations on getting your clock working. It looks great.

Great job
8 days runtime and 1 2 mitunes acuracy is a very good result for a 3 D printed Clock

I would like to print it but unfortunally I cant
The dimensions of two parts front and back frame exceede the size of my priter wich is 200x 200 x 200 mm

Since this size of printing machine is very common may be that many other people who wish to print your clock cannot
Can you help me ?
Do you deem possible realize a special version of the front and back frame made for example in two parts to be assembled ?screwed or glued may be connected also by steel rods as stiffners .

That seems like a good idea to split the two largest components. I will post some new files in a few days.

Roberto52, I added split versions of the front and back frames. Each component uses 2" lengths of 6-32 threaded rod to provide some strength. Epoxy the parts together just like the pendulum rod. Checking for alignment will be critical for these components since it helps control the gear spacing.

I cannot get the pendulum to swing well. My question is about the bearings. I got the ones listed in parts list but they are not metric as is the shaft. Do I have the wrong bearings.

You have the right bearings. The design uses a mix of metric and imperial sizes. The bearings have a 0.125" hole, but sometimes the tolerances allow it to be as small as 0.1245". 1/8" steel rod might be as large as 0.126" and would need to be sanded to get it to fit into the bearing.

3mm rod is nominally 0.118" and always fits into the bearing. It also has a bit of wiggle room in case the frame sags or one of the bearings is slightly mis-aligned.

Hi. I'm excited with this watch. I wonder if the brass rod should be solid or hollow.

It should be solid so the end can be slotted for the winding key. Hollow brass might not be strong enough, unless it has thick walls. It could be made from steel or aluminum. I think brass looks best and is relatively easy to machine. Aluminum would also be easy to machine.

Your comment got me wondering if it would be strong enough if it was 3D printed. It might be if it had a center hole for a piece of 3mm steel rod. I may try a remix for this.

Got a question. I printed my clock and have all of the free shafts spinning very freely. If you hold the shafts between your thumb and finger and spin the gear they spin freely and if you tilt the shaft they move along the shaft nicely. However my clock won't run for more than about 15 to 20 seconds when the pallet hits the escapement tip and stops.. I am using about 7 lbs of weight. My question is about gear4_54_18. It has all of the correct measurements along the shaft but how much turning resistance should it have since it is spring loaded. Do I maybe have too much spring tension? Any thoughts. I have done most of the trouble shooting. The pendulum swings for at least 30 seconds with the escapement removed and the weights spin the clock with the pallet removed. I am using superlube sparingly.


It sounds like you are really close. There is a video about wooden clocks https://www.youtube.com/watch?v=o9Tpt_sbzrk that covers most of the concepts to get the printed clock to work. Ignore the parts about waxing the pivots or blowing on the escapement to get it to spin. Almost everything else is relevant, especially the part about balancing the swing. After the clock stops, does the escapement still have power? Test it by manually swinging the pendulum left and right. The escapement should still tick with each swing. If it doesn't tick, then there may be binding or friction in the gear train. If it does tick, then maybe there is too much friction in the pendulum bearings. Each push of the escapement only adds a small amount of energy to the pendulum.

30 seconds of pendulum swing seems a bit low. I tried a quick experiment with two bearings and an Irwin quick-grip clamp to create a simple pendulum. Shielded bearings with the original grease swing about 2 minutes before degrading to a stop. Bearings that I thought were cleaned with solvent lasted about 2.5 minutes. I took off the metal shields and saw that they were still loaded with grease. They were really easy to clean without the shields and swung for over 10 minutes. 52 seconds into this video https://www.youtube.com/watch?v=AXmZhfvH9y0 shows how to remove the seals. Using really clean bearings without shields was the trick that got my last clock working great. I did add dry-lube to the bearings.

gear4_54_18 spring tension is not very critical. It needs to be strong enough to prevent the minute hand from slipping when the clock runs. If it is too strong, the side effect is a bit more friction when setting the time. The range between too strong and too weak is huge. The weight train pushes on the lower gears and does not interact with the spring.

Jimimonet, I did a better test by removing the escapement and testing how long the pendulum would swing in the clock frame. The swing gauge is very helpful. The test started by releasing the pendulum with +/- 5 degrees of swing. It degraded to 4 degrees at 1 minute 40 seconds, 3 degrees at 5 minutes, 2 degrees at 9 minutes, 1 degree at 15 minutes, and 0.5 degrees at 20 minutes.

It seems like you should be focusing on the pendulum friction if you are only getting 30 seconds of swing with a free-running pendulum. Do the bearings feel gritty when you turn them by hand? I buy random bearings on eBay, usually in lots of 8-10. Most of them are good, but sometimes 1-2 of them are a bit rough. Another problem is that the sellers come and go, so I cannot recommend a particular vendor.

Another question, did you run a needle file or drill through the back frame where the pendulum arbor passes through? One of the three clocks I made had a small burr that would stop the clock. Filing away a very tiny amount of material was all it took to double the pendulum swing from 2 degrees to 4 degrees.

Thanks for the advice. It is a little frustrating but I will keep at it until I get it. My pallet bearings do not have shields but I will clean then anyway, they look free of grease. I watched the video and learned a lot.

Thanks Steve I figured that after I posted . Have got 5 gears printed and most of the frame . I split the front and back frame in fusion 360 to fit my printer . All so put location pegs on the cut to give strength to the CA glue joint .

The weight shell will be trying to pull the frame apart, so pegs are a good idea, especially if they are horizontal. Please post a remix if you get a chance.

btw: How big is your printer? I may try to fit a smaller footprint for my next design. Are you able to print the full height of the weight shell? There should enough margin in the design to allow a reduction of about 1" in height if needed.

I love the look of this clock and have started printing . Dose appear though that gear 1 is missing

Gear 1 is the escapement. Everything was labeled 1 through 8 in the initial sketches. "Escapement" is more descriptive than gear 1. All of the other gears kept their original numbers.

Please post progress reports about how it is progressing.

Comments deleted.

Have you posted an updated back frame. I have not printed it yet. Again thanks for all of the great work. I am going to make printer cutting template for the various shafts and tubing. I would be glad to share it with you to include it in your parts if you want.

I just ran a few passes through the hole with a small round file. There might have been a burr, because very little material was removed. The pendulum swing increased from 1.5 degrees to 3.5 degrees.

I think the part number for the McMaster Carr 5/16" brass rod is wrong, should be 8953K138. Not 89535K138 as listed in the parts list.

You are correct, there is a small typo in the part number.

btw: I tested the design and made another working clock. The only change was to slightly enlarge the hole in the back frame where the pallet arbor sticks through. This reduced the friction enough to get around 3.5 degrees of pendulum swing.

I am making another clock to double check the design. So far, I have found a few minor issues.

1) My latest purchase of stainless steel tubing has different tolerances and 1/16" music wire doesn't fit inside it. The solution is to use the next size smaller. In my case it was 0.055" (1.4mm).

2) A couple if thin spacers are missing, so I will add them to the misc_spacers.stl file.


Really nice design! I look forward to printing this and making it my clock that I take to work. Winding it once a week is amazing!

Thank you Steve I have been looking for quite some time for a project like this. It appears quite an elegant solution to what is quite a complex problem Look forward to making it soon when I have a little more spare time

Thank you I also have been looking forward to this for some time.

Nice work Steve, I am looking forward to making this in the near future.