The Chronicles of my Moon Phase Clock Build

Super cool. That would explain the small bump in Moon Phase Clock sales the past 2 days. There were 5 moon phase sales and only 2-3 of everything else. Thanks for posting it.

I should spend more time on social media. My wife says I should post my clocks on tick-tock. :)

Sweet. I expected that sales bump to happen. My pleasure posting it.

😂 TikTok. Oh my gosh, that is so hilarious . Your wife has a good sense of humor.

Thank you sir.

Those are really cool. They would be great to change out with the seasons or to put daily quotes.

Know it all to well. Or did, lol.

Wow! That looks beautiful Phil. Good work. You were right, the contrast in black and white is amazing. Really nice photo too. Your moon looks like it came out with lots of contrast between the dark and white areas; looks great.

I agree. The black and white looks stunning.

I have a "man shed" in the crawl space under the house for woodworking equipment and a CNC router. It gets hot in the summer and cold in the winter. Plus, I have to worry about noise limits at night and on Sundays.

3D printers get to live inside the house. As long as they are not on the kitchen table or right next to the bed, they can be used all the time.

OK, this is lengthy. But it contains a lot of what I wanted to talk about in my new future post. So I'll just start here.

I think, for a 3D printed clock anyways, that your shallower impulse angle is in the direction of goodness. Considering all of the friction in these printed gears moving up the going train, and I have tested this clock with gears removed to feel it, there isn't a whole lot of torque left available in the escape wheel to push the pendulum much. Unless of course you want to stack a ton of drive weight on it. A shallow angle with a low pendulum amplitude might be best for these 3D printed mechanisms. I think I have about 2 hours total now just staring at the escapement. It's quite hypnotic. But you can learn so much by just watching. For now, until I learn more about it, your design looks good.

In case you are interested, and obviously it doesn't matter since my clock is running just incredibly well, I have noticed a double tap (bounce) on the entrance side (right) during lock. The escape tooth does a tiny little double tap on locking. The left, exit side does not do that at all. Balancing the beat doesn't change that. So I went on a research run.

What I think I discovered, but I have to read more to be sure, is that bounce is affected by the span of the pallets. The pallet pivot offset controls the amount of locking area of contact, or depth of engagement during lock, while the span between the pallets controls the amount of drop. It seems balancing the amount of drop between entrance and exit is ideal, but not necessary. Energy lost on one side is gained on the other. And the energy lost to the bounce I guess can be neglected.

For your beautiful design, and what really took my build to the next level, is attention to the quality of the working escapement surfaces; the locking and impulse areas of contact. That, and a well rotationally balanced escapement wheel. Observing our escapement, after initial locking, the escape tooth continues to slide up the pallet as the pendulum reaches it's maximum travel. From what I read, this is the whole purpose of the Graham design to eliminate recoil. Then the tooth essentially slides back down the pallet surface, and finally begins to exit with hopefully the impulse surface in full contact during the exit. Soooooo, as far as 3D printed surfaces in contact are concerned, where tiny, tiny forces really matter, in a single beat, we have 3D printed surfaces first slamming into each other at lock, then rubbing and sliding going up to pendulum max, then sliding back down during the approach to the impulse stage, then sliding and rubbing together again during impulse and final exit.

With all that said, the printing of a high quality escapement wheel is essential for the greatest efficiency, and the area that I personally struggled in to get it to run on your 7.5 lbs or less challenge, lol. Don't get me wrong, the clock will run fine on say 8.5 lbs with even a shabby printing job. But if one's goal, like mine was, is to build the most efficient, well tuned clock that one's personal skills and equipment will allow, then the tiny details really matter. This all only applies to the escapement wheel, as I think losses due to printing imperfections in any of the other gears are in the noise, because there is so much torque available, even all the way up to gear 2. But that one extra gear reduction moving from gear 2 to gear 1 is a game changer, lol. That final 4.5 : 1 reduction makes everything going on at the escapement super critical if your goal is to really dial it in.

The issue I had was with the consequences of elephant foot on the bottom (as printed) of the escapement wheel. One can employ elephant foot compensation, but that only applies to the very first layer only, and if one's first layer adhesion is not optimal, it can result in a lifted and failed gear print. Twice for me on gear 1 for some reason. Trying to produce a tight bonding first layer, for me, resulted in excess elephant foot. Which was hard to eliminate. A bit of elephant foot on the other gears had no affect. But on gear 1, not so. I ended up taking the time to manually debur all the escapement teeth with a hobby knife and magnification. And just going over every gear to be sure there where no imperfections in any of the locking and impulse surfaces. After doing that, and balancing the wheel, which I'll talk about in another post, all was good. Even with successful elephant foot compensation, I feel adhesion to the print bed in general causes the first several layers of the print to be deformed, which can affect the geometry of the critical working surfaces of the escapement teeth. But there is a simple fix.

My suggestion was going to be to consider adding say a 1 mm chamfer to the bottom (print side) of the escapement wheel (gear 1) to eliminate any dependence on the operator or the 3D printer to get this part right. A small chamfer would not affect the escapement dynamics at all, and would guarantee that the working contact surfaces are nice and flat, because the design tooth profile would start to be laid down a number of layers up from the bed after the chamfer was formed. The same could be done to the pallet, but deburring the ends on it is only 2 each, so no big deal there. The top surface of the gear teeth is probably OK, at least on my printer, but depending on one's nozzle and other factors, one can have final layer ironing effects that can cause a minor elephant foot like print on that top layer. So a small chamfer, maybe like .3 mm on the top may be helpful as well to eliminate that possibility.

Let me know what you think of all this. I hope it is helpful.

Good point about the elephant foot. Maybe I am spoiled, because I rarely see any elephant foot on the gears. For some reason, the frame often has a tiny bit, but this never matters.

I typically adjust the slicer compensation up to 0.25mm elephant foot compensation. This is mainly set to eliminate any infill on the first layer. The first layer prints with two slightly wider than normal passes and the second layer has three passes.

The escapement and pallet could have built-in correction added. 0.5mm should be enough to affect the first three layers. It would only be applied to the one short sidewall and the active surface. All other surfaces have no effect if there is some elephant foot, so no need to compensate. If you have to clean up by hand, then only the surfaces that actually touch would need to be fixed. This cuts the effort in half.

The first assembly guide had a section about cleaning up every gear using a fine tooth hand file. It usually took less than an hour to clean up every gear in the entire clock.

I am always on the lookout for ways to improve the reliability of the clocks. In my experience, the most common reason clocks stop, after fixing the obvious tight shafts and frame pinching, is gear engagement friction. The escapement suddenly stops spinning when the gears line up badly. It doesn't matter how efficient the escapement is at that point. The problem appears to be worse in 3D printed gears. I am always for ways to improve this.

Yeah, it's mostly a limitation on my end I think. I have old school, smooth PEI print surfaces, not the newer, textured, flexible, magnetic print surfaces. Good, reliable adhesion can be tricky sometimes. It's temperamental. Sometimes it sticks great, others not so great. And I hate using the glue stick stuff. I need to upgrade both of my printer's hot beds to the modern magnetic, flexible surface tech.

Thank you. It was a lot of fun finding places on the model parts to add some ornamentation. I guess for me I can never let well enough alone. Always have to change stuff around to feel like I done something, lol.

Looking forward to seeing your build. Thanks again for the comments.

Cheers!

Wow! Absolutely stunning additions to the clock!!! And great looking colors! Well done!

Keep it running with the slightly larger weight you are using. Most of the clock books recommend testing the clock to find the minimum weight needed to keep the clock running, then increase the value by 30-50% to have a good safety margin. A bit extra never hurts. It also helps keep the clock running if the frame tilts a bit when winding. I usually find that the most common time for my clocks to stop is about 20 minutes after winding. It usually means the clock has been tilted and the beat is slightly off.

If you are looking for your next clock, the 32 day easy build clock is about to get a refresh. 32 day mode is running on my clock with 7.5 pounds of weight. It needed the new gear profile to make the clock run reliably in 32 day mode. Wait until the update is released in about 2 weeks before starting the print.

Thanks for the comments Steve. I love this clock. Such an amazing design. OK then, the 7lbs 4 oz puts me just at about 30% more than my min. Should I bump that up a bit? My amplitude is currently 2.4°. Is that enough for reliable running? It's been over 24 hours running now without skipping a beat, and the time is keeping good, as good as only 24 hours can show for now.

I will keep an eye out for the new release of the Easy Build. You mentioned you were working on wooden designs. I also have a laser cutter that I built that can do 1/4" ply. I've been doing some research to see if my laser capabilities would allow me to build a wooden clock. For thicker than 1/4" parts, I would have to stack and glue, for example 2X ply to get 1/2" thick. I do have a small, desktop CNC, but it's capabilities are very limiting. May have to try a gear on it just to see if it can. But a 3D printed clock next I can do as well.

I still would like to talk about friction in general, but I will open a new thread in the General category for that soon. And finally, I have several constructive suggestions for improvement on this build, but need to check with you on how to tactically discuss those. Not sure about your stand point on using this open forum for that.

Cheers! Will try and get video up soon. File size is being a struggle.

30% upsize seems good. Check the beat after winding it and it should be fine. If it works, don't fix it. :)

I have been wanting to do a wooden clock for several years. There has always been a long list of 3D printed clocks getting in the way. I am currently refreshing several of the older clocks with a new lower friction gear tooth profile. Feel free to create a new topic about friction reduction or other improvements. We can all learn from open discussions.

Some of the assembly guides have pictures of sample solid wood gears. That is as far as the wooden clock design has progressed. Eventually, the designs should be applicable for plywood gears cut on a CNC, scroll saw, or laser. 1/4" (6mm) would be the typical thickness for the main gears and stacked to 1/2" (12mm) for the pinions. Hopefully, the frames can be simple flat stock, other than the dial that should be a bit fancier.

Sorry about the file size limit. If you want a confirmation on the escapement performance, then a simple closeup of just the escapement for a few seconds is sufficient. There is no real need to pan around the whole clock. That can be a separate video. Your static images already show lots of great detail there. YouTube allows much larger file sizes for the full clock video. Some people post links to a shared Flicker or Google drive for large files.

Steve

Thanks for the comments Steve. Agreed on the bearings where looser tolerance is better here. Trade tighter tolerances for looser, lower friction performance.

On the gear drill out, the fact that the central part is looser except for the ends of the bore is exactly what can cause issues if one's printer doesn't print along a perfect vertical axis on taller parts, or if the vertical axis shifts ever so slightly at higher Z's. Even though one might drill out the ends to slip fit, the two ends can still be out of axial alignment. Think of it as the top opening shifted slightly along X/Y compared to bottom opening while printing. Even though the arbor slip fits easily from either end, a straight arbor passing through those misaligned ends will cause friction. That is what happened to my gear 1, and only appeared on the longest of gear bores.

And I absolutely agree on the tri-color. I don't like it. It's to difficult to get consistent coloration, especially keeping the filament from rotating back and forth during a print. One would need perfectly aligned extruder drive pinion and idler, and maybe a more consistent filament delivery system. I've made great improvements, but still find it annoying. When I do it again, it will be bi-color.

Question I forgot to ask. I'm sure I'm good, but better make sure. I may have missed it, but it wasn't clear to me in the instructions. I concluded that Spacer 7a goes under Gear 7, but guessed if it went with the top hat up against the gear or top hat down against the frame. It made more sense to go with top hat up against the gear, so that is what I did.

Cheers!

A crooked gear shaft might cause the gear to wobble. Not sure if the misalignment was caused by the drilling process or a printer issue. I usually hand drill every hole using a pin vise so the drill bit follows the center hole. A slight amount of misalignment at the beginning usually corrects itself. A very low speed drill press might work better to keep the holes vertical.

There is a fairly large side tolerance on most gears, usually 0.15" or 0.2" before it becomes an issue. Any wobble under 0.1" at the teeth should be OK as long as there is no sidewall rubbing. You will have to determine if the gear needs to be re-printed.

The "hat brim" on spacer 7a can go either direction. There is a minuscule amount of support added to gear 7 if the brim is next to the gear. Some of the other clocks build the spacer into the back frame so there is no confusion.

MatterHackers dual color PLA specifies that the color is guaranteed not to twist on the spool. They ship with clear spools and the alignment looks perfect. I use Prusa printers and don't see any twisting between the spool and the part. Printers with long Bowden tubes might see more twisting. The slicer could have a big impact if it prints part of the object clockwise and another part anticlockwise. PrusaSlicer does a decent job maintaining a consistent direction.

Steve

Yeah, not sure on where the friction culprit actually was. That gear 1 I did print twice. The 1st had practically zero arbor friction, but the gear had lifted off the bed a bit and didn't come out good. The 2nd gear 1 printed perfect, but had the slight arbor friction issues. But I got it sorted out either way. I have been drilling with a pin vise by hand first. I only resorted to the drill when it needed I bit more finagling.

Thanks for the filament tips.

Hey there,

That's great. The parts pack is most definitely the way to go. I am exactly where you are at progress wise. I myself am just getting back from the holiday visits, and am just getting back to the clock project. I have all my parts printed. Need to do my post-printing clean-up of the parts and get them all ready for assembly. I have added several special touches. Will get some pictures up tomorrow hopefully.

Cheers and Happy New Year!

Thanks Phil. And you have a great Chrissy as well. No printing for me for a few days.

Thanks Phil.

Lol, yeah. It's the engineer in some of us. Just can't leave well enough alone. Problem is, doesn't always end up "better".

Thanks Phil.

No worries at all Steve. I really appreciate the effort. Don't spend another second on it. Your experience is mine as well, in general, with any reasonable attempt to convert STL to solid geometry. My fall back approach with situations like this is simple. I can print the gear and star field with no moons, print the 3D moons separately as disks, and bond those to the gear. I use that approach often, especially with multi-color parts (with no MMU).

Thanks again.

TurboCAD usually does a decent job converting surface models to solid. It falls apart when the images start as NASA quality high resolution images. All the moon images I found were either super high resolution or cartoonish. I started with the better quality images and had to fiddle with them to generate STL files that worked. Your solution seems perfectly acceptable. There is plenty of glue surface to hold everything together. The pendulum bob is another good candidate, since it needs to be printed with the front surface on the build plate.

I went on a small mission to attempt to determine once and for all if converting an STL file to a usable solid model is even reasonably feasible, and the answer is No. At least not affordably. There is a program called Rhinoceros 3D that supposedly does a pretty good job, but it's $1000 for a single seat. I use Fusion 360 for my CAD, and like most solid modeling apps, they use NURBS to define the geometry. Which is an equation based, parametric approach. While STL is a bunch of triangles with a list of coordinates specifying how they are all connected. Apparently, it is very difficult to determine the NURBS equivalent of an STL.

After a couple of days of experimenting, I was able to finally get the 3D moon into CAD as a recognizable solid, but even then, it still is not a true NURBS solid. It was still a tessellated mess. Bottom line, it is a much better approach to modify STL in a mesh based environment, and solids in NURBS. Forget about mixing the two easily. They are just two vastly different beasts. For editing STL's it's best to stay in the mesh environment.

I have been using Meshmixer, which is quite capable. Although I'm not that proficient with mesh mixing, I think with some practice, everything that one would want to do to change an stl can be done as a mesh edit. No need to convert to solids to shoehorn the editing into a more familiar environment. I just need to get better at Meshmixer and it's concepts.

Anyways, thought I'd share what I learned.

Cheers!

Hey Phil,

Good for you, and great choice on the SP14. It's beautiful. It's large. And I feel it has all of Steve's design evolution all built into this masterpiece of a design. Honestly, as a "Design for 3D Printing" guy myself, I am learning and appreciating some of the techniques Steve uses in his modeling to make continuous filament path walls, and feature geometry that avoids supports. None, anywhere in his entire design. Impressive!

The black & white will look awesome. Just something about the simplicity of the colors I think will do wonders. I think I saw some Zebra Stripe B&W filament out now, and maybe some Grey Scale stuff. All kinds of new filaments coming out that in B&W might make for some very interesting parts.

Myself, I'm still printing parts. But slowly all of a sudden because I'm printing gifts for my grandchildren, lol. Hopefully back up full time after the holidays. Good luck with your build, and keep the pictures coming.

Cheers!

I have no idea why there is only a flat moon step file. I will generate the file when I get a chance. It might take a day or two.

Oh right on. Thank you so much. No immediate hurry since I have plenty of parts to print still. But certainly happy to know it can be made available. I think you will like the remixing I'm doing. Maybe could make it available to others after it's finished, if you all like it.

Thanks again.