Fully 3D printed gear drive that might not suck

While working on another project I had an idea for attaching 3D printed gears, pulleys, etc… to shafts in a reliable way by taking a small (and cheap) pulley and knocking off the outer flange then modeling the inverse of the pulley teeth into whatever you are trying to attach to the shaft (see picture below). Essentially what this gives you is a compact, cheap, micro-spline connection.

I had been eyeing gear drive drives for a little while and was intrigued, but they were all too rich for my blood, so naturally when I had this idea I decided to make a fully 3D printed gear drive out of it. I knew that people had been playing around with 3D printing the drivetrains, but 3D printing anything that attaches to the motor had always seemed like a fools errand to me (I know, I know, some people kind of got it to work a little bit), so it was mostly just printing the wheel gear (which is still awesome). Well, with my handy dandy new cheapskate micro-spline I decided I was duty bound to try it because science. There are certainly pros and cons to a printed gear drive, but again, science.

I picked up a 15t 3m aluminum pulley which is just barely bigger than an 8mm shaft for like $3 from Aliexpress (https://www.aliexpress.com/item/32786813888.html?spm=a2g0s.9042311.0.0.6cc34c4dmGpk8l) and some Taulman 910 Nylon filament and got to work. I did some rough calculations and got that the splined connection should have a factor of safety on the torque it can withstand of like 8 or something ridiculous like that with the motors we use. I also built a little text fixture to determine how much torque it could take before the connection failed, and the pulley to shaft connection always failed before the spline (I was just using grub screws, so I realize this could have been better). Anyway this gave me enough confidence that this spline would hold up well, and it would just be a question of if the gear teeth could take it or not.

I threw together a test setup to see if there was any hope for the printed motor gear or if it would immediately blow up. Picture to follow (sorry, new account on this forum). Please excuse the ghetto-ness of this setup, but I wanted to test it out with what I had before going whole hog.

The herringbone gears make it a little tricky to assemble and so it requires a split gearbox, so you can see one half o the gearbox in the picture.

Last night I took it for the first test run, which mostly just consisted of me hammering the throttle over and over again up a hill to see if I could break it and the drive took it like a champ. I inspected the gears when I was done and they looked like they were in great shape with no noticeable wear (only about a mile of riding so don’t get to excited). The drive felt pretty great compared to my belt drive with what feels like a lot less resistance, and of course it was basically silent because of the herringbone gears (Although I don’t run FOC so my board still sounds like a spaceship, which I actually quite like).

I’m stoked that this drive made it this far, and it gives me hope that this might actually be viable. Two big questions remain: how long will the motor gear last before needing to be replaced? And when this does fail, how will it fail? As far as the first question goes if the gear can make it a few hundred miles before needing to be replaced that is a win in my book. Yes it is a hassle, but it would probably be a worthwhile trade off for me for the massive cost savings (this whole test cost me <$10 if you pro-rate the cost of the filament for the amount that I actually used). The second question is a bigger deal in my mind, because if the drive fails by throwing a tooth, seizing up, and launching me off the front of my board, that would probably be a deal breaker for me.

Anyway, I have high hopes and am interested to see how this all pans out. If anybody else is feeling adventurous and wants to build a setup for themselves, please go for it (although I don’t really have any CAD files I can easily share because they are highly tailored to my setup).


Here is the picture of my drive:


Very cool my dude. What printer do you use? What’s your setup?


Excellent project, cant wait to see what happens with the testing!


Thanks! I have a prusa MK2S which has been extremely good to me. I ran the taulman 910 at 90C bed temp and 255C hotend. I was worried about warping and other issues that people always seem to have with nylon but it printed with zero issues. I have seen other people say the same, so I wonder if the 910 prints easier than other nylon formulas.

My setup is a custom 8 ply maple deck, caliber trucks, a 192 kv 6374 SK3 motor running at 12s, 11t motor pinion and 55t wheel gear, and 6" penumatics on printed (petg) hubs.


Also very keen to see how this pans out. Well done.

1 Like

yep taulman 910 has way less warping than standard nylon. They put a lot of R&D in it. Also @mishrasubhransu is using it for 3d printed hubs. Great material. For your purpose I would see NylonX or more rigid carbon fiber better otherwise the nylon will skip teeth under load, being so elastic is not the best for your application.


I’m hoping that I would have a lot less ridgidity problems than I might otherwise, as the herringbone gears as ALOT of ridgidity over straight cut gears. Also, in my test ride I was giving full torque over and over again with no skipping problems, so that is promising. It remains to be seen if that keeps up over time. Regardless, something like nylonX was going to be my next move if this doesn’t pan out, although I want sure how well the chopped fiber plays in a geartrain. It seems like it would be overly abrasive.

1 Like

Well if you already tested than good like that. If you have problem you can also try to just replace on gear at the time and see if it is enough. For the rest porject is great and Taulman nylon is an amazing filament.

1 Like

Apparently, polymaker polimide is a lot stronger than alloy 910. 3Dprintinggeneral on YouTube made a video about it. Just watched it last night. It’s $60 for 750gram.

I would think that carbon fiber in nylon x would act like abrasive and won’t fare well in the long run. Pure Nylon by nature is low friction and is perfectly suitable for gears.


Maybe a pure nylon contact surface and a reinforced core for strength?

1 Like

I guess you mean this one https://www.youtube.com/watch?v=2QT4AlRJv1U

I hope the guy will do a comparizon between Polymide CoPA and NylonX

1 Like

Yeah, I just saw that as well and it looks like it could be very promising. One concern is that the Tg for the polymide copa is significantly lower (67C) than for the taulman 910 (90C). The CoPA also quoted some other temperature resistance metric that was much higher, but I am not familiar with that metric and can’t compare it to 910. I don’t know how much heat the gears will experience, but I know they are going to generate some heat, and it doesn’t take too much to get up past 70 C. If the 910 doesn’t perform I might have to give it a go.

1 Like

Alright, time for a testing update:

After the initial hill test which verified that the 3D printed gears could handle full torque, I really wanted to see how they would handle under load at full speed. However, I wasn’t super keen on being on the board to find out in case things went poorly. I found a broken treadmill on a classified ad and used it as a test rig. I clamped the board to the treadmill and had the board drive the treadmill (the treadmill was off) which provided the load for the board (see picture).

This test setup actually worked really well and was able to provide quite a high load for the board (at full throttle it took the board several seconds to accelerate to full speed). After a little while of testing and a few ramp ups to full speed at full load the motor pinion failed right when it reached max speed:

There is both good news and bad news here. The bad news is obviously that the gear failed, the good news is:

  1. The gear didn’t fail immediately (on the first try). Based on gut feel that could be totally wrong, I hope this means that this is close to the limit of what the geartrain can handle, and not simply well beyond it.

  2. When the gear failed it did so in a relatively benign way. rather than lock up and come to a screeching halt, the wheel kept turning, but with a very audible noise and increased resistance. The pinon gear was still even able to drive the wheel to some degree although it definitely didn’t like it. It didn’t seem like there was any violent jerk when it failed either.

Even though the gear failed I feel like this warrants further testing. Firstly, I was (intentionally) pushing the drivetrain much harder that I would under normal riding conditions. At least for the way I right I am never applying anywhere close to full throttle when I am going fast, and I hardly ever go at full speed. If my normal riding conditions are backed off from the test conditions a good deal, and the drive was on the edge of failure at the test conditions, I feel like there is fair chance that it will be fine under normal riding conditions. Secondly, I now know that when the drive fails it does so in a non-streetface-inducing way, which makes me a lot more willing to ride it. Also, interestingly enough the wheel gear looks like it is in perfect condition, I could probably just replace the motor pinion and keep going (although I am currently re-printing both gears right now for a fresh start). Again, the goal here is not a drive that lasts forever, but if I could get a few hundred miles out of each motor pinion that is good enough for me.

Again, if after extended testing under normal riding conditions this doesn’t pan out, I may still try this with other nylon formulations to see if it can work.

I’d love to hear your guys thoughts.


Great start! Good on you for testing for high stress conditions, depending on how much greater they are (factor of 2? 3?) this sounds like a good indicator of what they can handle. It sounds like they can do some low speed driving at least. Now, questions:

  1. What exactly were your test parameters? How fast/slow did you run it? How much total runtime?
    How much simulated weight?

  2. Are the printed teeth straight edged or rounded? Rounded edges will probably help the teeth engage better and prevent the smashed teeth I see in the second photo.

Really nice test set up! On the big size but really nice. You do what most of the company is not doing, nice! For the gears by the colour of the printed gears looks like PETG. If thst is the case you need to step up. I would aoggest to go to Nylon or similar. Search for a strong Nylon. I am recently test nylon gear for my CNC mill. I print it at the highest temp the filament allows and then boil it for 30 minutes and leave it cool down in the water. There are several. Articles online about it. For the moment I am using eSUN Nylon, since I already bought it but for instance Polymide CoPa is way stronger. That would be my next choice in case of failure. So far the gears is working well

Here some more printing parameters


Sorry I haven’t gotten back here in a minute. Been busy with life/building/riding.

@High-roller :

  1. Test parameters: Ran it all the way from stopped to full speed (192 kv motor at 49.2 volts - sag, so roughly 8.8 krpm) @ full torque. total runtime was not that much, probably around 5 minutes. I went straight to max conditions, and I really should have done a stepped approach. Maybe next time. I did~3-4 runs up to full speed before it broke. Simulated weight doesn’t really make sense here, since the board is not moving the weight on the board has no effect on the gear train. I will say that the simulated “load” on the geartrain was high, it took >10s to get up to full speed which is at least twice as long as it takes to accelerate me (110 kg) to full speed, so pretty high load.

  2. The printed teeth use an involute profile.


Thanks! Not PETG, Taulman 910. I printed PTEG gears at first just for fit-up, and rode it just for kicks and giggles, and the gears shredded in < 30 sec. I’ve definitely been amazed at what nylon can do. I’ve been looking at the Polymide CoPa as another option. Much higher tensile strength, but I dont think it is as hard as the 910 which is alloyed with Polycarbonate. The CoPa also has a lower Tg, so I don’t know how all these factors would balance out, higher tensile strength for lower hardness and Tg. All these things are important for a gear train, so the only way to find out which one wins might be to just try both. CoPa is pretty expensive though (like $65 for a 0.75 KG roll) so that kind of sucks…

Testing Update:

The treadmill test gave me the courage to do more on-board testing. I printed a new set of gears and got riding! For the most part I rode pretty normally, although I will say that I was a little more timid than normal. I still did plenty of hard accels and breaking, and I took it up to almost full speed (~24 mph, max speed should be around 28 mph) on a few occasions. I checked the gears after the first ride (~5 miles) and they looked pristine. They rode and felt fantastic and didn’t seem to have any hiccups or issues for about 25 cumulative miles.

At the end of my last ride I failed the gears once again, although this time it was under very different conditions. I had ridden about 6 miles (not my longest ride on these gears, which was ~11 miles), and was accelerating (not full accel, mayber 1/3 of max) up a small incline at a slow speed (10 mph?) when the motor pinion failed again. Once again the pinon failed in a very benign way, no jerking or jolting, just kept on rolling but you can hear the difference in the gears immediately, so that is very good from a safety standpoint.

It seemed super weird to me that the gears failed this time at such low-stress conditions after I had been riding them much harder than this for the previous 25 miles, and on the treadmill they only failed at extreme conditions. As I thought about it I could only think of a few possible reasons that they could have failed at such a low stress condition:

  1. Overheating - it could be possible that the Gears heated up beyond the Tg of the Taulman 910 (82C), but I don’t really think this could be the case. The Gears do get quite warm to the touch, but at 82C they would give me 3rd degree burns in < 1sec. It is possible that they were getting localized heating this high, but I really doubt it.

  2. Fatigue failure - The teeth could have weakened due to fatigue to the point where they could fail under low stress. I also think this is unlikely, as ductile materials tend not to fail in fatigue, and Nylon is extremely ductile and tough. I also don’t think this is the cause because (and I am no expert on material failure analysis) there was no cracked or broken teeth, just deformed teeth, and if it was fatigue failure I would think I would see broken, or at least cracked teeth.

  3. Tooth wear - The face of the teeth could have worn to the point that they got too weak and deformed. It also seems unlikely that under proper lubrication (see next point) the teeth would have worn this quickly. I feel like this conclusion might be a little more tenuous, but when I examined that teeth after 5 miles (which I realize is not much), they looked pristine, and I could see no signs of wear whatsoever. To me it seems unlikely that they would have worn much after 25 miles based off of what I saw after 5 miles, although it is possible that they could fail after just a tiny bit of wear, but I really hope that is not the case.

  4. Improper lubrication - My bet is on this one. The lubricant that I was using for the gears is called super lube. I picked this because it was easy to source and is a silicone based lubricant with PTFE add-ins, so there are no compatibility issues with nylon. The downside to this lubricant as I found out is that it doesn’t seem nearly viscous enough a gear drive. It probably has the consistency of like a triple antibiotic ointment. I put lots of lubricant on the gears, but everytime I would open up the gearbox, there was a ton of lubricant stuck to the sides of the gearbox that looked like it had just spun off of the motor pinion. My theory is that this ended up leaving very little lubrication on the actual teeth, which quickly got worn off in the 25 miles that I rode, and left the teeth with no protection which led to premature failure. This is supported by the fact that when I opened up the housing after the gears failed and inspected the gear teeth (on the wheel gear which was still intact) the contact surfaces of the teeth were almost completely dry.

I think the next thing to do here is to up my grease game. At the very least I think this is the next thing to try, if I try out better grease and they still fail quickly then we’ll know it wasn’t that. I’ve done a little bit of research trying to find better suited grease that is compatible with the nylon, but haven’t found anything yet. I am curious to know what most of the gear drives right now use. I know most of them don’t use nylon gears, but they are using POM gears, and the grease compatibility might be similar. @moon I think your 3G drive is also using a nylon gear? Were you able to find a good grease for it? Can anyone else point me in the right direction for better grease?


Absolutely true. I am also try to decide what to pick it up. But if you check the video on you tube with a direct comparison 910bvs CoPa is seems CoPa is also harder.
@DAddYE got the CoPa. Did you run some tests?

But if you check the video on you tube with a direct comparison 910bvs CoPa is seems CoPa is also harder.

Is this the video you were talking about? I can’t see anywhere in the video where it talks about the hardness of the materials.

Well by the definition:
Material hardness is the property of the material which enables it to resist plastic deformation, usually by penetration or by indentation. The term of hardness is also referred to stiffness or temper, or to resistance to bending, scratching, abrasion, or cutting.

There are different way to test it. Deformation is one of them. So looks like CoPa is harder to deform than 910. Do you specifically search for incision test?