Printing a 404:1 Planetary

The decision have been made the deadline set so now its up to me to print what has been designed at school as I am the one with the 3d printer this is the printing of the past 36 hours or so

namely the ring gears and sun gears. One sun gear is wrong so they will have to mirror it and I will reprint it.

Here by the numbers some technical info there are 3 sections section 1 and 3 are the inputs with 1 CIM motor each. The center stage is the output which is taken off on the big black ring the reason its 2 colors - its whats on hand. Material is Bridge Nylon - again its whats on hand. The design is full involute double helical. Mounting the rings is going to be “fun” as they need to be squeezed and distorted to slip over the planets - you can do that with nylon then each ring gets an outer ring out of something hard like HIPS to keep it in a nice round shape. There are 4 planet stacks which are connected there are going to be bearings on the planet carriers and we will use a 5/16 6 in bolt as the axle.

All gears are designed with a 14.5 deg pressure angle and a 30 deg. Helix.

Set 1 and 3: Module 1.50185mm (to match the center distances of the center set
Sun 23 tooth, Planets 31 tooth, ring 85 tooth no unit correction/tooth shift beyond what is required for them being helicals

Center set (2) Module 1.5mm Sun 22,Planet 31 and ring 86. All (sun, planet, ring) have a unit correction/tooth shift to allow the ring to be 86 instead of 84 teeth (84 would not allow evenly spaced planets like the 23,31,85 requires) Each planet set is unique with the center planet rotated based on its position by 0, 2.903. 5.806, 8.71 deg.

Stage 1 and 3 is 29mm and stage 2 is 58mm they are separated by 4mm each.

Goal is to lift about 400lb up a flight of stairs.

Well we will see - its either going to be a triumph or a lesson that things don’t always work out as designed on paper. Either way a lot of learning already went on and is going on so I see it primarily as educational value either way. If it works I will post the cad and probably a video of it in action. Well that is if we can stay within budget (parts in house plus about 100 bucks) to do it. Hence the Bridge Nylon ($36/kg) and the HIPS ($12/kg) CIM motors that will be used were donated by Team 1533


When are you 3d printing a cim?


I highly doubt that the CIMs will be 3D printed. It’s very difficult to get plastic to conduct significant amounts of electricity consistently. It’s even more difficult to 3D-print metal–though it is doable, it’s usually done using either Selective Laser Sintering or Laser Deposition processes, which are significantly more expensive (and hazardous–did I mention lasers and metal dust?) than most of the standard printers used in FRC. We’re talking $$$$$$ machine, $$$-$$$$ material compared to $$$$ machine, $$ material.

1 Like

I’d be interested in hearing the maximum torque before failure at the output. Printed gear teeth don’t tend to be strong enough to stall out the motor at the input in these sorts of designs.

If I gave you the impression we are printing CIMs I apologize. We are using CIM’s to power the gearbox. The Gearbox is 3D printed

Will try to get that number for you even though we reoetedly stalled a CIM and also a Gearmotor (71:1 one IDK the # atm) with 3d Printed Gears. But never 2. It depends on the size of the teeth and the material.

1st if you use double helical involute gears. You have more than 1 Tooth being engaged at any given time. This even more so if yo mage the gear wide hence the planet assembly is 124mm thick. Then the bigger the module the more “meat” you have in the tooth. And then we use Nylon which is tough. A slight drawback of Nylon over metal is that its also quite flexible so there is the chance that the tooth bends, skips and resumes shape - IOW snaps into place in a slightly different position. Hence the planet carrier which in theory would not be necessary on a design like that but in praxis will happen and kept this gearbox (a 12:1 prototype from being certified for the robot

We stalled that one repetedly to simulate the robot getting into a shoving match or trying to drive through a wall and after a while the planets skipped. None of the teeths were damaged. but the planets became misaligned and the gearbox jammed. Hence we did not use it in that configuration. We also know through tests that nylon - if printed and designed correctly will hold the 2mm key that is in the CIM (You print a copy of the 8mm to 1/2in hex adapters that andymark sells for the CIMs put it on the motor and put it in a Vice and it will start the motor. Its a short test as the wires get hot quickly and so does the motor and you don’t want to burn it out. Same for the output of a Gearmotor. Now PLA will shatter. All gears are designed to printed solid by the use of perimeters. That is why some bigger gears have holes to simulate infill and reroute the plastic to where we want it solid
Like this gear which will move and hold well over 100 lb (module 1.4) and was part of the inspiration for the gear sizes on the current gearbox. As this one was rated at 100lb (we needed about 75 and stopped at 100 because of problems with the test rig) the current box is 1.5 module cause we had the room and 4x as thick based on if 14mm will give us 100 lb then 58 (again little more room) should give us 400. But then we will see what happens in praxis. We had problems with that particular gear due to shaft deflection which was fixed by slightly rearranging the size and geometry and going from a 25 deg. Pressure angle to a 14.5 degree pressure angle with a slight undercut which kept the shaft deflection low enough to not skip. That is a reason too why this current gearbox uses 5/16 machine bolts out of steel as axles and not churros. One thing that is worrying us on this gearbox is weight as each planet with bearings, plastic and bolts will weigh about 400 grams (about 1 lb ) and they will be spinning at 800-1000 rpm so we might get some vibration from it so we are taking great care to make everything as symmetrical as possible. So this is the first planet stack fresh of the printer

We will see if we can clean it up enough or it might have to be reprinted its 3 planets 4mm spaced apart and 240 grams of Nylon This one is not shifted. Eache stack is unique as the center gear is offset about 2.9 deg, Now I am sure this would be quite difficult to machine out ouf steel but then that’s the power of 3d printing

Oh we would love to have one of those $$$-$$$$ machines lol No we are not printing Cims. We got them donated by team 1533 for that project. And its all done on a $1200 machine out of good old Bridge Nylon

This is very interesting so I took a look at the JVN calculator using your inputs. Here’s my math for you to take a look at:

Using the “Linear Mechanism” tab, CIM motor, 2 per gearbox with an efficiency of 50% (generous for plastic 3d printed).

Travel Distance: Standard slope of a flight of stairs is 37 degrees, with a vertical height of 12 feet. Hypotenuse of this is 20ish feet, or 240 inches.

Applied Load: Using that same 37degree angle and a weight of 400 lbs, the load to keep the weight in place is ~241lbs. However that’s assuming you have 0 friction, which is impossible. Assuming a friction factor of 0.41 for linoleum, the total load (when moving) is now ~372lbs.

Pulley Diameter: I don’t know what you’ll be using, so I kept it at 2 inches.

Gear Ratios: You have a 2-stage 404:1 gear ratio, so it looks like you have a 101:1 and a 4:1.

With all these factors, it looks like it’ll take you 181.64 seconds (3 minutes) to lift your weight up a flight of stairs. Certainly doable with your setup, however the real question is your gearbox’s durability.

You’ll have to get the weight moving first, and you know that Static Friction>Kinetic Friction. So your gearbox will see an increased load at the beginning of the trial, but with an expected stall load of 8617lbs it shouldn’t be that rough.

The main issue will be running for 3 minutes straight. You can reduce this time to under a minute if you go with a 10in pulley, but then your gearbox will see more load. It’ll be up to you to make that choice, so you can either go with a much smaller load for a longer time or a higher load for a third of the time.

It’ll also be up to you to design a solid winch system so the gearbox doesn’t see too much additional force. If you put the pulley directly on the gearbox shaft with the other shaft end unsupported, it’ll most likely fail. Part of this challenge will be to prevent failure through this mode.

All in all it looks like a fun project with a reasonable goal. Good luck, and I’m looking forward to seeing the results!

Right now no pulleys the student wants to drive it up the stairs - immagine either a 8 in gear like wheel coming out the center ring or a couple of sticks in a mechanism to grab the stairs with a stabilizing sledge /track setup. This is still being bounced around a bit but might end up the center ring and another one upfront - basically a tank driving up the stairs. We have some excited HS kids and its kinda how big a gearbox can we print and how much weight can we haul up the stairs. And I am ok with that as learning goes on

If you look at the gearbox the first ring is 85 teeth and the 2nd 86 The first/3rd stage is 23/31/85 so the planet carrier moves at 85+23/23 = 4.7:1 the 2nd stage is calculated R2/R2-R1 which is a 86:1 As the R2 has one tooth more each center planet has to be turned by 1/4 tooth and hence they are 31 tooth planets by 360/124 deg. Depending on load we expect the center ring to spin at about 6 to 8 rpm depending on load. Initially we were talking of about at 1000 to 1 or 800 to 1 gearbox but that fell away as we couldn’t print one that big on our printer or would have had to use a smaller module which can’t hold the load or make it taller which would have cost more $$ than we have. So this is the biggest gearbox we can do within the parameters and means currently available to us. So the most of the torque gets transferred through the planet stack between Ring 1&3 and Ring2 which is the output. So what worries me is less a tooth breaking but the layer adhesion between the planets letting go. So we are printing a little hotter and slower to give it the best chance.

Alright I don’t know if I was clear but I was joking, I do want to see as much of a 3d printed robot as possible👍.


This is very cool! It reminds me of a try and fail attempt of making a differential with 3D printed spiral bevel gears. I used an online gear tooth calculator to help try to work out the stress and strain requirements. For me it was the strain (deflection) that I wasn’t prepared to factor in. The plastic would give way and the differential would slip even though the stress could be handled fine, it ended up not being a practical solution. After this experience, my goal was to learn and predict what material is best suited for the load requirements, which then led me to learn how to use a CNC machine and later found others on the cnc forum, and like me came from a 3D printing background.

For lighter loads, 3D printed planetary solutions work great. I used Vex EDR gears and then 3D print the outer ring gear. Our 3D printers have a bit of fitting to get right though… Through trial and error I would tweak the diametral pitch (very small amount), so the ring would have the proper fit. So in a light load case, it would work great.

Oh, one other thing… A problem also occurred when the T6 1/2 hex shaft would apply load on a 3D printed bore for it. To address this, I used a versa hub, so when the same material was used to accommodate the shaft it could successfully delegate the load to the spokes… and then the 3D print could manage that just fine. My gears were mounted on the versa hubs… but even with this setup… the deflection on the teeth happened, and hence the saying… “a chain is only as strong as the weakest link.”

1/2 in hex Nylon can handle if very tight press fit. 3/8 not so much but we also had 3/8" hex eat out an aluminum hub so we don’t do 3/8" hex anymore. In our experienc plastic on plastic (Nylon) works fine Plastic on metal not so much as the metal tends to eat the plastic

I have checked out I’m curious if this is the filament you are using. The thing which has caught me by surprise is the Modulus “PSI” when 3D printed is 26,544. Now I know where I messed up in my calculations… if I use Solidworks Elastic Modulus defaults for PET and ABS, they are higher than this. I will research this a bit more later I think printed modulus is lower because of the amount of gap between the filament (e.g. 100% infill).

Yes that is it. It is not the ideal filament for this Taulman 910 would be better. IMO Better even than CF infused due to the superior layer adhesion compared to any chopped CF infused filament (not counting the Markforged ironed/layed in full strands) Now you can consider it 100% infill even though that is not the settings. Instead its printed with 6-9 perimeters on a .8 nozzle with the layer width set to 120% to optimize accuracy and layer adhesion (using SLIC3R as a slicer) further improved by going volumetric E. That means we are within +/- .08mm on outside layer position to calculated layer position. That with a .1mm “error” backlash will give the Nylon a little room to “flow” when starting to be used. as the layers on the outside are a bit rounded. with the .8mm nozzle a layer height of .32mm has been selected as tests have shown so far that the best layer adhesion is between 25 and 50% of layer width and at .32 the printer has to only do full steps on the stepper motor eliminating inaccuracies due to inaccuracies in torque and positioning on micro steps. Same has been taken into account on other stepper motors on this particular printer. So if we are successful and post the stls and ipts then it may or may not work as well on your printer as it has been optimized specifically for this one to get every last ounce of performance out of this design.

Reason we are not using 910 is due to the fact that it would have broken the $100 budget

Keep us posted on the project, it just now dawned on me that the project we have been working on is somewhat similar to this, so once you complete this iteration… and are back to gathering data and analysis in the SDLC. I thought we should compare notes. My requirement is around 170 pounds up a 30 degree incline, so I went with the Vex versa planetary. I’ve documented this here with a CAD here. More details in those links. One thing that stands out to me is the use-case for helical geometry where the versa planetary does not use this. I remember in my studies that helical and spiral offer more torque but the trade-off is that these forces are delegated to the mounting mechanisms (e.g. planet carriers). Anyhow, the gearboxes from Vex are a similar design, so it may be interesting to do a comparison and contrast and see if there is room for improvement.

Forces are always relegated to the mounting mechanism. Helical also distributes the load over multiple teeth that is why its used in an automotive transmission for example. Now I single helical gear needs a thrust bearing (or 2) as you have a force that tries to slide the gear of the other gear or the rack (standard trig applies if you neglect friction) A double helical (herring Bone) like we use almost exclusively eliminates that as 1/2 of the gear pushes one way and the other the other and cancle themselves out .Also double helicals are self aligning so if one gear is held in position the other will follow. We used that successfully to hold a double rack in place between some pinions (4) Making helical gear and especially double helical out of metal with subtractive manufacturing is very expensive and difficult as it requires specialized machinery that is very expensive. Furthermore most gears in cheap gear boxes are not involute either. that means they clank together instead of smoothly roll together. Now with 3DP (additive manufacturing) there is no $$ penalty in making complex shapes (except maybe the effort to CAD them) So making double helical gears - at least for us is the way to go. So you can just mount one to the motor and the other will follow even if its kinda “floating” on the shaft which makes alignment - especially for teanagers much easier. Now we have lifted on a rack and pinion setup with an adymark Gearmotor properly geared 100 lB 20 inches on the work bench repeatedly. What the kids did not test as they felt they did not have the time was if the assemply will hold the weight to when working as stilts lifting the robot up (one in front one in back as an elevator) So it did not last too long as the “stilts” part broke when it misdeployd and a team mate crashed into it trying to quickly push us into place to climb. Feel free to ask any question and I will keep you up to date. Right now we have to reprint a couple of the gears as we detected a small math error but big enough to interfere with the proper meshing of some gears.


Ah, this is true, buying a metal spiral bevel gear cost around $250, so I think the perfect storm is to make the double helical on a resin printer as the prices are affordable, we use this model. I’m still researching how rigid this material can be; I found this example. Thanks for explaining the benefit of this geometry, it may turn out a resin printed gear may be a more affordable solution to compete against manufactured metal gears!

Let me know how it goes We discarded resin due to the fact that they are more limited in material than FDM and prints are in general said to be much weaker. Now when it comes to gears its not only hardness that counts but also wear resistance. So we have run and abused Nylon gears to no end with no visible damage. One of our students felt that PLA was much harder and should make better gear and replaced the Nylon ones with PLA ones on our ball takein. Now PLA is harder and has a higher PSI rating than Nylon but after 1 match we had to pull it as the teeth showed considerable wear damage due to grinding against each other and some were cracked as PLA is brittle. Guess that is why we don’t see Glass gears even though its much harder than steel. Resin can produce more accurate parts. But with what I know right now its not that useful for relatively heavy duty aplication. And the other reason why we did not use it is because there is smells and chemicals etc. on a much higher rate than with FDM But if it works I would like to know as I’d like to keep up on technology

How is the project going so far?