Experimenting with Cycloids

Was asked to look into a cycloidal drive

Have managed to draw some cycloids. I thought it would be nice to have one thats self aligning. Of course its 3DP

So here is the first attempt on a double helical (herring bone) cycloid)

3mm offset 10mm pins and here the (of course) 2 part ring - as how else would one assemble it - printed in Hips on my chiron we are using I think a 6805-2RS bearing in the center.


Basically works as expected I was - as I feared - a little too generous on the tolerances but for a first prototype I did not want it to not go together due to tolerances that are too tight

And my first attempt on the crank


Which too needs work

Overall for a herringbone disk that size it should be a little thicker the tolerances a little tighter But then something like this never works on the first try. The reducer will be 18:1 and coupled with our 41:19 bevel setup should give us 38ish :1 so with a Cim and a 2in dia pulley on the output should lift a 150 lb robot at about 12 in/sec at about 30 Amps give or take a couple. Not this one as its mostly hollow - right now I go for fit. It will also have 2 disks offset by 180 deg. to balance it. The pins for the output right now I will use 3dp bushings on until everything pans out and I probably pick a bearing for that at that point

8 Likes

Note that the shear forces on the camshaft are equal to the gear ratio times the output torque divided by the cam offset. Reinforcing the cam with a through-steel shaft would be a good idea.

I like the herringbone cycloid, although I think it could reduce efficiency. Very interesting idea. Looking forward to hearing how much torque you can push.

I’ve done some testing on a 12 to 1 and 11 to 1 2 stage printed cycloid. 1st stage rotor is .25" thick. So far no problems with this stage. The second stage however has been a problem. Went to 1/2" thick and tried several different filaments. All suffered eventual mush out. I believe the main problem is tooth pressure. PLA, PETG, and CF Nylon have all eventually failed when tested by repeated stall with a mini-cim. I’m currently in the process of testing with a pure polycarbonate filament. This has forced me to tackle all the issue of printing PC. Had to modify the printer to handle an active heated chamber. Also, have to compensate for more shrinkage than the other filaments. So far the results are promising. In the process of some redesign. Doubled up on some bearings, tweaked cad for more precise fit and now attempting to balance the rotors. I ignored balancing until now. First protos are like a phone vibrator on steroids. So far a 3d print cycloid looks very possible for First applications. The question is what reduction is past the plastic rotor tooth pressure limit.

On the 3d printed one that I made on the other thread, it was a 25:1, and after about 10 lbs. of lifting force the teeth deformed enough to slip. It appeared the teeth were limiting in that configuration, though at 25"1 the teeth are pretty small.

I am worried about the above described mushing up material flow. I intend to attempt to tackle that with multiple disks as that would engage multiple teeth the helical approach too is to have at “hand over” time 2 teeth engaged And on the shaft I have considered to make the axle a tight fit and mount the ends on bearings the plastic would still take a lot of sheering forces. also trying to figure a way to print it lying down. maybe cut it in half. I worry though of introducing a “rattle” due to size changes as its not that accurate then due to the “squish” on the bed. Now I could print it with a raft but the target material (nylon) is extremely difficult to print with a raft and not have the raft become a permanent part of it. Especially with the bigger nozzle I have. I think I first will finish the gear box tester I am working on (mostly software left on that). The problem with stalling a (mini) CIM is that this will be an FRC unrealistic thing. We are limited to 40 Amps. As I dont have the funds I got to write the code for some arduinos and roll my own controller. If money would be no object I would take a roborio, PDP and connect 2 Talon SRX to it. then I make 2 gear boxes or 1 and use a toughbox mini or something off the shelf. Then I connect the 2 with a belt and pulley. Then I put the one of the to be tested one gearbox on full speed (or any desired speed) and control the load one (which can be a toughbox mini or any other gearbox that back drives) to put enough load on to make the amps on the to be tested gearbox rise to the desired level. The talon SRX has a feature that reports the Amps back. And you can use the speed you feed to slow down/speed up to keep the amps steady we use that in auto to “crash” into a wall and stop when we reach 8 amps). Now you can have that run at extended times at lets say 30 amps. The above with our 41:19 bevels should give a 38ish:1 reduction which with a full CIM should lift 150 lb (full weight robot with battery and bumpers) at 30ish Amps (depending on efficiency) Now the setup would have to have a means to properly tension the belt. And it would be also a test for the pulleys. Be aware the Gates belts - according to specs - will rip somewhere around 300 lb of load so if you test beyond that use multiple belts in parallel. Well it wil lbe a while especially as I am also trying to move and have quite a few other challenges. But ATM the gearbox tester moved up in my priority list as I have quite a few people asking me for results. If/when it works I will make it available

If you give me a link I look at it. But that is why I don’t want to go under 3mm offset and go with 10mm pins. If you have a 10mm thick disk and 6mm pins you should be able to lift 60-80 lb with most filaments. Also print the disks solid 100% infill to keep them from warping and for 150 lb the walls on the crank at any point should be >5mm solid again or you will get deflection (skipping) because of it warping/deforming. Even if you clamp an 8mm steel rod and hang 150 lb from it it will bend so hence the offset has to be big enough that the deflection just causes a little “slop” under load.

1 Like

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.