pic: CGX-115, cam-counterweighted cycloidal drive

[asid61]

loading photo ...

[Scott Kozutsky]

As I understand this mechanism, the cyclodial gear nearest the cim drives against the mounting plate of the cim, which would be fixed against the main structure of the bot. The other cyclodial gear is driving against another plate which is mounted to the arm.

Because both cyclodial gears advance in the same direction, wouldn't this gearbox just provide a powered zero output?

BTW, great work, it's actually because of you that I know that these gearboxes exist at all.

[asid61]

Originally Posted by Scott Kozutsky

As I understand this mechanism, the cyclodial gear nearest the cim drives against the mounting plate of the cim, which would be fixed against the main structure of the bot. The other cyclodial gear is driving against another plate which is mounted to the arm.

Because both cyclodial gears advance in the same direction, wouldn't this gearbox just provide a powered zero output?

BTW, great work, it's actually because of you that I know that these gearboxes exist at all.

That's an interesting way to look at it. You would be correct if both cycloidal gears had the same tooth count. However, because the second stage gear is actually minus one tooth, the result is a little different.
This is the order of motion:
1. The miniCIM drives the camshaft
2. The wobble gear rotates around the CIM axis, but it itself turns once for every 30 rotations (the first 30:1 reduction)
3. The second wobble gear, bolted to the first one, rotates once for every 30 rotations as well.
4. Due to the eccentric nature of that rotation, the arm is pushed 1/30th of a rotation for each rotation of the second gear about its axis, creating the second 30:1 reduction.

It's a bit easier to see if you have the cad model working. The arm is forced to rotate, essentially.

Don't thank me, thank s_forbes! He provided the CAD model for the first cycloidal gears and the spreadsheet to calculate the gear ratio and dimensions.

[Scott Kozutsky]

That's really clever. I had the cad open on the grabcad site but the explosion was clear enough to get me to (mostly) figure out what was going on.

Just clarifying; because of the difference in gearing between the two cyclodial gears this goes from a powered zero to moving the difference between both "stages" every revolution.

The gear ratio should be the same as if this were a standard 2 stage cyclodial gearbox with the same number of teeth, as I understand it.

[asid61]

Originally Posted by Scott Kozutsky

That's really clever. I had the cad open on the grabcad site but the explosion was clear enough to get me to (mostly) figure out what was going on.

Just clarifying; because of the difference in gearing between the two cyclodial gears this goes from a powered zero to moving the difference between both "stages" every revolution.

The gear ratio should be the same as if this were a standard 2 stage cyclodial gearbox with the same number of teeth, as I understand it.

100% right.

[Scott Kozutsky]

Wouldn't it be a 28*29:1 = 812:1 gearbox then? The output 'stage' has 30 pins therefore 29 lobes and the input has one less lobe.

[asid61]

Originally Posted by Scott Kozutsky

Wouldn't it be a 28*29:1 = 812:1 gearbox then? The output 'stage' has 30 pins therefore 29 lobes and the input has one less lobe.

If it was a normal reduction, yes. But in this case some witchcraft happens where the reduction on the second stage is just "# of pins on the ring gear" instead of "pins-1". I'm not exactly sure why, but I suspect it has to do with the way a planetary gearbox ring gear is exactly 1 less in the gear ratio, only in reverse. After all, in this case we are driving the ring and not the carrier.

[nuclearnerd]

I'm a bit confused, this picture shows a big old mounting flange, but others show the versaplanetary mounting pattern. Did you draw both? I ask because we used 3 stage versaplanetaries this year at 250:1 reduction. We found we were really pushing it with the strength of the hex shaft, and especially the holding power of the two #10 screws. If you've got such a big reduction, consider engineering it to use another output and mounting arrangement to bolt directly to frame members and/or sprocket bolt circles.

[asid61]

Originally Posted by nuclearnerd

I'm a bit confused, this picture shows a big old mounting flange, but others show the versaplanetary mounting pattern. Did you draw both? I ask because we used 3 stage versaplanetaries this year at 250:1 reduction. We found we were really pushing it with the strength of the hex shaft, and especially the holding power of the two #10 screws. If you've got such a big reduction, consider engineering it to use another output and mounting arrangement to bolt directly to frame members and/or sprocket bolt circles.

There are two separate designs. This one is a larger, higher reduction gearbox meant for moving large arms at a 900:1 ratio off a miniCIM or CIM. The other one (CGX-116) is a VP stage with a lower reduction of 594:1, but can only output to a hex shaft.
I'm seriously considering designing a better VP output stage or output shaft that can hold a larger shaft and maybe output to a 1.875" hole pattern. I totally agree that the mounting options and hex shaft are limiting factors here.

[Mike Schreiber]

For anyone struggling to visualize this (as much as I hate to refer someone to Wikipedia) There's a good animation here.

https://en.wikipedia.org/wiki/Cycloi...idal_drive.gif

[asid61]

So apparently the radial loads on the camshaft (which is attached to the CIM shaft in this case) is equal to Torque/(eccentricity * gear ratio), which assuming you want to have a 200ft-lb (2400 lb-in) torque on this maximum comes out to
2400/(0.05*900)
or about 53 pounds of pure radial force on the CIM. A CIM at stall on a 12t pinion experiences about 22lbs of pure radial force. At the moment I have a pair of 8mm bearings in the output shaft, so it's probably safe, although on smaller gearboxes like the Versaplanetary version where this is not the case there could be savage failure modes. This is mainly a problem for cam-counterweighted cycloidal drives, as they don't have 2 wobble gears on 180* phase offsets to cancel out the forces.

[Cothron Theiss]

This all looks great! A few quick things before I dive back into your CAD:
You're missing a few parts in your GrabCad folder for this assembly, like the small bearings around the first wobble gear, or the countersunk screws bolting the two wobble gears, and just a few small parts here and there.
Also, where did you get the 50mm ID 65mm OD bearings? Just wondering what supplier sells those.
And how did you determine the dimensions for the lightening pocket on the camshaft?

[asid61]

Originally Posted by Cothron Theiss

This all looks great! A few quick things before I dive back into your CAD:
You're missing a few parts in your GrabCad folder for this assembly, like the small bearings around the first wobble gear, or the countersunk screws bolting the two wobble gears, and just a few small parts here and there.
Also, where did you get the 50mm ID 65mm OD bearings? Just wondering what supplier sells those.
And how did you determine the dimensions for the lightening pocket on the camshaft?

As usual, I forgot to do a Pack and Go. Try downloading the STEP file (GrabCAD has an option for that). Alternateively if you download the massive Parts folder it's all in there.

Recently I discovered the 6700 series bearings, and less recently the 6800 series bearings. These are very thin-section bearings, with the 6700 series being a mere 4mm wide and usually having a difference in OD between 6-7mm, so you can get a 30x37mm bearing. These are crazy useful for a lot of designs, and cost just a few dollars on Ebay or Aliexpress, as they are standard sizes. I believe I am using 6810 bearings in this design. There's an amazon supplier for 6800 series bearings that offers Prime, but they are vastly overpriced and usually cost $10-50 each compared to $1-5 for Aliexpress.

To get the lightening pocket just right, I first drew the maximum size pocket on the camshaft, in this case somewhere around a 0.09" outer wall and 0.125" thick material between the front and back pockets. I then made an assembly consisting only of the cam and the off-center rotating components including all the the gears, screws, and bearings. I mated the center of the cam to the origin and took Mass Properties readings to determine the center of mass, and trial-and-error adjusted the depth/wall thickness of the pocket accordingly. In this case, I ended up with a 0.09" wall and a 0.187" wall between the two pockets, and under 0.0001" of theoretical concentricity.