What's new
What's new

Machining a Gates Poly Chain GT2 Sprocket

jcorsico

Aluminum
Joined
Feb 17, 2009
Location
Maryland
Hi guys. I need to machine the tooth profile for a Gates Poly Chain GT2 sprocket, but I can't find the exact tooth profile anywhere. I think this was patented by Gates, so the tooth profile should be public information. This is for a prototype sprocket (quantity of 1).

I tried the Gates web site, but their CAD models say that the tooth profile is just an approximation, and shouldn't be used in production.

Does anyone know where I can get the tooth profile?
 
This Appears to be the patent. They have been intentionally vague in the patent to help protect their IP. If you want to know the exact profile you are going to have to buy a belt and layout the tooth profile as best you can.

https://patentimages.storage.googleapis.com/b3/49/66/8b4b777e751902/US8070634.pdf

There is some more info on it on page 101 in the pdf below.

https://assets.gates.com/content/dam/gates/home/knowledge-center/resource-library/catalogs/old-pc_carbon_manual17595_2011.pdf

Thank you very much.

Yes, the patent appears to describe the high level mathematics behind the tooth shape. But not the specific sizes of teeth chosen by Gates. Argh!
 
What size GT2? 8mm GT2 will run on a 8mm HTD pulley, not sure if there are design differences but they do work in practice, the larger sizes are different for certain.
My old Gearotic has GT2 5mm and 8mm, a new version might have the larger ones now that the patent has expired
 
FWIW,
I've reverse engineered similar profiles scanning the part on a basic 1200dpi flat bed scanner. Import into cad sketch environment, take relevant measurements from actual part. Create an accurate single tooth profile, polar array, make solid and CAM it.
 
This Appears to be the patent. They have been intentionally vague in the patent to help protect their IP. If you want to know the exact profile you are going to have to buy a belt and layout the tooth profile as best you can.

If I was doing this work, I'd want a sample pulley in as close a size to the one I wanted to make as possible. It's the pulley tooth profile that matters here, not the belt (which has a dynamic change of shape relative to the pulley).
 
What size GT2? 8mm GT2 will run on a 8mm HTD pulley, not sure if there are design differences but they do work in practice, the larger sizes are different for certain.
My old Gearotic has GT2 5mm and 8mm, a new version might have the larger ones now that the patent has expired

The belt is a Gates Poly Chain GT Carbon, with 8 mm pitch. Gates specifically says that this belt needs to run on a Poly Chain GT2 sprocket, and cannot use an HTD sprocket or a Poly Grip sprocket (very confusing that "Poly Chain" and "Poly Grip" are different products).
 
FWIW,
I've reverse engineered similar profiles scanning the part on a basic 1200dpi flat bed scanner. Import into cad sketch environment, take relevant measurements from actual part. Create an accurate single tooth profile, polar array, make solid and CAM it.

If I was doing this work, I'd want a sample pulley in as close a size to the one I wanted to make as possible. It's the pulley tooth profile that matters here, not the belt (which has a dynamic change of shape relative to the pulley).

These are both good ideas.

I think the belt tooth is intentionally a different shape from the sprocket gully (they are not mirrors of each other).

So I could buy a Gates sprocket of the appropriate size, and then scan that sprocket to CAD.
 
You can reverse engineer an existing sprocket with a decent set of calipers, radius gages for outside radii and gauge pins for inside radii.
 
If I was doing this work, I'd want a sample pulley in as close a size to the one I wanted to make as possible. It's the pulley tooth profile that matters here, not the belt (which has a dynamic change of shape relative to the pulley).

Does the tooth profile on the belt change shape? Or does it pivot similar to the way a traditional drive chain does?

I was thinking it would pivot like a chain, making it possible to solve for any pully tooth profile via iterative subtraction/slicing.
 
The belt is a Gates Poly Chain GT Carbon, with 8 mm pitch. Gates specifically says that this belt needs to run on a Poly Chain GT2 sprocket, and cannot use an HTD sprocket or a Poly Grip sprocket (very confusing that "Poly Chain" and "Poly Grip" are different products).


I understand that. Just FWIW we've been violating most of those those recommendations for years, and with higher loads and higher RPM than Gates says those belts will tolerate. I'm pretty sure they come up with slight changes to the tooth shape just to make the older out of patent belts technically obsolete in the catalog
The belts run on the major diameter of the pulley, not the tooth space, the tip of the tooth is most important, as you mentioned the body of the belt does not ride in the bottom of the tooth space and there is daylight between them under load. Get an 8mm GT2 belt and wrap it around an HTD pulley and see how it fits.
 
Does the tooth profile on the belt change shape? Or does it pivot similar to the way a traditional drive chain does?

I was thinking it would pivot like a chain, making it possible to solve for any pully tooth profile via iterative subtraction/slicing.

No, it changes shape for every tooth count, that's one of the things that make HTD/polychain/polygrip belts difficult.
 
Does the tooth profile on the belt change shape? Or does it pivot similar to the way a traditional drive chain does?

I was thinking it would pivot like a chain, making it possible to solve for any pully tooth profile via iterative subtraction/slicing.

Yeah, like Mud said. Also, the "last bit" of tooth design gives you a little quieter running, a longer run life, etc., but you can get away with approximations if you're careful and don't mind the small performance hit.
 
Yeah, like Mud said. Also, the "last bit" of tooth design gives you a little quieter running, a longer run life, etc., but you can get away with approximations if you're careful and don't mind the small performance hit.

Sorry to bump this thread but I am still a kind of confused, probably in part because my post wasn't very clear.

When I say iterative slicing I was thinking something along the lines of a motion sim, like a frame by frame of how a hob cutter would take a chip. In theory it should return a very close representation of the involute of any shape you feed into it.

Motion-Sim.png


Wouldn't this work to generate the correct profile or something close to it?
 
When I say iterative slicing I was thinking something along the lines of a motion sim, like a frame by frame of how a hob cutter would take a chip. In theory it should return a very close representation of the involute of any shape you feed into it.

Motion-Sim.png


Wouldn't this work to generate the correct profile or something close to it?
Technical name 'odontograph' and yes it would work, that's how they used to design hobs on paper, but what software did you use ? Pretty.

(Looks a lot like Cadkey 7 from here :D)
 
Technical name 'odontograph' and yes it would work, that's how they used to design hobs on paper, but what software did you use ? Pretty.

(Looks a lot like Cadkey 7 from here :D)

Acad 2d - Mostly automated via a lisp routine I was working on for standard involute gears.
 
Sorry to bump this thread but I am still a kind of confused, probably in part because my post wasn't very clear.

When I say iterative slicing I was thinking something along the lines of a motion sim, like a frame by frame of how a hob cutter would take a chip. In theory it should return a very close representation of the involute of any shape you feed into it.

Motion-Sim.png


Wouldn't this work to generate the correct profile or something close to it?

Maybe. Putting my "not an engineer" cap on, the method shown seems to prioritize the root of the sprocket, which to my mind is not the critical feature to design around. It's flank engagement/disengagement.

If you can rerun the process looking at that it might be more useful. Whatever gets you full flank mating as quickly and smoothly (minimize rubbing) as possible will to my mind be most effective.

Leaving a gap at the root may also help with reducing the noise of trapped air being squished out.

Again, not an engineer. If there are some floating about to yay or nay my thoughts that would be welcome.
 
Maybe. Putting my "not an engineer" cap on, the method shown seems to prioritize the root of the sprocket, which to my mind is not the critical feature to design around. It's flank engagement/disengagement.
What he did will show how an imaginary perfect rack (belt) will mesh with an imaginary perfect sprocket. That's the start point. Modifications come next, if you want to make any.
 








 
Back
Top