profile shifting

[Neon22]

I see numerous rules about profile shifting appearing when pressure angles and Teeth get below a certain ratio. E.g. "Gears that are smaller than 32 teeth for a 14.5° Pressure Angle, or 18 teeth for a 20° Pressure Angle, have a Root Circle smaller than the Base Circle, resulting in the teeth being undercut."

This doc shows how to determine if undercut will occur. Pitch shift can be both positive (usual) and negative(rare). Negative looks cool - I expect it weakens the gear tooth so undesireable but does look interesting. pg 604 ( doc starts at 595).

• http://www.khkgears.co.jp/en/gear_technology/pdf/gear_guide1.pdf seems simple - basically shifting the addendum and dedendum while leaving base circle in same place.

Should we detect undercut and auto shift or just annotate and ask (and maybe add a page about undercut to the tabs) ?

[jnweiger]

This here http://lcamtuf.coredump.cx/gcnc/ch6/#6.2 confirms that profile shift is all about the addendum. The tricky part, appears to be that two meshing gears should have complementary profile shift: The smaller gear has an enlarged addendum, while the larger gear has a reduced addendum.

[Neon22]

YEs :( and apparently they won;t mesh unless matching. so can only be applied in cases of a gear-train.

However other solution is to undercut. which will work but is harder. That csg code will do it but its a lot of code :(

[Neon22]

So do you think the right thing is to:

• Allow user to enter profile shift parameter. as a ratio (?) split between two gears.
• Report in annotation this shift and that meshing gear is expected to have shift of X.

Then we cover this issue - separately from undercuts (for which this is one strategy). Then in gear-trains - we could utilise this feature.

[jnweiger]

Yes. I'd make profile shift a percentage. Two gears would mesh, if one has -X % and the other has +X % profile shift. -100% should mean that the addendum is 0.

And yes, then it is one strategy against undercut. My undercut warning already mentions this.

On Mon, Apr 7, 2014 at 5:03 AM, Neon22 [email protected] wrote:

So do you think the right thing is to:

• Allow user to enter profile shift parameter. as a ratio (?) split between two gears.
• Report in annotation this shift and that meshing gear is expected to have shift of X.

Then we cover this issue - separately from undercuts (for which this is one strategy). Then in gear-trains - we could utilise this feature.

Reply to this email directly or view it on GitHubhttps://github.com/jnweiger/inkscape-gears-dev/issues/9#issuecomment-39692707 .

[jnweiger]

I believe, this was too simplistic. Real profile shift should shift the entire contour inwards or outwards. Not just the endpoints.