Crown Gear

[Nate]

...
but cutting them is a challenge other than simply using the 3d stl like a model to be engraved, but by algorithm, near impossible.. so far as Ive been able to figure out.. and engraving would leave a lot of roughs to deal with from the tool
lines, ( the reason tangent shaving works so well, no tool lines.)
...

Pardon my naivitee, but is "engraving" the same as tracing contours with a ball-nose end mill?

[Nate]

...
   The problem with them is that they really lend themselves only to engraving type of cutting on 3d , not 4th axis as you
cant shave the tangents in 4th axis.

When I started generating bevel gears, I set them up so that the (idealized) contact line between the teeth would always go through the tip of the pitch cone.  That makes the gears symmetric and allows for a system of many different bevel angles that all mesh with each other, but it's not the only option.

Now, let's suppose, instead, that we design a set of bevel gears so that the idealized contact line between the teeth is parallel to the shaft of one of the gears.  Then we would end up with an asymmetric bevel gear pair - one of the gears could be an involute spur gear, and we'll call the other one a crown gear.

I need to do some pencil and paper work to be sure that the it works out, but this sort of crown gear should have the nice properties that:

For a 90 degree drive, it should be possible to 'tangentially shave' the tooth flanks using a taper nose end mill and a 3 axis CNC machine (or, for the masochist, even on a conventional 3 axis machine with a rotary table).

It should mesh well with a range of involute spur gears that have the same pressure angle and pitch.

It should be mechanically viable with decent efficiency, little noise, reasonable meshing tolerances and so on.

(This seems like such a good idea that it's probably either wrong, or someone else thought of it well before I did.)

[ArtF]

Hi Nate:

  Yes, when I use the term "engraving" I really mean 3 d raster cutting on a plane. It leaves really nasty toolmarks unlessdon ein a tight raster, which is why I resist it so much as a bevel technique. As you your idea of using a
tangent to shaft as the idealized contact point, I cant quite picture how a normal spur can run on a 90 degree bevel, dont the teeth have to narrow toward center?

Thx
Art

[Nate]

...
  Yes, when I use the term "engraving" I really mean 3 d raster cutting on a plane. It leaves really nasty toolmarks unlessdon ein a tight raster, which is why I resist it so much as a bevel technique.
...

Would tracing geodesics (in particular, geodesics that are close to radial) and restricting the raster to the tooth flanks produce any significant improvement?

...
As you your idea of using a tangent to shaft as the idealized contact point, I cant quite picture how a normal spur can run on a 90 degree bevel, don't the teeth have to narrow toward center?
...

Yes, the teeth have to narrow, but they don't have to narrow along a radius.  I need to think about it a bit more, and I'll see about modelling it.

[ArtF]

>>Would tracing geodesics (in particula r, geodesics that are close to radial) and restricti ng the raster to the tooth flanks produce any significa nt improveme nt?

  Im really not sure, I have tried mental exercises like figuring the effect of only using radial paths from cone center ,
but its hard to say what shape of toolpath would make the toolmarks so they best mesh together without having
to file all day

Art
..

[JamesTSG]

Or make the gearsets slightly oversized, put them in a jig, and then run them for a while dry or with a cutting fluid (or even abrasive powders!) to let them "mill themselves". 

[ArtF]

  That IS one way. Id be interested in peoples experiences having done that. The STL's
and models GM puts out could be used in Vectric and such to generate a path for engraving
type cutting.. I may try that method to see what I get in the new year. Ive designed Auggie
to be able to shut down anytime and resume the next day, so I wont mind so much those
multi hour engraving paths..

Art

[Nate]

The approach I'd been originally thinking of seems to involve undesirable undercuts.

How about using "improved peg gears" as crown gears?

[ArtF]

Think Id prefer two disks, a length of brass rod and just make them from pins..

Even those teeth look like they need to narrow toward center...

Art

[Nate]

More fiddling...  I guess they don't need to be peg gears to work as 'straight sided bevels'.