Sorry I didn't see this sooner. If the sides of the bars are parallel, then they are also parallel to the centerline of the bars. Then the geometry should be the bar centerlines, and a thickness offset of half the bar thickness. Since a thickness is being used on the bars, then offset the geometry of the ends by the same amount. This is the kind of part that ROLLDIE was designed for. I will answer any questions directly.

It seems like that part would be better turned than milled.

If the index position can go to .5 degrees, change the FS format statement of the indexer -- for instance: FS 6 1.15(ltn...) The last number 5 is the fractional format of the previous position (the 'tenths' position). This can be a 5 or a 2 (2 tenths of the previous place). In this case the indexer FMT statement will refer to FS 6.

One micron is about .00004, so 3.7 microns is about .00015. You could set the tolerance to either .00015 or .0001. To break up an arc you can use this formula: chord = sqrt(8*err*r), where err is the chord height tolerance. You could analyze the linear results of breaking the spline into linear segments by analyzing the spline dynamically, and comparing the chord length with the radius at a given point by using the above formula.

Jeffrey, can you put up a part file? If it's a real cam, maybe I can help.

Rolldie should work fine. You don't even have to model the slot -- just the centerline. And you can do everything in the flat. Ordinary axis substitution performs cutter offset BEFORE axis substitution. Rolldie 'rolls' the data first, then offsets.

You could tip it up at 45 deg. and use an end mill with inserts. Cutter maintenance is easier than with a ball mill.

This particular part could also be made with ROLLDIE. All that is needed is 4 diagonal lines in the flat. These would be the centerlines of the blades. When you offset from these lines, the opposite sides will be parallel.

Program a 90 degree arc on the circumference. It's not clear if there is any axial movement. If there is, you could program a diagonal line in the flat, and use ROLLDIE.

Try flattening both ways -- in the along and across direction. The truth is somewhere in between The mid-line of the chosen direction will be true length, and the indivual curves perpendicular to that will be true length, but the outer boundary may not be accurate. There is no attempt to account for stretching (or compression).

Mr. M, use the slot option. When you're making a slot, the sides will not be parallel if you use the wall geometry. If you only have the geometry of the slot walls, you must create the geometry of the slot center. The good news is that ROLLDIE works from the flat, so you don't have to wrap any geometry.

More thoughts: what is your cutter grinder setup? Is it CNC?

The toolpath to grind the cutter would be more complicated. Also, the cutting action wouldn't be much different -- the cutting angle on the outer part of the ball (which appears to be all that's used) would be almost the same. The real difficulty is creating a toolpath where the grinding wheel breaks out on the next tooth while it is cutting the face of the current tooth. It still may be necessary to make a toolpath for a V-wheel rather than a dovetail-shaped wheel in order to get clearance between the wheel and the shank, OR, put the flat face of the dovetail wheel on the draft surface, and the angled OD on the plane cutting surface. Are you really 98?

Oops! The "back wall" of the tooth should have an equal angle everywhere with the tooth face, the primary 18 deg. plane. Rather than making a cone FROM the next tooth, I should have made a DRAFT surface TO the next tooth, with the angle of the draft surface referenced to the 18 deg. plane. Then the toolpath is in the plane of the tooth face, using a dovetail-like wheel. The intersection of the draft surface and face-plane is on level 46. DRAFT15.MCX-6

The tooth design has a lot of restrictions. If the face of the tooth is NOT a plane, the wheel which grinds the tooth will probably gouge somewhere. Also, the plane is tilted so that the wheel will not hit the shank. Also, it looks like the picture. The design slices a sphere at 18 degrees from the axis, going thru the tool tip. (I really just sliced a circle). Any plane slice thru a sphere is a circle. The back of the tooth is just a cone at 15 degrees to the slicing circle. Then if you intersect the cone with the slicing plane, the intersection is a toolpath. That's all that's needed if the wheel is a dovetail shape, with a 15 degree angle. If the wheel is a V shape, however, the toolpath is more complicated.