@keng brings up a good point about the tool definition. MeshCAM assumes that the diameter of the cutter increases from the flute diameter directly to the shaft diameter at the end of the flutes without taper. It interprets this definition
MeshCAM will avoid the shank on 3D parts where the surface is sloped, by offsetting the toolpath outwards from the part. I have not checked its behavior on 2.5D parts (where all sidewalls are vertical) but will do so.
@3dsteve, your toolpaths look fine from the screenshots. You used a reasonable stepdown for the waterline, and from the toolpath shot there wasn’t a big jump down to the pencil pass at the bottom of the material.
Since you didn’t add supports to tab the workpieces into the rawstock, another possibility could be that the pieces shifted as you began to cut through and pinched the cutter.
@Darren, it is always good to use a pencil finish on 2.5D machining. Waterline isn’t guaranteed to cut to the bottom of the material, even when the stepdown is an even divisor of the material thickness. Due to internal calculation roundoffs, MeshCAM will usually not take the bottom possible waterline cut, and that holds true also for roughing cuts. MeshCAM might not take the bottom possible roughing cut, so always assume that the finishing cuts might encounter a material depth equal to a roughing stepdown. My rule of thumb is never make the roughing stepdown deeper than I am willing to let the finishing cutter bite.
It is always a good idea to use the same cutter for finishing as pencil in 2.5D machining. On 3D parts where you are just tighening up corners is the place where smaller-diameter pencil cutters are usable. But flute length is always important to keep track of there too.