Antonius, thanks. My model already used the same connection types as shown in the You Tube video (some different references). Good to know the datum points have to be separate features. My were already separate simply by luck.
The idea onb the limits on the pin is a really good idea, the flipping problem is very annoying.
I've not had trouble when I only have a few links, but when I put the whole chain together (104 links total) is when performance becomes a problem.
My path has relatively complex geometry, it only has two sprockets, but it does have a couple of glide paths that change the profile. I think that would make the rotational motor give me fits to get close to a constant speed.
Funny, yesterday my search in the forums showed no similar topics, today I see several others. None have the solution for the performance issue though.
Files uploaded as document; http://communities.ptc.com/docs/DOC-3642