CIP motion technical details?

I'm trying to get a better handle on how coordinated motion works on CIP drives like the 6500. From what I gather:

• CIP Sync is used to get the clock in the drive to match up with the clock in the ControlLogix. Variance in the clock values is typically low but increases with each Boundary or Transparent clock. Real world variance measurements appear to be non-existent, but it is implied they are in the single digit microsecond range.
• The PLC does all of the curve calculations to get set points and sends a few of these points, along with clock times the axis should be at the points, every course update period. How many point/time pairs are sent or if it is actually just one point, I don't know, can't find any documentation on this.
• The drives, running at a set cycle time (250microseconds?) handle the PI controller cascade and try to hit the point(s) at the designated times that it received from the PLC. The drive does linear interpolation to define points between the ones provided by the PLC
• Gearing to commanded position is handled by the PLC. It generates the master and slave points at the same time. Gearing to actual position requires the PLC to retrieve the position feedback from the master to generate the slave points (is this feedback data time stamped or jittery regular CIP communication?). The delay for the feedback causes the slave to lag in reaction to master position/speed changes (can this be compensated for?).

Can someone correct any incorrect assumptions of mine and/or fill in the blanks. Maybe point me to documentation that covers some of these details. I've found 100 pdf documents telling me how to add and configure drives and axes and zero that tell me what is actually going on. The above list is from word of mouth from others, so the details are suspect.

I can find out since we are a member of ODVA and one of our guys go to the meetings. I don't know the details because I know I would not do things the way Rockwell does them. I do know this. The PLC generates the curves and downloads the next point one point at a time. The motion controllers are relatively simple but they do a second order interpolation between the points.

What you call gearing to the target position isn't really gearing in the traditional sense. If the PLC is generating both target position it simply downloads them both to the two axes at the same time. There is NO delay. We call these synchronized moves because the axes are following their own target position but all the target positions are the same.

Gearing is required when reading the actual position from an uncontrolled axis. In this case the master actual position must be read and resent to the slaves. Yes, there is a delay but I think Rockwell tried to compensate for this delay. I don't know how well Rockwell does this.

We keep the master and slave in the same controller so the master position can be used without delay and synchronously. If the master must be used between multiple controller the master encoder can be daisy chained so that all the slaves immediately see the master without any delay.

If the PLC is only sending down one point per course update period, then you don't have very good control of the axis. For 60 CIP drives AB recommends 30ms coarse update. I've done applications on other platforms with that many axes where I ran entire cams in 30ms; only having one real point in that time frame would be a disaster, regardless of how much interpolation was going on.