Carousel ATC - Indexing for directional rotation and comp questions

Hi everyone,

I am in the process of designing a carousel-style ATC for my router and hope to use the carousel.comp by Andy Pugh. 

I'd love to make use of the feature of bidirectional rotation to make things go faster. This raises the question of designing the index and identifier codes. 

1) Regarding gray vs. index in carousel.comp
Does the comp use an index signal (pocket position) + gray (pocket identification) in gray mode? If so, in which order? Move to index and then check ID?

2) Regarding index signals
I have thus far used inductive sensors with modified screw heads for targets in all my CNC axis in homing applications. This is easy to do as the machine always approaches the sensor from the same direction.  Using an 8mm inductive sensor usually means that it will trigger at some percent of overlap with the target. Accordingly, the direction of approach makes quite a difference for the final position. In the case of an ATC, this may be catastrophic;).
Hence I have two questions:
2.1) Does carousel.comp consider this and overruns an index to then approach it from a defined direction? If not, would that be a feasible addon?
2.2) If not 2.1: Are there any suggestions in designing a sensor target that reacts the same in both ways? I was thinking about using a triangle... 

As always: Thanks;)

I've just run in to a very similar - if not the same - problem.

ATC project

Feel free to read through my ongoing ATC thread, but just today I realized that using proximity sensors alone will not garauntee the carousel will stop at a precise location - regardless of the flag shape or sensor specifications.

Even if you tune the flag shape to be perfect (same trigger point approaching from either direction), changes in the rotational speed or deceleration will cause carousel.comp to stop the rotation at a slightly different location.

You could tune flag, speed, and decel, but changes in the weight of the carousel (i.e. different tools) will likely lead to a non-precise stopping point.

If you can live with a single-direction ATC you could have it stop on the trailing signal edge, then back up slowly (like an axis homing routine).

Otherwise you need a physical mechanism that ensures a specific rotational position (geneva drive, shot pin, ratchet & pawl).

I'm stumped on this one.  Looking forward to seeing if can come up with a solution that doesn't result in a 2-minute tool change.