Surface Hits without Focus

Well what I did after all was turn Control to Auto, Find Surface to YES, and Surface Variance to Yes. It takes a little longer but the Optiv hit the surface point with no problems, even our test part that is thicker than the others came out well.

Just make sure you have a wide enough range for the focus. I usually have problems if it's not painstakingly obvious that a material is out of focus, then in focus, then out of focus again.
You could have 1 quick but wide surface point, set it as your z-axis and then take another couple points with a much smaller focus range, if that helps

Edit: When I say wide, I mean larger Z values

Well what I did after all was turn Control to Auto, Find Surface to YES, and Surface Variance to Yes. It takes a little longer but the Optiv hit the surface point with no problems, even our test part that is thicker than the others came out well.

Usually in class (back when I worked for hexagon), I would have people leave Control on Auto : then the computer handles the Range/Duration. It's faster to program, but slower to execute. And that's usually what I'll do for a 3-point plane that I'm using for my manual alignment. Top Light (maybe Ring), minimal zoom, auto for range/duration. It's quick and easy, and good enough for manual.

but not for a DCC alignment. This is when you push the machine to max.

#1) full zoom. You need to zoom all the way in to get maximum resolution.
#2) no more auto. Set a reasonable range and duration. This is part dependent
#3) you're at full zoom. You're gonna need more light
#4) don't use 3 points. for god's sakes don't. Use many. You're in DCC, it ain't like you're sitting there and watching it, you should be off doing something else while it runs! 3 points is the minimum, which we should never be using.

per your cylinder program:

I can't tell if the axis of the cylinder is lined up with the camera. it should be. I don't think the vision probe should be used to try and capture a cylinder's OD when perpendicular to the camera, it'd never be able to focus on the curved surface. Or at least it shouldn't be able to, it's possible that it might give you SOME data, and I wouldn't trust it. If I were forced to put a cylinder perpendicular to the camera, i'd 100% be touchprobing it. Hope that helps.