So he might be implementing something completely unnecessary by your definition. Your servo system already does exactly the same. By your misunderstanding of controls.
I gave examples for the outer loop back, I don't have to repeat them. Feel free to continue ignoring them.
Nick
Nick, I am happy to correct YOUR misunderstanding of controls. You brought up glass scales as solving the problem of, "This simply means that the contol and the user knows exactly where the axis are. No matter how big the threshold, no matter what backlash, no matter how many steps lost, no matter wether one axis stalled due to overload."
Those are definitely problems every CNC machine should consider, but none of them are really problems that glass scales are intended to solve with the possible exception of backlash, and even there, not so much. They simply provide a more real time input to the backplash compensation which we can set up by other means. None of these require us to close the loop back to the controller either.
Glass scales are used to minimize problems like thermal expansion and leadscrew error. BTW, I can even tackle leadscrew error with Mach3 and glass scales. Since it has leadscrew mapping built in that is easy to use, I simply mounted a set of high quality glass scales temporarily and used them to measure my leadscrew errors. Took an afternoon and now it is done.
Glass scales are not particularly easy to reconcile with the servo encoders the drives need either, so when you close the loop back to the controller, there are really two loops there. The glass scale provides compensation that is applied after the primary loop has been used for the gross positioning.
Because it is only these things that glass scales matter for, as I mentioned, most commercial CNC machines including Haas and Fadal don't have glass scales by default. This is also the reason why simply closing the loop at the servo drive is sufficient--because all we need to solve the problems you did mention is to close the following error and that's what the drive does. Your outer loop examples therefore do not need an outer loop, and I have described in detail what you should have, which is what the glass scales are really for.
But there is a reason that glass scales are so rare in the field. They simply aren't needed for most cases and there are other solutions to the problems they solve.
Commercial VMC's without glass scales, and most don't have them, solve the leadscrew error by doing laser mapping. They are periodically re-calibrated in the field. They minimize thermal expansion errors with discrete temperature sensors.
Machinists also play a role, they minimize thermal effects by warming up their machines and using wear offsets based on measurements taken along the way. Or by using in process probing like the RAMTIC strategy I mentioned. Isn't it interesting that a company like Renishaw is building precise probes on VMC's without glass scales? If you're clever, you can do so.
Nick, we can keep going as long as you like. I've been doing CAD/CAM since college when I paid for my tuition by porting CAD/CAM software from DEC PDP-11's to the new generation of M68000 workstations. I know how the stuff works.
Cheers,
BW