gtom wrote:is there a example how to integrate the stepgen in the pid?
Well, doing it in HAL is not that complicated.
The problem is with the results.
The idea is that steppers have a different torque/velocity curve than a regular servo.
If a servo gets behind the commanded position you simply increase the power so it catches up.
When a step motor gets behind commanded position it's usually because it doesn't have enough power. Increasing the command to it will only get it to lose more steps, and stall eventually (actually quite fast).
The only thing you could do is connect the position error (difference between commanded and feedback) to the adaptive-feed input to the motion controller. That way emc2 will slow down all axes if one motor starts to fall behind. In this case one can only hope the motor which started to fail will magically heal itself and catch up eventually.
It is (in theory) possible to drive a stepper like a servo, but you need a really high count encoder for that. At least 100 pulses / motor step. And you need to have a control (pc), drive, motor-encoder combination that can do things together. This only works if the PC control talks to the step driver and together they can decide on slowing down, increasing, etc.
And if you consider more than one axis, which all move synchronized, you see why it easily gets problematic.
I think the best way to make a stepper system work is to design it properly, so that it's never close to areas where it can stall or lose steps.
If you have encoders/linear encoders and you can watch if an motor stalls it's even better. IMO a motor will only stall if you had a collision, or something else which isn't part of a regular machine operation cycle.
Regards,
Alex
