Re: [CAD_CAM_EDM_DRO] Linear vs. Rotary Enc.

John D. Bussema wrote:

>About backlash.. . . don't linear encoders deal with backlash (with regards
>to leadscrew lash and all other sources of mechanical slop) much more
>accurately than rotary's, being that they are located at the end of the
>mechanical system, essentially?
>

Yes, as far as the accuracy of the linear encoder's readout, this is
true. BUT,
if there is backlash between the motor and the linear encoder, then
there is a
position where the motor is not connected to the table movement, every
time the
motor reverses. This is a severe discontinuity of motion, and drives servo
systems that depend on a linear transfer function crazy!

> About EMC's control loop. . does EMC have a
>way to define an acceleration time constant? Would that effectively deal
>with this problem of the servos searching for position, if you were to
>lengthen the time they take to decelerate at the end of a movement in one
>direction?
>
>

No, once you have this discontinuity, there is nothing you can do after
the fact to
fix it. You can cover it up with deadband, but that is an AWFUL
solution. Time
constants would be worthless, as it would just make the table rock back and
forth slower. It would also make all other movements slower.

Jon