Re: [CAD_CAM_EDM_DRO] CNC division

Ian Wright wrote:

> Hi,
>
> I've been thinking about running a dividing head / rotary table under cnc
> control and was wondering how the division is usually arranged. As most
> divisions will not be whole numbers of steps on a stepper or encoder, you
> obviously can't just tell the thing to round up or down the error as you
> will likely finish up with a 364 1/2 tooth wheel instead of a 365 one (for a
> year calendar on a clock). So, with a slight bias towards EMC, can you have
> the cnc software do the necessary computation to get closest divisions or
> would you have to do it with a hand-calculated canned cycle for each number
> of divisions - or perhaps some other way?
>

It would seem that you would have to increase the mechanical resolution
to a degree that eliminated the problem or at least reduced it to an acceptable
value. Your example above implies that your resolution is approx 1/700 of
the rotation of the table. This is very coarse. My control (AB8400) has a
resolution of 1/1000 deg when the axis is run in the rotary mode. This
means I would want to have a mechanical resolution (encoder feedback)
at least equal to this to minimize errors. I have a 90:1 ratio table (4 deg
per turn of the handle) which I plan to drive with a servo motor at a 2:1
reduction. The encoder (500 line - 2000 quad counts) is on the motor giving
me 4000 counts per turn of the input shaft for a resolution of .001 deg per
count.
If you were using the usual 200 step motor direct connected to the input
of a 90:1 table, each degree would be 50 steps. If you half stepped the
motor, you are now down to 1/100 degree per step. If you add in some
timing belt reduction between the motor and table, you can increase the
resolution even further.
Anyway, that is my $.02 worth on the subject....

Larry...