Re: A Z80 guy needs help!

Hi,

This is a for a "G-Code in, step/direction out" project that will
take about a year to complete. All motion is vector based and works
on the assumption that position is the running sum of precise
velocities, each applied for a precise period of time (1/1,024 of a
second).

For circular interpolation this would be:

Xv = R sin (theta + delta) - R sin (theta)
Yv = R cos (theta + delta) - R cos (theta)

Replace (theta) with (theta + delta) for next time period.

Delta is the angular distance to be moved in 1 time period and is
dependent on the value for R and the programmed vector velocity.

This requires solving for sin/cos in the shortest time possible to a
24 bit accuracy. I am using a Chebysheff polynomial expansion
coefficients to converge the series in the fewest possible terms. For
this I have a "roll your own" math package in assembly to do it.

It seems I will have to learn C as well (learning makes my head
hurt!).

Mariss

--- In CAD_CAM_EDM_DRO@y..., Russell Shaw <rjshaw@i...> wrote:
>
>
> Your samples may be rapid, but the calculations only need to
> be done at a rate that suits the control loop response time.
> If you keep the calculation rate in a separate slower loop
> compared to the parts that keeps track of the instantaneous
> servo movement, slower code and more efficient use of the
> cpu can be had. It depends on what the exact control methods
> are.
>