Re: mill threading, Acceleration, Axis

Dave, Ballendo, Jon and the list,

Sigmoid? is that what you call the curve? I think mine looks like
that. I have a table of delays that correspond to a table of step
rates. I just walk the rate table and calc how many steps to let go by
before bumping up to the next rate in the table.

I pre-calc the deceleration point (I know the number of acceleration
steps to go from starting rate to ending rate for a trapezoid ramp),
continually "watch" for the next point to bump up, after the final rate
is reached, no bumping until the deceleration point, where I start
looking for decel points. Triangular acceleration, just start decel at
the midpoint.

OK, it's not perfect yet, but it's working! Only doing single axis move
for the rotary table or a linear axis, (not that I won't keep expanding
the code!).

I think MY shop processor IS computationally challenged! Certainly the
little hand held I want to use is!

Alan


dave engvall wrote:

>
> ballendo@... wrote:
>
> > Alan,
> >
> > Jon already gave you some good advice. I'll add in a different
> > direction...
> >
> > The "best" acceleration "curve" for a stepper motor is generally
> > considered to be the "S" curve. This is where we start acceleration
> > slowly, then accelerate the acceleration (yep, read that one twice!),
> > and then do the reverse to slowly move into the desired step rate.
> >
> > It's two exponential curves back-to-back. The benefit here is that
> > the acceleration RATE is low at the beginning and end of
> > the "ramp"(curve), where the inertia of the load is against us. And
> > the acceleration RATE is HIGH in the "middle" of the ramp, so we get
> > to our desired rate as quickly as possible.
>
> Do you gain anything by tableizing the sigmoid curve in normalized form and just scaling it....or are the processors today so fast
> they are not challanged?
>
> Dave