Re: List question

Ron Ginger wrote:

> From: Ron Ginger <ginger@...>
>
> Well, I sure drew out the lurkers with that one :-)
>
> and it seems to have opened up a couple new threads as well, so great,
> lets forget splitting and keep talking.
>
> The question of coordinating 2 (or even 3) PIC servos- or for that
> matter any servo is one that has bothered me. There are several servo
> boards Ive sen that are single axis, with serial port control. Can these
> be used for coordinated movements? I notice the PIC has a command to set
> the time of a movement, so I asume you could calculate the longest time
> and program both boards to that time.
>
> I would really like to hear from someone that has done this

Well, I really haven't DONE it, but I've worked with several systems
that do it, and I have picked some of the intricate code apart.
Clearly, one processor that can do it all makes it easiest.
That's what EMC does now. Since a single routine generates
all waypoints on the toolpath, and then passes the positions it
expects all axes to be at, at every tick of the trajectory generator's
clock, the axes should stay well in sync under normal conditions.
Anything where the separate axes have their own system clocks
could get out of sync, and at least could be 1/2 a tick out at any
moment. If the servo loop runs fast enough, this error will be small.
HP makes a real neat servo chip, not too different from the LM628,
but much cheaper. BUT, it doesn't have fractions in the velocity,
meaning that, say, X can move at 1/2, 1/3, 1/4, etc. the speed of
Y. But, X can't move at .53725 times the speed of Y! Well,
that's unusable for multi-axis work. You couldn't even do a decent
circle with it. Most of these things don't use time, they use velocity.
So, the main program computes velocity such that all the moves
take the same time. An easy computation, although you also need
to scale the acceleration, so all axes accelerate for the same time
before coming to their respective constant velocities.

Jon