Re: Re: motion control chips

Alan Rothenbush wrote:

> From: Alan Rothenbush <beer@...>
>
> > From: "Elliot Burke" <elliot@...>
> > Subject: motion control chips
> >
> > Forgive me if this has already been mentioned. There has been a little talk
> > about motion control boards, conparing the low priced spread with Galil
> > boards. Well, while browsing the HP optoelectronic catalog (really, I don't
> > have much time to waste), I came across the HCTL-1100 General Purpose Motion
> > Control IC.
> > It has lots of features, is intended to control stepper , DC, or DC
> > brushless motors, has encoder inputs. It does all the motion control
> > calculations, and can be synchronized with other HCTL-1100's.
> > It has a 24 bit counter, which beats most of the decoders I've looked at.
> >
> > This may be the right chip for a motion control board.
> >
> > Has anyone looked at this part?

Yes, I've looked at it seriously. It has a MASSIVE drawback in multi-axis
systems. It is an integer only device, meaning that extremely slow movements
are difficult, and setting up to have one axis move fast and one move much
slower is quite difficult. When making a 2-axis linear interpolation that runs
almost (but not quite) parallel to one axis, the two axes make their moves
such that they both accelerate simultaneously to the desired velocity, both
reaching the desired velocity at the exact same time. they then run for
the exact same time until they both start to decelerate at the same moment,
and then finally reach zero velocity simultaneously. This requires either
floating point or fixed point arithmetic. The National Semi LM628 and 629
series have BOTH a 24-bit integer and a 24-bit fixed point fraction for
both velocity and acceleration. This makes the type of coordinated moves
mentioned above possible.

Even coordinating the LM629 may not be so simple. It should do coordinated
linear moves just fine, but it is not clear how you make it do something like
a circular interpolation. You can keep jamming new velocities into the motion
control chip, but how can you be sure the 2 axes are really still in sync?

Jon