Newsgroups: comp.robotics
Path: brunix!uunet!cs.utexas.edu!rice!caesar.rice.edu!adnan
From: adnan@caesar.rice.edu (ADNAN Sarmad)
Subject: Re: Motion Controller Chips
Message-ID: <1992Mar25.002915.14415@rice.edu>
Sender: news@rice.edu (News)
Organization: Rice University
References: <2431@ucl-cs.uucp>
Date: Wed, 25 Mar 1992 00:29:15 GMT
Lines: 49


>each side of the chassis. The wheels have optical encoders which provide fe
>feedback to two HP1100 Motion Controller chips. These chips are supposed to
>monitor and adjust the speed, acceleration and position of the wheels to keep
>the robot moving in a straight line, etc.

>If you can help, I'd be eternally grateful.

Okay since you are going to be eternally grateful and that probably 
is long time.

>(a)	I'm using 'trapezoid profile mode' which calculates and implements
>	acceleration/deceleration profiles to get to a specified point. This
>	works pretty well except that I can find no way to interrupt this
>	planned move (e.g. emergency stops or changes of plan) without just
>	shutting off the motors and losing control of the wheels.

In order to turn off trapezoid profile mode 
either reset the F0 flag in software or use the limit switch input in 
software

>(b)	There appears to be no way to tell whether the movement is complete
>	other than polling the chips to check whether the actual position
>	is equal (or close enough to) the final position. Something
>	interrupt-driven would be more useful.

In order to do interupts you have to wire the prof line (pin 12) to an 
interupt input on your computer. 

>(c)	The controllers tend to stop a bit short of the target and leave the
>	motors whining with frustration at not quite being where they want
>	to be. It would seem to make sense for such a chip to try to increase
>	the power gradually until the motors reached their destinations.

This is a problem in your selection of the control constants in your 
software, use of the correct settings of the gain, pole and zero would 
cure this. Also use of a higher resolution encoder would help minimize 
this.  HP makes some 1000+ count encoders that are ideal for the 
HCTL-1100.  

>Any ideas/experience/suggestions will be very welcome.

As  for your request for a better chip, I do not know of any other
single chip commercial solutions that are in the same price-range and
offer better control.

--

                  Sarmad Adnan (adnan@rice.edu)
