home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!comlab.ox.ac.uk!michael
- From: michael@uk.ac.oxford.robots (& Stevens)
- Newsgroups: comp.robotics
- Subject: Re: Stepper motor programs wanted
- Message-ID: <MICHAEL.92Sep9002510@lucrece.uk.ac.oxford>
- Date: 8 Sep 92 23:25:10 GMT
- References: <gate.LJuLqB1w165w@toz.buffalo.ny.us> <1992Sep8.202500.27325@oswego.Oswego.EDU>
- Organization: Dept. Engineering Science, Oxford University, UK
- Lines: 36
- In-reply-to: nacs4160@Oswego.EDU's message of 8 Sep 92 20:25:00 GMT
-
- In article <1992Sep8.202500.27325@oswego.Oswego.EDU> nacs4160@Oswego.EDU ( Marc F. Houde) writes:
- I'm looking for programs to run a stepper motor accurately at high speeds. It
- is to be used for a project in holography. Please send any info to:
-
- Three years ago I was also looking to control steppers. At the time I
- had varying velocity demands coming in to a drive motor, simple I
- thought. Open loop positioning, so all you have to do is work out from
- the velocity demand what the step interval should be.
-
- When implemented things actualy looked harder. The system had a fixed
- time quantisation interval (tick). This tick quantisation could be
- quite significant compared to the step period.
-
- An addition all this had to be done very quickly, the only time was on
- the interupt period of a simple 8 bit microprocessor. This made me
- think of Breshenham's line draw algorithem. Fast, integer only,
- naturally fitting to a quantisation problem.
-
- The stepper motor problem required that the velocity (gradient) could
- be varied on the fly, and could also be +ve, and -ve. Two descision
- variables are required to cope with this modified case. The big
- advantage of this was the algorithm has no residual error. If the
- velocity required doesn't quantise in time exactly the output will
- still run at the correct velocity 'on average'. The residual error is
- carried over to the next time step and added to whatever the new
- velocity is. Formally: The stepper will be exactly at the same
- position as the integral of the velocity demands to get position.
-
- At the time I did a quick literature survey to see if others had
- applied a Breshenham like method, but nothing came up. Has anyone else
- done something similar? Anyone interested in the algorithm?
-
-
- --
- Michael Stevens, Robotics Research Group,Dept of Engineering Science,Oxford,UK
- INTERNET: michael@robots.oxford.ac.uk
-