Society of Robots - Robot Forum

General Misc => Misc => Topic started by: Steve Joblin on October 19, 2007, 08:47:20 AM

Title: Motor Control Using Back-EMF techniques
Post by: Steve Joblin on October 19, 2007, 08:47:20 AM
I understand that a good method of motor control is to use back-emf.  I can't seem to find much about it via posts on these forums and/or Google.  Can someone point a newbie like me to a good source that provides an "idiots guide" and some samples to get me started?  It seems like a great technique for motor control, but I know of only one company (Acroname) that is using it in a product (http://www.acroname.com/robotics/parts/S11-3A-EMF-HBRIDGE.html (http://www.acroname.com/robotics/parts/S11-3A-EMF-HBRIDGE.html)). Why don't more hobby motor controllers utilize this technique?
 
Title: Re: Motor Control Using Back-EMF techniques
Post by: hgordon on October 19, 2007, 11:09:55 AM
We're using it, though we are measuring pulse widths using a "zero crossing" technique rather than measuring voltage.  You'll find some information on our explorations here -

    http://www.surveyor.com/cgi-bin/robot_journal.cgi/2007/06/10#099

The motor control board for the new version of our robot has this circuit for 2 motors with only minor modification - you'll find the schematic in the middle of the page of

    http://www.surveyor.com/blackfin/

Title: Re: Motor Control Using Back-EMF techniques
Post by: Admin on October 19, 2007, 11:53:12 AM
The back emf solution would only work for perfectly flat terrain . . . add slopes (or collisions, or change robot weight) and it breaks . . . note that back emf has error rates as well.

Ive seen it used in conjunction with encoders to detect slip (wheels spin, but no emf from torque).
Title: Re: Motor Control Using Back-EMF techniques
Post by: hgordon on October 19, 2007, 12:10:55 PM
The back emf solution would only work for perfectly flat terrain . . . add slopes (or collisions, or change robot weight) and it breaks . . . note that back emf has error rates as well.

It definitely requires some signal processing to get good data, but we only sample for a few milliseconds at a time, and this can be compensated for by slightly boosting power levels (PWM width).  There's no dead reckoning technique that will produce perfect results, but the benefit of back-emf is that it's REALLY cheap to implement - our parts cost for 2 channels is under $1.00.