DPRG List  

[DPRG] Please speak to me about SR518/Dynamixel servo deadzone/deadband, why I'd want it, and what to do with it

Subject: [DPRG] Please speak to me about SR518/Dynamixel servo deadzone/deadband, why I'd want it, and what to do with it
From: dlc dlc at frii.com
Date: Sun Jul 21 12:35:28 CDT 2013

I think that you've nailed it.  Noting in this world is perfect. The 
dead zone keeps a servo from continually oscillating and from responding 
to minute, unimportant perturbations.  "Old" hobby servos are analog 
which makes a dead zone difficult to define.  The better hobby servos 
that are digital have dead zones defined, and even programmable.  Good 
hobby RC transmitters allow you to program dead zones and "dead-ish" 
zones (exponential steering for example).  When you have infinite power 
(and robust gearing) or short operational times you can eliminate or 
minimize the dead zone (think milling machines or cruise missiles) when 
you want insane accuracy.  It has been my experience when programming 
servo algorithms that I use dead zones more often than not.  With cheap 
hobby servos I program the dead zone into the joystick since I can't do 
it on the servo.

As usual, YMMV, this is just my experience with servos and dead zones.


On 7/21/13 11:01 AM, Paul Bouchier wrote:
> Hi David.
> I would think a small amount of deadzone would be desirable in many 
> applications to conserve power & wear. It doesn't seem it should be 
> large, but it should be non-zero, otherwise the servo will "hunt" 
> around the desired setpoint & may never settle, OR the slightest 
> disturbance/flexure of the mechanical parts may kick it off into 
> re-seeking to the target location. If the target location is typically 
> in one spot (e.g. straight ahead for steering), this will tend to wear 
> the gears in one spot, and will consume more power than is necessary, 
> which is an issue for battery-operated machines.
> I suspect the reason for this feature that's not seen on typical hobby 
> servos is that dynamixels have much higher accuracy, so would be more 
> prone to this kind of issue. Regular servos may have a fixed deadzone 
> -- I don't know. Copied the list in case anyone else has more informed 
> comment.
> Paul
> -----Original Message-----
> *From:* David Ackley [mailto:dackley at verizon.net]
> *Sent:* Thursday, July 18, 2013 9:12 PM
> *To:* Paul Bouchier
> *Subject:* Please speak to me about SR518/Dynamixel servo 
> deadzone/deadband, why I'd want it, and what to do with it
> Hi Paul
> I'm finally starting to dig into the details of the SR518/Dynamixel 
> servos and discovered that they seem to have a large 
> programmable/non-zero deadzone/deadband. To me, this programmable 
> deadzone/deadband appears to be functionally similar to gear backlash 
> which one tries to avoid and it seems to be something that would make 
> smooth/accurate motion control almost impossible. What is it for, why 
> would I want to have it, and how do I get 
> smooth/accurate/non-oscillating motion control with a servo that has a 
> deadzone/deadband?
> I'm confused -- but then I guess that I get more easily confused these 
> days =|;-{
> Thanks
> Dave
> _______________________________________________
> DPRGlist mailing list
> DPRGlist at dprg.org
> http://list.dprg.org/mailman/listinfo/dprglist

Dennis Clark          TTT Enterprises

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.dprg.org/pipermail/dprglist/attachments/20130721/26dab56d/attachment.html 

More information about the DPRG mailing list