[etherlab-users] Is Ethercat Master Enough to Control Servo

Raz Ben Yehuda rbenyehuda at manz.com
Thu Sep 13 08:09:29 CEST 2012

On Thu, 2012-09-13 at 00:03 +0200, Thomas Nelson wrote:
> On Sep 5, 2012, at 2:13 AM, Raz Ben Yehuda wrote:
> > On Sun, 2012-09-02 at 11:47 +0200, takeshi ikeya wrote:
> > > Hi Tahir!
> > > I think ENOUGH..
> > > If you need quick response, you'd beter use it under RTAI (Linux).
> > > 
> > I must add that did not find the current design suitable for moving
> > servo drives. For this reason I modified etherlab to support slave
> > DC
> > bus time ( currently etherlab is App time based).
> > 
> > > 
> > >  takeshi.ikeya at gmail.com
> > 
> > -- 
> > https://sites.google.com/site/ironspeedlinux/
> > 
> I have to disagree with these blanket assessments.  Much is dependent
> on the hardware/software platform that the Etherlab EtherCAT
> installation is deployed on, and this must be taken into consideration
> to realistically assess what the resulting system is capable of.
> In my case, I have a client that is successfully operating two Copley
> AE2 dual axis servo drives in a high-precision robotics system at a
> 250 uS cycle time (4 KHz) on an 1.6 GHz Intel Atom N270 with a 82574L
> Gb master interface under Linux 3.2.28-rt42 (RT_PREEMPT).  Jitter is <
> 10 uS even with the system performing other non-RT tasks, such as the
> network service for the remote application control/status UI.

What timer do you use ? you cannot nonosleep. you use the hardware i am
using. and i connected scope to my parallel interface and saw in-scope
how messy the clock is.

Did you profile you code  ?  a single context switch costs about 10us on
these machines- i know this for a fact. if you did not do the cycle task
in kernel space in the interrupt context you are likely to have more
than 50us jitter. use ftrace and see for your self.
if you did the cycle task in interrupt context then you must have used
RTC. RTC is emulated by HPET in intel based systems, HPET has 5% jitter
as-is , and I am referring to the non-emulated hpet ( ie hpet0 ).
check this link:

> Using RTAI or Xenomai was out of the question because we needed to use
> off-the-shelf industrial x86 SBCs and only the mainline kernels have
> the HW support required.
> Tom Nelson
> Consulting Engineer
> Granite Computer Sciences, LLC
> Milford, NH USA


More information about the Etherlab-users mailing list