[etherlab-users] Knowing when the packet has finished cycle

[Jun Yuan]

Hallo BigNose,

I've just read the documentation of igh ethercat master, especially the
chapter *4.2 Native EtherCAT Device Drivers*.
It says, there is no need to have an interrupt, because the "communication
is highly deterministic: A
frame is sent and will be received again after a constant time" in the same
network. Bad that I didn't read it before.
So now I think, what we need to do, is to find out that constant time for
our specific network by trial and error.

Best regards,

Jun

---------- Forwarded message ----------
> From: +BigNose <bignose at bluewin.ch>
> To: etherlab-users <etherlab-users at etherlab.org>
> Cc:
> Date: Wed, 14 Dec 2011 10:05:13 +0100
> Subject: Re: [etherlab-users] Knowing when the packet has finished cycle
> Hallo Shahbaz,
>
> I encountered the exact same problem. So I'm happy, this topic get's
> discussed.
>
> The problem is, that in a realtime control application, it is common,
> that you wanna have the sensor values as fresh as possible. And as
> Ethercat is very fast, this yields a good starting point.
>
> But just like you, I also encountered the problem, of how can you get
> ASAP the data from the master.
>
> IMHO the main problem here is, that the interrupt capability of the
> driver has been dropped. So that now it has to be polled.
>
> BTW: Why exactly has the interrupt capability been dropped? What is
> the advantage of this?
>

"The interrupt-less operation is desirable, because hardware interrupts are
not con-
ducive in improving the driver’s realtime behaviour: Their indeterministic
incidences
contribute to increasing the jitter. Besides, if a realtime extension (like
RTAI) is used,
some additional effort would have to be made to prioritize interrupts." -
chapter 4.2

To always poll, if the master has received, is IMHO a bit ressource
> hungry, and as you said, I wouldn't know how to implement this
> anyway, as you push the Master into an error, if you do it.
> A maybe workable solution would be, as you said, if one can surely
> determine, after how much time, one can surely 100% read the master.
>
> So I'm really very interested in this topic.
>
> Thank you
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.etherlab.org/pipermail/etherlab-users/attachments/20111214/28f8a700/attachment-0005.htm>