[etherlab-users] "EtherCAT WARNING: <N> datagrams TIMED OUT!" problem

[Anthony Sharobaiko]

On Thu, Apr 10, 2008 at 09:16:26AM +0200, Florian Pose wrote:
> Hi,
> 
> this is really strange. This combination seemed to work perfectly
> here. Let's try to eliminate some possible causes...

This is all strange indeed. It all works now.
Coupler was reconnected back to TwinCAT machine in order to capture ethercat
packets with wireshark (windows version of tcpdump) during "correct"
operation. Wireshark has failed to capture anything, but it seems that something was
changed in the coupler through TwinCAT (this is the only explanation I have).

Both problems are now gone and I can not reproduce them any longer. If I get
another (new and clean =)) EK1100, I'll try.


ec_mini example worked for me. Btw, was ecdb.h meant to be a complete database?
If yes, here is a minor addition:

diff -ruN include.orig/ecdb.h include/ecdb.h
--- include.orig/ecdb.h	2008-04-11 12:59:16.000000000 +0400
+++ include/ecdb.h	2008-04-11 13:02:48.000000000 +0400
@@ -71,6 +71,10 @@
 #define Beckhoff_EL2004_PDO_Outputs 0x3001, 1
 #define Beckhoff_EL2004_Outputs Beckhoff_EL2004, Beckhoff_EL2004_PDO_Outputs 
 
+#define Beckhoff_EL2008 0x00000002, 0x07D83052
+#define Beckhoff_EL2008_PDO_Outputs 0x7000, 1
+#define Beckhoff_EL2008_Outputs Beckhoff_EL2008, Beckhoff_EL2008_PDO_Outputs
+
 #define Beckhoff_EL2032 0x00000002, 0x07F03052
 #define Beckhoff_EL2032_PDO_Outputs 0x3001, 1
 #define Beckhoff_EL2032_Outputs Beckhoff_EL2032, Beckhoff_EL2032_PDO_Outputs

> On Wed, Apr 09, 2008 at 04:58:20PM +0400, anthony wrote:
> > datagram timeout warnings appear when I switch on the only slave
> > device (EK1100 coupler) connected directly to realtek NIC.
> > /sys/ethercat/master0/info shows same number of sent and lost frames.
> > Number of received frames is 0.
> Does the activity indicator at the EK1100 show, that the frames are
> received at least at the slave?

AFAIR, indicator didn't flash at that time.

> > Also note that starting/stopping ethercat master somehow changes
> > realtek card MAC address (00:11:6B:95:32:FD -> ff:ff:db:b6:ff:ff),
> > preventing subsequent launches without setting MASTER0_DEVICE to FFs.
> 
> I never cam across something like this. Do you have the possibility to
> try another network adapter type (like e1000) for the same kernel, or
> the same network adapter on a different kernel? I'd like to know if this
> is a problem of the network device driver.

I also have a spare e100 card (intel 8255x), but http://www.etherlab.org/en/ethercat/hardware.php
states that it is supported under 2.6.18 only (I was unable to find adeos
patch for this kernel version in rtai distributions). Are there any plans to
support it in newer kernels (like .19 =)) ?


-- 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.etherlab.org/pipermail/etherlab-users/attachments/20080411/59b577e3/attachment-0003.sig>