[etherlab-dev] Slave state change from SAFEOP to OP without PDO update

[Kim H. Madsen]

Hello,

I have noticed an issue with the etherlab EtherCAT master. If the slaves a split into two domains (with different update cycles) then the state change SAFEOP->OP can be requested by the master without a slave have received output data. This makes the slave stay in SAFEOP with an error: EtherCAT ERROR 0-19: AL status message 0x001B: "Sync manager watchdog".

Is this a bug or a missing

[Florian Pose]

Post by Kim H. Madsen
I have noticed an issue with the etherlab EtherCAT master. If the
slaves a split into two domains (with different update cycles)
then the state change SAFEOP->OP can be requested by the master
without a slave have received output data. This makes the slave
stay in SAFEOP with an error: EtherCAT ERROR 0-19: AL status
message 0x001B: "Sync manager watchdog".
Is this a bug or a missing feature?

In general, it is recommended to start sending cyclic process data
immediately after ecrt_master_activate(). What are the sample times?

- --
Viele Grüße,
Florian

[Kim H. Madsen]

Later we found that also older TwinCAT versions have this problem. So we have solved the problem in the slave code by deferring the acknowledge of the state change and only throw the error if the slave haven't received output date within one second.

Best regards
Kim

-----Original Message-----
From: etherlab-dev-***@etherlab.org [mailto:etherlab-dev-***@etherlab.org] On Behalf Of Florian Pose
Sent: 4. januar 2013 11:03
To: etherlab-***@etherlab.org
Subject: Re: [etherlab-dev] Slave state change from SAFEOP to OP without PDO update

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Kim H. Madsen
I have noticed an issue with the etherlab EtherCAT master. If the
slaves a split into two domains (with different update cycles) then
the state change SAFEOP->OP can be requested by the master without a
slave have received output data. This makes the slave stay in SAFEOP
with an error: EtherCAT ERROR 0-19: AL status message 0x001B: "Sync
manager watchdog".
Is this a bug or a missing feature?

In general, it is recommended to start sending cyclic process data immediately after ecrt_master_activate(). What are the sample times?

- --
Viele Grüße,
Florian
-----BEGIN PGP SIGNATURE-----
Comment: Using GnuPG with undefined - http://www.enigmail.net/

iEYEARECAAYFAlDmqMUACgkQABFOIMygR8w+yACePBapkWTplk0xT1V5Wroh8ZR8
bdoAoJ9Lgs3Ugh4GOqWwka1EKiy6RYwt
=eYl0
-----END PGP SIGNATURE-----
_______________________________________________
etherlab-dev mailing list
etherlab-***@etherlab.org
http://li