Has anyone attempted to implement an ethercat master stack on the ICE boards? Or if not, does anyone have any sample code to program the PRU as an Ethernet MAC layer?
I know this sounds like a silly thing to do, but we have a requirement just come up to support multiple masters on a single EtherCAT ring. Only one master will ever be available at a time, but the nature of EtherCAT is that you can't turn off devices - hence the 'redundant' master needs to pretend to be a slave until switched in. Hence the need to support a master stack on a slave device.
Suggestions welcome!
Mat
Can you please tell if there will be some time allowed for the EtherCAT slave to switch over and become a master.
It is certainly possible to use the PRU as a classic 802.3 MAC but to switch to this function from an EtherCAT slave function, firmware swap is likely to be required and that needs further thought.
Thanks.
Maneesh
Hi Maneesh,
time isn't a worry. This is a last-resort fallback position, its just that our hardware is inaccessable so we would look to bring the reserve master up remotely (easiest way is a long-distance re-flashing exercise).
Mat,
Technically, it is possible to accomplish what you are saying. The PRU MII ports will be connected to EtherCAT network and operate as slave in the normal mode. When there is a particular event that triggers for it to become a master, then, in a few seconds PRUs can be loaded with a different firmware to make the MII ports as regular IEEE 802.3 Ethernet ports. Such switch-over will disrupt the ongoing traffic on EtherCAT network... do you have a plan on that side of things? Or may be since the main master is already down, it is acceptable that the traffic is disrupted. And what about the software/functions that the EtherCAT master is itself doing at application level? Are those replicated on the stand-by master or is the stand-by master only doing a very limited amount of work?
The Ethernet MAC functionality itself is definitely possible on PRUs and may become available in future... but I can't say if or when.
Regards,
we're assuming any failure of the original master will kill the bus, hence the new master will come up 'fresh' - re-initialising the slaves as required.Same again on the application side - there won't be any handover.
So, sound like a plan is forming. Just a question about how much needs developing vs what we can get as sample code (forgive me here, but we're doing early-stage prototypes, so need to minimise any development work). The two areas:
PRU MII code
Master stack application code.
Would be really good to know if either of these is on the roadmap anywhere - even for the future?
Both MAC support and Master support are likely to become available in future but we do not have a timeline to share right now... sorry about that.
for EtherCAT master stack code you may either contact Beckhoff/EtherCAT technology group or look in the open source world. There are some available. Also ETG forums are a good source for such info or questions.
Regards.