Collision Errors on Modicon Momentum

[vodoca 0]

Hello, We have Modicon Momentum PLC and there is about 80 millions of Collision Errors (and growing cca 1.4 error per second) logged in Ethernet Statistics. In the manual they write only "This field should allways be 0". It is working for us though. We only have one problem that two machines on ADO 740 50 with ENT 110-01 stop simultaneously and irregularrly without any appearent reason. It is happening now for about 6 months, but I have calcluated that this collision errors accumulated over cca 2 years. My question is, what "Collision Errors" mean and could that by any chance make two particular machines on one ADO 740 50 module go off simultaneously? Edited 17 Aug 2015 by vodoca

[Quant 21]

Hello Vodoca, may be it's only firmware issue of M1E CCC, so that firmware update of CCC is needed. Cpuld you tell me which firmware has CCC9x0?

[vodoca 0]

It says only Exec ID: 899 - is it the firmware version? I found nowhere any hints how to perform firmware update.

[Quant 21]

never saw such ID. The hardware id of M1E should be 000A 0003. The current firmware version of 171CCC9x0xxx is 1.31. Are you using concept softwe? Then you will find also ExecLoader, which can read/write firmware to the M1E.

[Crossbow 295]

Typically collisions on an Ethernet network are caused by network overloading. Make sure you do not have any Ethernet hubs, but be sure to always use Ethernet switches. Hubs are dumb devices which simply retransmit all data, while switches will learn the devices attached to each port and only forward messages to the appropriate ports. This reduces the number of packets on the network and the chance for a collision. Also make sure the switches are compatible with 100MB traffic. Older devices could only talk at 10MB, so your network bandwidth may be used up because the network is not running at full speed. What other devices are connected on this Ethernet network? Office PCs? Printers? Internet access? Any non-PLC traffic on the network can always cause lag, just like it slows down your Internet connection when someone is downloading a large file. Industrial Ethernet Book has more info if you are not familiar with how Ethernet works. http://www.iebmedia.com/

[vodoca 0]

We have switches and an optical network dedicated to PLC (+ visualization PC) ethernet traffic. I guess it is pretty difficult to buy a hub nowadays. Among PLC and modules there are also 2 PCs in the network, one with 3 NICs and one with 2 NICs. The secondary network (wired separately) is dedicated to other than PLC traffic (remote desktop, other devices). 1 PC is connected using 3rd NIC to the routed VPN network. But the question is "What mean Total collisions and why it should be allways 0" As per here there are different types of ethernet network collisions and not all of them should be allways 0. http://www.cisco.com/c/en/us/support/docs/interfaces-modules/port-adapters/12768-eth-collisions.html#topic1 Edited 24 Aug 2015 by vodoca

[vodoca 0]

I got an answer from Schneider technician that it rather indicates a network problem than a firmware error. I am unable to find out the firmware version now, since Exec Loader requires PLC to be stopped even to show info. I have also found out that Momentum has half duplex meaning it cannot transmit and receive simultaneously. Still I am not sure if we should somehow deal with the problem (about 1 millon collisions are incremented per week) and first of all I do not know - does it mean normal ethernet collisions and why should it be 0 when collisions are part of ethernet network by design.

[Quant 21]

yes, to read firmware information from M1 the PLC will be stopped. Here is the firmware on my CCC:

[vodoca 0]

I have found that in Ethernet statistics there is a Firmware version field. We have 1.23. We have measured and tested optical network and replaced one brown looking switch and now we have 10 collisions per second measured on Momentum. But our problem did not re-appear since then, it seems to be better even with worse collision rate, so I suppose that the assertion that "This number should be allways 0" is wrong. When all the network devices except one STB NIP 2212 island were disconnected from the PLC, the collision rate went down to cca 1 collision per minute, but did not disappear entirely. We plan to replace normal switches with manageable ones to further improve this problem. Also I have read the manual for the newer Momentum, in the old momentum it is stated that: This field contains the total number of transmit collisions. Edited 16 Sep 2015 by vodoca

[vodoca 0]

I have created test environment and I must confirm that Collision errors on older Momentum are quite normal and probably harmless. I have one regular office switch and PLC connected only with one PC, and even then I receive collision errors. "This number should be allways 0" is valid probably only for newer Unity Momentum processor with 100Mbs full duplex auto-negotiation.