Ethernet/IP Comm Issues

We’ve got an assembly cell that’s controlled with a 1769-L35E Compact Logix processor. All remote I/O, HMIs and barcode scanners are communicating via Ethernet/IP. We have a total of 22 devices communicating via Ethernet/IP on this cell. The barcode scanners are Metrologic IS3480 with RS232 out. We’re using Realtime Automation gateways 435NBA to converter ASCII to PLC on the scanners. When we first built this cell we noticed a glitch from time to time where the gateways locked up and had to power them down to reset. We also noticed the RedLion G310 HMIs would drop comms and blank out for a second or two, but quickly returned. We made some conf. changes on the gateways and it seemed to resolve this issue until now. We recently increased production rates on this cell and now our comm. issue is back and in a big way. Last week we replaced our 24 port open hub with a 24 port Hirschman managed switch hoping this would shield the gateways and HMIs from having to deal with the I/O traffic, but it didn’t change a thing. I’m now thinking we need to upgrade from our current $250 barcode scanner to a higher end Ethernet/IP scanner in order to remove the gateways. My questions is, could the gateways also cause the blanking issue with the HMIs? It does seem like when the gateways lockup we notice the HMIs get worst. Any help or ideas would be greatly appreciated. Any recommendations on an affordable scanner?

Thanks,
Jim

We use several ethernet cards and hubs to prevent this type of event. For example our plc will contain 3 ethernet cards. These separate drive comms, plc to plc comms (also remote. I/O),and hmi's systems. Perhaps reducing the number of devices by installing a separate ring will help.

I would not use a "hub" anywhere on a system, and would only use a "switch".

Hubs do not do anything to prevent EtherNet "collisions", which can have a disastrous affect on the data transmission bandwidth.

Use switches for port expansion (e.g. place a switch in a panel, 1 port = PLC, 1 port = HMI, 1 port = to rest of system (via your managed switch), 1 port = free for programming laptop). These need not be "managed" switches. Hirschman make excellent 4-port DIN-rail switches suitable for this.

Mtech's advice to use separate EtherNet comms modules is not valid for your CompactLogix system.

Another thing to consider is the communications loading on your L35E controller. On the "Controller Properties", "Advanced" tab, there is a "System Overhead Timeslice" parameter (default = 20%), which specifies how much cpu time to allocate for system tasks, which includes comms.

If you haven't already done so, you could try increasing this value. Note that in firmware versions 13 and 15, this will have a detrimental effect on your application scan time, but from V16 and above you can set a higher timeslice %age, and check the box to "Run the continuous task", "During unused System Overhead Timeslice..". This is an optimal setting, releasing the processor back to the continuous task when there is nothing left to service in the "overhead" period.

HTH

Switch. Yes, that's what mean. Excuse my technical terminology. Just handing out suggestions regardless of operating systems. Best of luck.

Thanks guys for the reponse. Sounds like both of you feel it's a traffic issue. Actually the Unicom 24 port open hub I mentioned above was labeled as a switch, but was only $80. We purchased the $1,100 Hirschman switch as a managed switch. I've got 2 of the Unicom switches. Do you guys think it would be worth while running into the Hirschman from the PLC, then branch from the Hirschman to the 2 Unicom switches? We could connect all I/O to one Unicom switch and HMIs & barcode scanner gateways to the other Unicom switch. I can actually do this pretty quick tomorrow morning.

Thanks again,
Jim

First, do a simple test. Download Wireshark, and connect to one of the unused ports on the Hirschmann switch and capture a short trace.

If you see a lot of "ENIP" and "CIP" multicast traffic, that's a sure sign that the switch is not configured correctly and is not constraining multicast traffic using IGMP.

Hirschmann RS20's had me chasing my tail a few years ago until I read the user manual more carefully to learn you had to commit changes to NVRAM or else the'd be discarded on the next power cycle.

Ken,

I captured a 2 1/2 minutes log with wireshark and it had a lot of CIP and several TCP. I currently have it setup as I mentioed above. The PLC and PC are connected into the RS20 Hirschmann, then I'm branching from the Hirschmann to 2 of the unmanaged switches. One switch with 7 I/O Ethernet adapters only and the other with 6 HMIs & 3 barcode scanner gateways. The line has run over 5 hours this morning with no issues. It's the longest it's ran without a reset since we increased production rates. I'm current talking with John at Hirschmann and he is looking at the wireshark data I just captured.

Thanks for the help,
Jim