Powerflex 525 Multi-Drive comm issues

[stretch_af]

I had to replace 3 PowerFlex 40 multi-drives that each had 3 PF4M daisy chained off the DSI port. The PF40s had comm-e cards installed for ethernet. A 525s were installed. I updated the program, rebuilt the nodes, then downloaded the program. The machine has a 1768-L43 CompactLogix ver 20. Stratix 2500 controlling traffic.

All parameters were set in the 525s to replace what was in the PF40/Comm-e setup. Now every time I plug anything into the 525's ethernet port, I get a F105 fault [C Connect Err], all its children get a F81 [DSI Comm Loss].

The PLC won't take ownership of the drive, I can't use CCW over ethernet, faults too quick. CCW works through the DSI port.

IP address set
C123 - 19200
C169 - Ethernet/IP
C171-175 set and verified
Node address in the PF4Ms didn't change.

I feel like I'm missing something stupid.
Did I miss a parameter?

 

[Ken Roach]

It sounds more like a fault, damage, or wiring problem with the Ethernet port.

Is this a 525 built-in Ethernet port, or the 25-COMM-E2P two-port DLR ring capable interface ?

When you say you get a fault "every time you plug anything in", do you mean that you plug directly into the drive, or into the switch it's connected to ?

Any chance you've created a ring with non-DLR devices, which might really be a loop ?

 

[stretch_af]

I got things working, I am still unsure of my issue.

The 525s have the embedded ethernet. With only the PLC, an HMI, a PF755, the new 525 and my laptop plugged into the switch, the 525 would get the F105. Unplug everything but the 525 and my laptop, F105. Plug my laptop directly into the 525, F105. I couldn't keep the 525 on line long enough to get the PLC to connect to it.

On a whim, I connected my laptop to a different ethernet network the CompactLogix has. For some reason, with my laptop off the drives network, all the comm issues disappeared. I got online with the PLC, correlated the VFDs, all is good.

So my laptop was faulting some new PF525s but nothing else? I'm confused.

 

[Ken Roach]

Well, that's some weird ****.

I suppose I'd try to run Wireshark or maybe try a different NIC or dongle to figure out if somehow my PC interface was sending traffic out that interfered with the configuration or operation of that drive.

WEIRD.

 

[stretch_af]

Same issue again this morning. Plugging my laptop into the switch faulted the 525s, nothing else.

So apparently those 525s didn't like my Ethernet/IP driver, or how I had it setup. Deleted that driver, used the generic Ethernet driver. Everything is working now.

 

[Ken Roach]

Are you familiar with, or are you willing to try using, Wireshark ?

Years ago I had a similiar-but-not-the-same issue with some Windows machines and the EtherNet/IP driver.

Instead of sending a "List Identity" broadcast command every 1 second or so, it sent them every 2 or 3 milliseconds. Because every EtherNet/IP conforming device has to answer that message, some low-powered devices couldn't handle it and faulted their connection to other things or became "blind" to ordinary RSLinx communications.

I could never figure out if it was a Windows problem or a switch problem (handling Broadcast packets) or an RSLinx problem; I just knew there were combinations of hardware and configuration that didn't do it. And it was probably 15 years ago.

I will certainly understand if you prefer to mark the problem solved and move on. I just don't like to let unsatisfying conclusions like "the EtherNet/IP driver faults PowerFlex 525 drives" go unexplored.

 

[stretch_af]

I have heard of Wireshark, never had a need to use it however.
I will try it out, hopefully find a root cause.

 

[tom_stalcup]

I ran across this in one of the rockwell manuals, when I was getting ready to do some pf40 to 525 conversions. No idea if it has anything to do with your setup or not, but it's something that I've been mulling around in the back of my head in case an issue shows up.

"The embedded EtherNet/IP adapter is a multicast device. In most situations, an IGMP snooping (managed) switch is required. If more than one or two embedded EtherNet/IP adapters are connected to the switch, a managed switch is required—otherwise the drive may fault on a Net I/O Timeout network loss. The embedded EtherNet/IP adapter, RSLogix 5000 (version 18 or greater), and a ControlLogix or CompactLogix controller will support unicast. Unicast setup is required when adding the drive to the I/O. When all embedded EtherNet/IP adapters are set up as unicast devices, then an IGMP snooping (managed) switch is not needed."

 

[Ken Roach]

Tom, I think that advisory about multicast vs. unicast is essentially saying "this adapter is relatively low-powered and can't filter out large volumes of multicast traffic, so follow the classic design rules of EtherNet/IP network architecture and switch selection and configuration".

My speculation is that there might be something related to the Ethernet NIC that Stretch is using.

If there's something that causes it to flood the network with Duplicate MAC Detect packets, or with CIP List Identity commands, then maybe the PowerFlex 525's are faulting for similar reasons: there's too much broadcast or multicast traffic for them to handle.

 

[tom_stalcup]

Thanks Ken.

The other thing running around in the back of my head was the igmp snooping issue with cisco/stratix/whichever switches from a while back.

I'd triggered that particular issue on our network, when I connected a laptop to the controls network that had a network bridge setup to the wifi adaptor.