Replace fiber run with a Wireless Network?

[IN2-H2O]

We have a remote Flex I/O rack that used to be fed via a multi fiber cable. Since the fiber run has been damaged several times, we switched to a wireless device to run the HMI and have ditched the flex I/O rack temporarily.I am now being challenged to find a way to get the flex I/O functioning again.

I have suggested replacing the flex I/O rack with a micrologix controller and passing the inputs thru the HMI with DDE, as I have done in the past. It is simple I/O, inputs only (2-16 input modules).

It was also suggested that there may be a way to keep the existing flex I/O rack and simply replace the fiber adapter module with a wireless adapter or other device that would function over the wireless network and keep the network topography the same.

I would greatly appreciate if anyone has any experience with this or could suggest an alternative.

Our PLC's are all Contrologix with ethernet, controlnet and devicenet networks available.

 

[Ken Roach]

Which network adapter is the FLEX I/O chassis using now ? I don't think there's ever been a FLEX adapter with fiber onboard, so it could be ControlNet over fiber or Ethernet over fiber.

If you're using a 1794-AENT Ethernet adapter, you can certainly run it over wireless. The reliability of the connection and the load you put on that wireless link are the challenges.

ProSoft's WiFi hardware is both physically robust and well tuned to industrial protocols like EtherNet/IP that rely on cyclic data exchange.

What is the wireless hardware being used for the HMI now ? Which version of ControlLogix firmware are you running ? Are the ControlLogix running any I/O over Ethernet right now ?

 

[IN2-H2O]

We are using a 1794-AC15/C controlnet adapter that goes from controlnet to fiber thru a pairing of 1786 CET Repeaters, 1786 RPFM Fiber repeater and then back to controlnet with a 1756-CNB/D communication adapter in the main backplane.

Wireless is currently run through a pair of Linksys WAP54G wireless routers but will soon be converted to more robust and secure devices by head office in the next couple of months. (They are 10 times the price so they better be more robust.)

Controllers are curently on V15.4 but will be pushed to V19 during our next major outage. All I/O is run over Controlnet/fiber right now. Only Ethernet being used is for the actual process network between PLC's and HMI's.

My main fear with wireless I/O is that someone will reschedule the I/O when the remote rack is powered down and unavailable. It's non critical I/O so I guess it doesn't really matter but I like to keep the process as bullet proof as I can so it can be worked on without getting me out of bed in the middle of the night.

 

[Ken Roach]

The good news is that EtherNet/IP I/O connections don't use a schedule like ControlNet does. As long as they're in the ControlLogix I/O tree, the controller makes connections to them independently.

The principal issue is that most WiFi systems don't deal with small, frequent packets of data as well as they deal with typical Internet traffic, which is usually relatively large packets or streams of data. A small connection like this might work fine with a Linksys WiFi router, but larger or faster amounts of data might not.

In my opinion, Linksys is the absolute minimum product for any industrial application. It's like getting one of those Kraft cheese and crackers packets: technically it's lunch.

The most important thing when introducing wireless into an industrial Ethernet network is to make sure you can filter or restrict the traffic that goes onto the wireless system; you don't want broadcast or multicast data that doesn't need to go onto the wire to get there. That means a decent infrastructure of managed switches and a good understanding of how the various types of Ethernet data exchange work.

In V19 you will be able to create a "unicast" connection over EtherNet/IP so the cyclic data exchange won't use multicast. There's less flexibility (no multi-consumer connections) but it's easier to configure and maintain the Ethernet switch configurations if you don't have to deal with Multicast.

You could install a 1794-AENT adapter on the FLEX I/O and because it's just input data you could test it out by sending MSG instructions to the Input Assembly. That wouldn't require any online I/O connection creation.

How fast does this I/O change state ? If there are momentary pushbuttons, that's going to determine how fast you need to poll the FLEX adapter. If it's just state sensors or selector switches, you don't need to poll as fast.

 

[IN2-H2O]

The I/O is connected to pushbuttons on a set of joysticks so it will require a fairly fast response. Our corporate IT is overdesigning the system as we speak so the network will be the least of our worries. They are replacing our $900 worth of Cisco devices with $23K worth so it better handle it.

I am going to order a couple of the 1794-AENT devices and start playing with them in the next couple of weeks. They sound like the best way to go and were exactly what I had in mind so thank you for that advice. I will be sure to post back my results as it progresses..

 

[Ken Roach]

Hard-won experience tells me that cost and sophistication aren't necessarily predictors of reliability with wireless gear.

Make sure that both you and your IT department understand and measure the effect of frequent UDP packet transmission.

EtherNet/IP is very easy to measure for reliability, since every I/O packet includes a Sequence Number. Both the Connection Statistics in the 1756-ENBT and 1794-AENT and the traffic itself (get ready to learn about Wireshark) can help you determine if all of your packets are getting through correctly.