Ethernet noise reduction strategies

I have a small CIP system that I'm having issues with. Basically, it consists of two ML1400s and a C-More HMI. Switch is located about 5' from the HMI, which is in a large SS enclosure that holds three VFDs and six motor starters. The Ethernet cable from the switch to the first PLC is about 100' long, and about 200' to the second PLC. The furthest PLC "runs the show". The closest PLC only reads the position of limit switches to validate valve position. CIP is started and monitored from the HMI. I read validation signal from the slave PLC every second (MSG over CH1, reading one word)

Here's the problem:

Every once in a while, I lose the validation signal from the PLC, which causes the system to stop. I set up a maintain timer and it works fine (latch currently holds for 7 seconds). The timer tricks the main PLC into thinking that switch position is validated.

I'm not fond of "tricking" the system, because moving a valve while running CIP could be expensive - read PRODUCT LOSS.

I've been trying to diagnose over a VPN connection, and several pumps are starting within a few seconds of each other, so I can't nail down which starter or VFD is causing the signal loss. To make things worse, I was involved in the project only after most of the pumps were wired, so I have no idea which may have ground loops, which wires may be run next to the Ethernet cables, etc.

I know this is a general question, but is there a recommended way to "filter" electrical" interference off the Ethernet cables?

I guess I could set up a serial network, too, but pulling cables isn't going to be fun.

I know of no 'filter'.

If the cable is run close to power wiring, you could re-route.

Otherwise, I've heard of others with similar problems moving to either shielded cable or fiber optic in this situation.

The old trick of winding the cable around a ferrite core doesn't do much for Ethernet. You really just have to physically separate the network cable from the noise source.

I've found network cables coiled up on top of drives, and run in conduit with VFD output.... bad news.

The MicroLogix 1400 has a pretty good "ES0" function file to show Ethernet port statistics, so you could look around in that to see if you start to accumulate errors at one particular time. It won't be easy but the results might help guide your onsite folks toward to problem.

As well as Dan's good advise. The Ethernet cable should be in it's own Conduit or Teck Cable. It should be routed away from Power Conductors in Cabinets as much as possible, and when it has to cross them, it should be at a Right Angle.

For your situation, you could try inserting another Ethernet Switch half way on you long run. You could also try putting snap on Ferrite Chokes at each end of the Ethernet Cable right before it plugs into the Switch or PLC.

Something like this.

http://www.radioshack.com/product/index.jsp?productId=2103222

Or this inline unit.

http://www.coilcraft.com/modjack.cfm

Stu....

The reason everyone loves ethernet is that ethernet's hardware layer is so well designed. Every single NIC card incorporates a 'magnetic' transformer that provides incredibly good isolation and common mode rejection.



It's my understanding that the 'magnetics' transformer was orginally it intended to prevent emitted noise (FCC), but the side benefits of common mode rejection and isolation for those of us who have been burdened with serial comm links, which are not as well designed, are fantastic.

Here's what Lantronix says about the magnetics in a blurb for those using Lantronix's embedded network interfaces where the remaining design is the connection to the outside:

Selecting LAN magnetics
For communications on a local area network, a magnetic filter module is required. The magnetics provide impedance matching, signal shaping and conditioning, high voltage isolation and
common-mode noise reduction. 10/100 Base-T Ethernet utilizes Unshielded Twisted Pair (UTP) transmission cable. The UTP wiring is prone to noise pickup that leads to conducted and radiated noise emission. The magnetics help to filter out the noise, provide good signal integrity and good electrical isolation.

The recommended Magnetics has a 1:1 transmit turns ratio, 1,500 volts isolation, center taps on both sides of the transformer, a choke on both the transmit/receive channels and has been designed for use with 10/100 Base-T Ethernet networks. The following are some additional factors to consider when selecting the Magnetics for LAN communications.
Common-Mode Choke.

For 10/100 Mbps communications, a common-mode choke is necessary for a system to pass FCC testing. The choke presents a high impedance to common-mode noise but a low impedance for differential-mode signals. For a WiPort application, both transmit and receive channels must have a choke and both chokes should be wired directly to the RJ-45 connector (see figure 1).

Common-Mode Rejection
Common-mode rejection is the ability to reject a signal which, referenced to ground, has the same amplitude and phase on both inputs. This signal, which is usually the result of noise or a small impedance mismatch, produces a small differential error voltage at the input terminals of the Ethernet. Subsequently, this error gets amplified right along with the desired communications signal. The magnetics plays a major role in knocking down this common-mode noise down to an acceptable level.

Lantronix Application Note: How to Connect a Lantronix Embedded Module to a Wired Ethernet Port
http://www.lantronix.com/pdf/appnotes/Connect-LTRX-Embed-Module-to-Wired-Ethernet_AN.pdf

My experience with ferrite cores was dramatic in the DC world, when signals from field transmitters would peg off scale when the tech's walkie talkie was keyed. The installation of ferrite cores dramatically reduced the impact of the RF on the DC, but that was an internal addition, not an external one.

I suspect that the ethernet packet is being swamped by normal mode noise and I'm just not sure how a ferrite coil at one end or the other can ameliorate that. I'd be glad to be proven wrong, though.

Interesting article Dan. Well, I guess my suggestion of Ferrite Cores is redundant then.

Stu....

Thanks, guys. Your answers are all along the lines I was expecting. I guess it never hurts to hope for a simple solution! I'll have the plant reroute shielded Ethgernet and see what the results are like. Can't hurt.

Thanks for you input.

Da, great stuff
RG you never said what brand of switch are you using? Is it a industrial grade unit?

Is fiber an option?

If you have to do a Re-Pull go with fiber. Shielded ethernet cable and rerouting may work and it may not. There is no doubt with fiber.

You may want to see if the drives have VFD cable or not. Shielded and Screened is best. See if the drives have EMc plates and or cores. If not add them. Cheap insurance.

The radiated nosise from the drive output cables could be the likely culprit.

The Plc Kid said:

If you have to do a Re-Pull go with fiber. Shielded ethernet cable and rerouting may work and it may not. There is no doubt with fiber.

You may want to see if the drives have VFD cable or not. Shielded and Screened is best. See if the drives have EMc plates and or cores. If not add them. Cheap insurance.

The radiated nosise from the drive output cables could be the likely culprit.

Click to expand...

I'm sure the drive output is the culprit. The drive is the best part of 60 yards from the CIP supply pump, in the same cabinet as the network switch and HMI. Adding a load reactor helped a lot.

I think I'll try the fiber. Like you said, I have to pull something...

+1 for pulling fiber

There is a good whitepaper on AB.COM/networks/media/ethernet web site, regarding the selection of physical media for ethernet.

I recently had to do some research for a large installation, and at first I thought: "What's the big deal? CAT-6 and away we go.". Not so.
There's a bit to it all, and the document is worth reading.