Ethernet First Time User

[Noely 0]

Hi, I am thinking of trying out ethernet on a small application for the first time. Basically all I have is a SLC 5/05 processor and I have a PC with RS View Scada to connect to it plus I want to install a Panel View 1000 in another location. I think it would be a nice little project to start on but I just need a bit of advice on what I need and is there any good downloads etc on how to go about addressing etc. Any help would be great Noely

[rpraveenkum 5]

First of set the IP address to the SLC 5/05 in Channel configuration Channel 1 through DHCP/Bootp server or through RS232 port Assign IP address to PV 1000 in configuration menu Set the sub net mask same to all three devices like 255.255.255.0 Set IP 192.16.19.X to pV1000 192.16.19.XX to SLC5/03 192.16.19.XXX to PC first three members (192.16.19) should be same for all device & last mamber should be different to all devices (x) want More details about ethernet http://www.ab.com/networks/site-index.html

[Ken Moore 23]

I would recommend you use industrial switches, not the commercial ones available at you local electronics shop. You application is small and depending on the distances involved you may only need one switch. Each segment will be limited to 100 meters. If your distance is a longer you will need additional switches. Since you will not have ethernet I/O, in my opinion unmanaged switches will be sufficient. If your distance is great, then fiber will be needed. Use good quality ethernet cable, I recommend Cat5e shielded plenum grade. For my control LAN, I use the same addresses scheme as the business lan, and just replace the first octet with a 10. So if you business lan uses something like 163.78.145.x, I would use 10.78.145.x for my control lan. This is just my personal scheme, you can use all 10's if you like 10.10.10.x

[BobLfoot 301]

I would recommend you visit the Rockwell Knowledgebase and read several articles. Like those lsited below. 1. Answer ID 5280 - Setting a SLC 5/05 for Ethernet communication. 2. Answer ID 34842 - RSLinx Max Block Sizes to SLC 5/05 on Ethernet. 3. Answer ID 17573 - SLC 5/05 Ethernet passthru to DH485 or DF1 PV on Channel 0. 4. Answer ID 19851 - Ethernet to DF1 Passthru Example. 5. Answer ID 21708 - EtherNet PanelView to a SLC 5/04 using a 1761-NET-ENI. 6. Answer ID 6653 - Ethernet PLC 5's or SLC 5/05's shows as an unrecognized device in RSWho, although they respond to ping requests. Also the event log for the Ethernet driver in RSLinx shows "unable to bind to socket" And the "billion" other Ethernet related answers they ahve for your reading pleasure. Others have made thier suggestions about number shceme's I'll be a little vaguer. Pick one that has room to grow. Don't fret if you skip a few addresses initially. Keep an IP Address Chart !!! Excel Works great for this. List the MAC ID and Desired IP and a Plain English device desrciption. Use the Chart LUUUKE!!! LOL But seriously you'll be glad to have that chart someday.

[rpraveenkum 5]

see this post for Industrial swithes http://forums.mrplc.com/index.php?showtopic=9503 Edited 14 Dec 2006 by rpraveenkum

[Noely 0]

Thanx a million for all this info. Just wondering bout cables. I see on my panel view that i need a cross over cable to program directly to it. I think i dont need a cross over cable if i use a switch or is this right and what exactly is a cross over cable. Thanx again

[rpraveenkum 5]

see tech note http://rockwellautomation.custhelp.com/cgi...amp;p_topview=1 Ethernet CAT 5 Cabling Pinouts Computers / Laptops should be connected to hubs with a straight-through cable. When making connections to the HUB, either the Computer / Laptop, or the HUB must be powered down or erratic network operation may result. Ethernet Twisted Pair Wiring Information (RJ45) Wire #1 White w/Orange Stripe Wire #2 Orange Wire #3 White w/Green Stripe Wire #4 Blue Wire #5 White w/Blue Stripe Wire #6 Green Wire #7 White w/Brown Stripe Wire #8 Brown STRAIGHT-THROUGH CABLE WIRING Wire #1 --> Wire #1 Wire #2 --> Wire #2 Wire #3 --> Wire #3 Wire #6 --> Wire #6 CROSSED CABLE WIRING Wire #1 --> Wire #3 Wire #2 --> Wire #6 Wire #3 --> Wire #1 Wire #6 --> Wire #2 Edited 15 Dec 2006 by rpraveenkum

[paulengr 44]

Except for 1 Gbps Ethernet, "Fast" (100 Mbps) and "slow" (10 Mbps) Ethernet use the exact same cable in the same way. There are 4 twisted pairs in the cable. Two pairs are the "transmit" and two are the "receive" pairs. The voltage DIFFERENCE across the pairs determines the signal. The signal formats are very complicated, especially for "Fast" Ethernet so you will need to do a lot of reading if you want to understand it (no real need). The other two pairs are not used normally. In a standard or "straight through" Ethernet cable, the transmit and receive lines are attached to the same pins. If you connected two devices together (say a PC and a Panelview), then the transmitters would be connected together and the receivers would be connected together...not what you want. So a "crossover" cable flips the connections so that they are wired to each other. A switch or hub has all of it's ports "reversed" compared to the devices so that the standard ("straight through") cabling works. When you terminate your connections, use TIA 368B on both ends for a "standard" cable. That is the standard for everyone except some phone companies that felt that they needed to be nonstandard. Those people use TIA 368A, which reverses the transmit/receive pairs. SOOO...if you need a crossover cable, terminate it with TIA 368A on one end and TIA 368B on the other end. If you are buying premade cables obviously it doesn't matter whether it uses 368A or 368B or not since both types will interchange (it just flips which pair of wires is normally transmit vs. receive).

[Peter Nachtwey 15]

See this 192.16.19.X is not a non routable IP address. It will conflict with someone else. http://www.mtmnet.com/PDF_FILES/NonRoutableIPaddresses.PDF We recommend 192.168.WWW.NNN. See the document. Where WWW is the work center number and NNN is the node or device within the work center. The IP mask should be 255.255.255.0. This keeps the devices or nodes in work center 1 from interfering with the devices in work center 2. The first device 192.168.WWW.1 is usually the gateway for the router. This allows traffic to go outside of the work center but only when necesarry. Every work center has its own router. This keeps the traffic separated and is VERY important when using Ethernet/IP.

[BobLfoot 301]

Peter I also notice that 10.BBB.CCC.DDD and 172.16.CCC.DDD thru 172.31.CCC.DDD are non-routable. What is the decision process to choose between 10... 172... and 192.... Any rules of just preference?

[Nathan 7]

Ken's advice is good. I wanted to expand on a few points: On the cable type: Go with shielded cable, Cat5e should be sufficient for the foreseeable future, but you might run Cat6 or 7 if the price is similar enough - depends if you ever plan on running gigabit. "plenum" cable is rated for airspace (plenum) in commercial buildings so that its flammability and smoke producing characteristics are acceptable. Your cable should be enclosed in conduit, so this isn't necessary (but something that I would recommend if the price is similar enough - there aren't any downsides). As far as using 10.10.10.x, this advice is better than has been justified. 9.9.9.x, for example, would be a poor choice. Ken has chosen a non-routable address range, meaning that internet routers will ignore these addresses as "invalid", providing another security factor for your application. This why home routers always default to 192.168.x.x/24. The unconventional thing is that he seems to have chosen to only use 255 addresses (depends on the mask). The textbook example would be to use 192.168.x.x in this case, but you can divide up network addresses however you want (subnetting). edit - I didn't read down to BobLFoot's response - Read about RFC 1918 non-routable ranges have been defined as "illegal" and are thus ignored by internet routers. 1. You won't run the risk of having a conflict with a "real" address on the internet 2. It will be significantly more difficult for an attacker across the internet to communicate with your computer directly - it won't be possible via a routed connection to that address. 3. Most importantly now days because of all the NAT devices - it's simply a matter of convention that indicates that you have some idea what you're doing with networking. Traditionally, using these addresses meant that you didn't want to have communication with the Internet. However, with the shortage of real addresses in IPv4, NAT, enabled devices have become commonplace. In a nutshell, all traffic is directed toward the router, which keeps track of which node on the network it should go to. Technically speaking, you could set up a NAT enabled subnet with addresses in other ranges with a router that's connected to the Internet via a real address on the WAN side, that same router could have an address on the LAN side that should be valid, but doesn't belong to it, and everything would work fine. With NAT it's more a matter of convention. All that said, if you expand to work with IT and they see that you're using 9.9.9.x, they probably won't give you the time of day. I posted a bit more on non-routable addresses here on the IA forum Edited 17 Dec 2006 by Nathan

[Gerry 13]

There are a lot of conflicting opinions on this subject. Even within Rockwell there is disagreement. However, one thing I recently read was that unshielded UTP is susceptible to interference from electical fields whereas shielded cable (STP?) is susceptible to interference from magnetic fields. The item also noted that magnetic fields were more prevalent than electrical fields in common industrial environments and that ethernet switches have filters that reject any signals less than 100 Hz. Elsewhere I've read that you should only use shielded cable if you're running it in metal conduit.

[Noely 0]

Thanks lads, Hope ye think i'm not been smart here or ungreatful by saying keep on talkin there between yerselfs as i'm learning more and more as ye go on. All that info has been of great help. Thanx again and happy xmas

[paulengr 44]

This is a rather long rambling post about several observations Ethernet-wise but they are based on actual tests that I've done recently and investigations I've done with my "RF engineer" hat on. The following is EMI 101 and how it applies to Ethernet. Basics of EMI: The impedance of "space" is just a few hundred ohms. It is relatively easy to "leak" signals into space for that reason, especially as the frequencies increase. At microwave and higher frequencies, the challenge is frequently just trying to design the circuit to keep the signals "inside". A simple 90 degree bend can act as a transmitter/receiver! Anything feature of a circuit more than 1/10th of the wavelength is generally considered electrically "short"...electromagnetics doesn't matter. But 1 Gbps Ethernet signals are running at 100 MHz, or about 2 meters in wavelength. This means any electrical feature bigger than 20 cm (about 8") begins to take on significance in the world of electromagnetics. In other words, anything other than a very short patch cable. This doesn't mean you need an RF engineer to design Ethernet...just that those issues are out there. There are generally two types of antennas, aka "radiators". One type is called a "loop" antenna. This is just that...a loop of wire. The other basic antenna type is called a "dipole". You don't need both legs...a single leg will still work. This is simply an open wire that is not connected to anything. For now let's assume that we don't have tuned antennas...we are going to assume that the antennas are several wavelengths long. Also, we are working in the near field. In the "far field", the electromagnetic wave tends to even out so that it is 50% E-field and 50% H-field ("magnetic"). Although either antenna can exhibit either behavior (depending on length), generally "loop" style antennas are most efficient for coupling H-fields. Open wires are best for coupling E-fields. Antennas "couple" the signal...they can both radiate or transmit the signal, as well as receive the signal. In most cases, this transmit/receive dichotomy is a two-way street (they are equally effective in either direction). The only exception I can think of is that some lenses and certain circuit elements (circulators) are one-way devices. Now think about a good source of H-fields. Every 60 Hz AC circuit in your plant is a loop and it is actually rare to find "open" wires anywhere for obvious reasons. So most plants abound with H-fields. If you've ever worked with an oscilloscope and an ungrounded circuit, you know what I mean. An example of an "open" conductor would be a welder, which would make an E-field. Incidentally an electrical arc is not a half bad transmitter if all you want is bulk RF signals (remember the original radio transmitter was a spark gap). Now, let's switch over to looking at the EMI problems that we should expect from various control protocols. I spent a lot of time studying this stuff since I was keenly interested in moving to an "everything Ethernet" model for performance and longevity reasons. I was rather unimpressed with the zillions of failure modes that plague Devicenet (CAN in non-AB lingo) and Fieldbus, which pretty much left either Arcnet ("Controlnet" in AB lingo) or Ethernet as options. The major consideration for my polant was how robust Ethernet is relative to DH+/RIO, The first problem was just trying to find electrical information about AB DH+/RIO. This is not easy but it appears that these are MOSTLY a variation on RS-485, the same physical layer in CAN and Modbus (among others). There is something else funky with the signal formats but I was unable to determine what the actual signal format is. Either way, conceptually DH+/RIO are subject to the same problems. Since it is relatively low frequency, DC interference and induced signals (such as 60-Hz coupled signals in improperly grounded shielding) are death. Other than that, it has a small voltage advantage (+/-12 volts differential signal vs. +/-1.5 volts) going for it, and the fact that it never exceeds anything over 1 MHz including any harmonics at the blistering speed of 230 kbps on high speed RIO. These types of interference problems simply don't exist on Ethernet since it has some pretty strong filtering to clean out any and all low frequency signals. Whenever you read about Ethernet, you will see a reference to the "magnetics" which is a couple ferrite beads or inductors and a 1:1 isolation transformer. This means among other things that until you get into a several 10's of decibels of interference rejection against any D.C. or even low frequency (60 Hz) signals. I haven't tried directly connecting 60 Hz power over an Ethernet connection but the interference rejection is so good that if I had the equipment to sacrifice, I think it might survive it. Some of the tests that AB published when they began investigating Ethernet suggest just how much interference would be necessary. For instance they ran an Ethernet line directly next to a long run of power leads for a large motor and they ran an Ethernet cable all over a welding robot to try to get so much as a bit error without any success. I forgot where I found it but I did find electrical test results of ScTP and UTP cable. The interference rejection was 1 dB better for ScTP. There was an excess link budget of about 30+ dB as I recall, meaning that the interference would have to be 1000 times stronger to cause problems over Ethernet. 1 dB increases it to about 1250 times...a small but modest improvement. Getting back to EMI...twisted pair wires have 3 key electrical characteristics that designers have to be aware of if they are designing twisted pair communication lines (you don't have to...CAT 5E is already "designed"). First, the wire has some inherent inductance related to the length and you can't really do much about that except to keep the wire as short as possible (Ethernet has a maximum length specification). Second, there are "eddie currents" that create self-inductance. The metal that makes up the wire can help that somewhat. Finally there is capacitance which is the easiest one to improve. Two wires in parallel over a long enough distance are effectively the two plates of a capacitor. The only ways to reduce the size of that capacitor are to make the wires as close together as possible, and to make them as small as reasonably possible. The same specs. that push twisted pair designers to make CAT 5E cables very small also reduce the size of the "loop" that the twisted pair forms. In addition, twisting the wires means that it minimizes the exposure to EMI at any given point (rolling the wire around won't enhance/decrease EMI). Overall, CAT 5E is very noise immune almost by design, especially to H-fields. The screen (technically, ScTP is "Screened twisted pair", not "shielded twisted pair) creates a Faraday cage. This means that any outside E-field will induce opposite charges on the inside surface of the shield, but the electrical field is then cancelled and the inner wires are unaffected by E-fields. The surface area of the twisted pairs themselves again is very small and the fact that the wire is twisted means that it has a lot of "self" shielding...E-fields will tend to create alternating charges on the wire which will self-interfere and cancel out. The shield does have the ability to create a "loop" antenna which can easily pick up H-fields (magnetic fields) and concentrate them around the shielding. The "loop" is around the diameter of the shield. The worst thing you can do though is to ground it at both ends...then you have truly created a real loop across the entire shield and the two grounds. This is the reason for the "grounding at one end only" rule that installers so frequently fail to do correctly. Now consider RMC...this is effectively the same thing as a screen...or in this case, an actual shield. Again if you followed the rules correctly with bonding, it should only be grounded at one end. If not, you get ground loops...or antenna loops in this case. Still, UTP is pretty rugged against this sort of thing as I stated earlier. So the effect of a shield (RMC or bought with the cable) is to create a Faraday cage. It should almost totally nullify E-fields, but it can create additional H-field concerns. Hence my statement that shielding flips the degree of protection between E-fields and H-fields. For most plants, I'd lean towards giving them every last ounce of H-field shielding I can get. If it is a welding cell for an automotive plant or a large AM radio transmitter, then I'd be giving serious consideration to E-fields and ScTP would be the way to go. In general, Europe seems to prefer ScTP and the U.S. trend seems towards UTP. I'm not sure why. If only there was an "AC" Ethernet cable...then I'd have the shield and the conduit too in one package. Electrically, consider the specs of CAT 3, CAT 5, CAT 5E, and CAT 6. The difference between CAT5, CAT5E, and CAT6 is small. CAT5E is simply CAT5 with tighter specifications (those inductances and capacitances). It is interchangeable with CAT5 but I can't hardly find CAT5 anymore (compared to CAT 5E). CAT 3 handles up to 10 MHz bandwidth. CAT5E handles up to 100 MHz bandwidth. CAT6 handles up to 250 MHz bandwidth. CAT 7 is already in development/released and will go significantly higher (I've heard 600 MHz but I really have no idea...the goal is probably to support at least 100 Gbps and maybe 1 Tbps). All of this stuff was originally meant for telephone wire (CAT 3 is old PBX telephone wire). The phone company was interested in tight specifications because they are trying to push relatively low frequency signals very long distances (miles), which will cause skew, low pass, echoes, and other nasty effects on voice calls if the wire impedance is not under control. But it was readily available so it was repurposed to communication (original Ethernet specs. called for coaxial cable). In terms of costs, get rid of all that CAT 3. Cut it up and throw it away. I don't even try repairing it in the few pieces left in the plant (it's all at least 10 years old anyways). The cost difference between CAT 3 and CAT 5 is insignificant. Even worse...I've had switches (mostly CISCO) REJECT CAT 3 links, even if you use only a 10 Mbps device. Although this sounds like CAT 5 is limited to just 100 Mbps (since CAT3 will only handle 10 Mbps), there is much more to the story. That would be equivalent to claiming that analog phone lines are only capable of handling 3 kbps (since phone lines are designed for just 300-3000 Hz signals). That would preclude 28.8 kbps, 56 kbps, and certainly xDSL which is capable of anywhere from a few hundred Kbps up to several Mbps. In reality, 10 Mbps Ethernet uses Manchester coding which is just about the lowest level of "trellis shaping". Each bit is actually either a positive or negative voltage for half of the bit followed by a zero for the other half. This causes the signal to naturally have a null (no signal) close to "0 Hz". This is why 10 Mbps Ethernet can pass right through the ferrite beads and the isolation transformer that is used in the transceivers without distortion. 100 Mbps ("Fast") Ethernet uses 5 different voltage levels and restricts which combinations of voltage levels are "legal". This is also known as trellis shaping. It causes the signal to have a null close to 0 Hz again as well as upper limiting the signal so that it is all contained in <30-40 MHz, certainly well within the 100 MHz bandwidth of CAT5E. 1 Gbps (gigabit) Ethernet goes even further. It fully utilizes the other pair of unused conductors (only 2 of the pairs in a CAT5E cable are used for 10/100 Mbps Ethernet, one in each direction). In addition, it adds even more voltage levels and more shaping. The end result is that 1 Gbps Ethernet fits comfortably within the 100 MHz bandwidth of a CAT 5E cable without any trouble at all. The extra bandwidth of CAT 6 is useless. . 10 Gbps Ethernet is not really very standardized yet since I haven't seen any "head-to-head" interbrand comparisons of interoperability among the few vendors that offer 10 Gbps Ethernet over copper. At this point you really can't fit on the old faithful CAT5E anymore, so CAT 6 (or fiber) becomes a necessity. I suspect that CAT 6 will probably be an interim and the next jump will be to CAT 7. In the mean time, I'd be asking myself whether I really want 10 Gbps+ copper given how "tight" the signal is or whether it would be better to just go straight to fiber...where I can get 10 Gbps or even 100 Gbps (or more...) today....and ultimately if the network is ever going to grow to that point. On fiber none of this matters because with DWDM you can always take the same old single mode fiber that you originally ran and get an almost arbitrary amount of additional bandwidth out of it. There's another problem Ethernet-wise that sets in on the 1 Gbps+ link speeds: latency. Although the sustained speeds are as claimed, there are several timing delays built into the Ethernet standard that are necessary. 100 Mbps Ethernet is just about "optimal" in this respect. You can't reach true 1 Gbps speeds because of these delays on Gigabit Ethernet and the problem gets worse as you increase the speed even higher. The old "deterministic vs. statistical" argument finally swings over to favor deterministic networks...hence FDDI token passing and IEEE 1794 don't suffer these bandwidth limitations. If there was such a thing as 10 Gbps Arcnet, it might actually make a comeback as the network of choice. So in the end so far, it seems like unless there's no cost difference between CAT 6 and CAT 5E (or CAT 7), I don't see any electrical advantage to running it. I've heard the same "buy CAT 6 because it does Gigabit Ethernet" claims before. It is uninformed (1 Gbps works just fine). There is a slight physical advantage in that in order to get the higher bandwidth, CAT 6 conductors have to be tighter packed in the cable, making it truly round (no "bumpiness" to it), which makes it easier to pull...but then if you are stressing your cables that much, you will probably manage to stretch or kink a cable and have to redo it anyways.

[Ken Moore 23]

Nice post, very informative.

[robh 12]

This was very helpfull. Thanks

[Gerry 13]

Long, maybe - but most informative. I appreciate getting the backgound to the "rules-of-thumb" that I have read in various places and mentioned in my previous post.

[rpraveenkum 5]

some interactive information about Ethernet http://www.ab.com/networks/interactive.html