Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Hardware

Grounding Ethernet Cable on a Ship? 58

Posted by Cliff from the dealing-with-a-literal-floating-ground dept.
wrschaeffer asks: "On a ship (a vessel at sea) where a floating ground exists, we are running shielded Ethernet cable to connect between 5 and 10 PC in our LAN. We are unsure on the grounding requirement. ABS (American Bureau of Shipping) has no requirements concerning grounding of Ethernet cable (just smoke and waterproof standards). We have many possibilities: ground one end; ground every end; ground every end and install a isolation transformer at the punch-down block; or don't ground any end. What should we do?"
This discussion has been archived. No new comments can be posted.

Grounding Ethernet Cable on a Ship? More

Grounding Ethernet Cable on a Ship?

Comments Filter:

  • hmmm (Score:3, Funny)

    by the_other_one ( 178565 ) on Saturday November 23, 2002 @11:16PM (#4741173) Homepage

    I thought the general idea was to not run aground on a ship.

  • Ground one end (Score:5, Informative)

    by jcwren ( 166164 ) on Saturday November 23, 2002 @11:20PM (#4741191) Homepage

    I would only ground one end, and the end at the hub, since that would be the most common point of the network.

    But this raises the question: Why are you using shielded ethernet? Ethernet is differential, and thus pretty immune to common mode noise.

    I've had 25m ethernet cables draped across running engines (test equipment) with HEI ignition, and not experienced any problems (I can't say that I checked the error counters very often, but in high speed data collections, I wasn't seeing any unexpected decreased throughput).

    In addition, I live on a houstboat, and I have ethernet run in the same conduits as AC power carring 5 to 15 amps on 4 different circuits, and don't experience any problems (max cable length is probably 20 meteres)

    • if you know so much about ethernet, why are you living on a houseboat?!

    • Re:Ground one end (Score:3, Informative)

      by Zarquon ( 1778 )
      But this raises the question: Why are you using shielded ethernet? Ethernet is differential, and thus pretty immune to common mode noise.

      Probably to reduce emissions.. Most boats have a lot of sensitive electronics.. and when you're under EMCON, you don't want anything avoidable to emit. Also, consider you'd want to make at least a minimal attempt to harden the equipment against any kind of EMP.

      Had a friend that worked at a naval shipyard.. spent weeks trying to locate, identify, and eliminate undesired emissions from various (uncertified and certified) electronics.
      • Cat 5 cable - especially shieled cable - isn't likely to radiate. The internal signals are differential, which is a "balanced" transmission line, and isn't going to radiate to much because of that fact.

        The more likely concern is picking up other frequencies on the cable itself on the shield!

        So - the guy is likely really asking about how to ground the shield - not the transmission lines themselves. At least I hope so - grounding the transmission system itself isn't to good an idea.
        • Doesn't take much.. Pinching the cable or bending will commonly shift it and undo the twisting in that area.. I suspect the main reason, though, is to avoid induced currents, whether from an emp or leakage from the radar / transmitters.
    • You probaly don't live on a big metal ship either.

      On an ocean going ship, shielded cable & grounding is a must. Static discharge from bulkheads is more than capable of corrupting data or even zapping a NIC. Other sources of interference include active radars or even shortwave transmitters in certain circumstances.

      Depending on the type of vessel, there may be other things going on to interfere with things. Fishing vessels has powerful motors and machinery. Naval vessels have fire control radar and ECM. Cruise liners have everything from big speakers in the disco to large electric ovens and water heaters.

      • These all seem like good resons not to ground the cable. If there are static discharge from bulk heads then just think of that discharge going into a NIC. Because I knew a guy that keep having NICs fail in a lab at the other end of the building. Turns out the thinnet was grounded in a couple of places and was up to 90v, and blowing up the transceivers. Run the cables in conduit and let that be the shield for long runs.

    • Re:Ground one end (Score:3, Informative)

      by Halvard ( 102061 )

      One of the reasons for the 100 meter spec for distance of Ethernet runs is potential voltage. If you are on a ship with long runs of Ethernet, you should definately ground the shields using good straps to the hull or to studs that aren't covered in paint. I would ground both ends and perhaps along the middle. I'd also use something akin an APC ProtectNet [apcc.com] or Tripp Lite DNET1 [tripplite.com]. The longer the run, the higher the potential voltage on the line and ground these to the hull as well.

      The shield, while it could carry a charge, is mostly about radiation -- into or out of the shielded cable. If you are running your Ethernet along power like the fool on the houseboat (inducting 60Hz into his Ethernet) you are asking for collisions, power spikes, etc. The potential charge on Ethernet conductors itself over longer runs probably will spell trouble all by itself without assistance from 60 cycle.

    • I would agree with grounding one end ... the general rule is to ground all signal lines to just one point so that you don't end up with any ground loops. This is the rule we followed here when we were running 10base-2 and grounded the cables at the coax hub.
      In addition, I live on a houstboat, and I have ethernet run in the same conduits as AC power carring 5 to 15 amps on 4 different circuits, and don't experience any problems (max cable length is probably 20 meteres)
      Well the electrical inspector would thow a fit over running low voltage in the same conduit as AC power. It's a big no no and should never be done for safety reasons (besides inducing AC current in the ethernet).

  • Big problems. (Score:4, Informative)

    by TheSHAD0W ( 258774 ) on Saturday November 23, 2002 @11:26PM (#4741219) Homepage
    I've heard of ground differentials as large as 90 volts on ships -- a serious problem, and one you wouldn't expect on a metal vessel surrounded by nice conductive salt water.

    I would recommend you run some fairly heavy -- 12 gauge should do -- ground wires in between the points you will be running the ethernet, and use the cables to interconnect the equipment ground points. Once you've done this, there shouldn't be any dangerous potentials on your ethernet lines. While you're at it, I'd also suggest you occasionally check for currents running on those ground wires; if you hook an ammeter around the wire and see more than a few dozen milliamps, you've got a bigger problem than ethernet grounding.
    • Excellent point. I had not thought about the potential that would exist between say the bow and stern of a ship. I imagine just moving through the water would generate charges.

    • Good advice. (Score:3, Informative)

      by Futurepower(R) ( 558542 )

      Good advice. Run heavy ground wires between the points. Then use shielded Cat-5 with the shield grounded at one end. You don't want the shield to carry current; it is only to prevent common mode pickup.

      "Common mode" is capacitive or inductive coupling to both wires of a pair at the same time. Common mode voltages can overload the Ethernet common mode voltage range. The common mode range is a few volts, so you also need the ground wires to insure that the range is not exceeded.
    • Good point, if you have that kind of potential between two parts of a ship, there may be even bigger problems than the Ethernet shielding.

      What happens if the actual computer and network equipment at each location is grounded there? You've just run 90 volts into an Ethernet port.

      Definitely make sure all equipment grounds are connected.

      Sounds like a nightmare.

      • What happens if the actual computer and network equipment at each location is grounded there? You've just run 90 volts into an Ethernet port.
        And if you don't ground the computer at the other end and it's referenced to the data line, someone touching the computer could get 90 volts through them. Whatever you do, you've got to be careful, and eliminating the possibility of 90-volt differentials (perhaps by running heavy copper braid along the route of the cable and grounding that instead) would appear likely to help (so would eliminating the currents which cause the 90-volt differentials). Just don't allow the possibility of ground currents running through your data wiring.
  • If it is a military vessel, then they would have to consider the degaussing equipment that is on most military ships. That's the only reason I can imagine using shielded cable. Unless of course it was running around high emf emitting equipment such as radar. If it is not a military vessel, grounding is not really necessary. However on a military vessel, the requirements imposed by the government may be different.

  • Ground at the switch. (Score:4, Informative)

    by Anonymous Coward on Saturday November 23, 2002 @11:27PM (#4741225)

    Allen-Bradley's Web site has a handy little pdf document [ab.com] describing a bunch of industrial Ethernet wiring practices, which would likely work well on a ship. Page 3-5 looks like the definitive spot for grounding shielded Ethernet cable -- connect the shield at the switch end, disconnect the shield at the remote end, and give the switch a good ground.

    Hope this helps.

    • This is the best piece of advice posted to this discussion. There have been some suggestions elsewhere in this discussion to ground the ethernet cables at multiple points. This seems to be the intuitive thing to do; however, it isn't---such suggestions are assuming that the entire hull of the ship is at the same ground potential (not a reasonable assumption). Use the Allen-Bradley instructions and you will stay out of trouble.
  • Interesting question. I'm unclear how normal ethernets of whichever cabling avoid creating groundloops in normal installations on the signal return line, or the old coaxial BNC-thin-net's shields. Maybe the NICs have isolation transformers.

    I presume you are talking of how to ground the shields of your shielded cable. To be effective as shields for EMC purposes, you will want the shields grounded to something. Normally, we'd connect any shield at one end to case-ground and leave it open at the other, to avoid ground loops. However, in the rest of the world we are required to bond our RF and Power grounds. Your power ground is floating. For EMC purposes, I'd think it would be the RF ground not the floating AC ground. In which case, you'd want it isolated from computer AC ground / case ground at both ends. How to ensure the AC ground stays floating from the RF ground must be covered in your shipboard electrical codes, good luck.

    I'd recommend discussing this with a qualified marine radio installer. They'd be aware of the grounding practices for digital cables in shipboard installations.

    Bill N1VUX (licensed for AMATEUR radio only)

    • "Interesting question. I'm unclear how normal ethernets of whichever cabling avoid creating groundloops in normal installations on the signal return line, or the old coaxial BNC-thin-net's shields. Maybe the NICs have isolation transformers."

      Each NIC transformer-couples Tx and Rx data. The transformers often have a non 1:1 ratio so they are usually impedance matching as well as isolation. There's usually a 1Kv-rated ceramic capacitor from the transformer center tap to chassis ground to bypass static charges.

  • Only for coax (IMHO) (Score:2, Interesting)

    by AnimalSnf ( 149118 )
    As far as I understand grounding is only a concern for coax. When using CAT 5 patch cable that isn't a problem because the signaling is differential, thus any noise that induces current in one wire also induces current in the other wire, the difference between the two still remaining zero. (Similar techonology is also used in USB)
  • If you ground both ends you'll set up ground loops - from what other posters have said, that sounds even worse on a ship. Ground neither ends, and you don't get much benefit from using STP. Always, when running any kind of cable, ground _one_ end.

  • don't ask slashdot this (Score:4, Interesting)

    by Aniquel ( 151133 ) on Saturday November 23, 2002 @11:44PM (#4741297)
    You shouldn't be asking us this. The high schoolers that lurk here are going to make jokes or give their 'best guesses', the rest are going to give you anecdotal evidence (ie, 'what I've heard'), and those that really know what's going on will be ignored. This is something for an electrician, someone that does this stuff for a living. That sure isn't me, and it sure isn't 99% of the slashdot population.

  • Bad grounding is a killer - literally. (Score:3, Informative)

    by stienman ( 51024 ) <adavis&ubasics,com> on Sunday November 24, 2002 @12:28AM (#4741442) Homepage Journal
    Bad grounding kills. I'd rather work with an ungrounded system than a badly grounded system.

    First of all, ethernet doesn't need a ground.

    If you're already dedicated to it, however, you do need to look into it, because an ungrounded shield can be dangerous as well.

    For ideas, do a web search on grounding rs-485 network cables. rs-485 is a differential signalling system similar to ethernet that is meant for long serial runs (kilometer or more) where grounding is a an issue.

    In most cases, you need to find out what the voltage differential is, possible current flow, and run a suitable grounding wire along with the cable since the shield may not be adequate to the job, and may still leave quite a voltage differential. Sometimes, however, it's appropiate to allow a voltage differential and ground through a capacitor and/or resistor.

    In other words, there's no single way to do it, and the way you do it will depend on many things, not the least of which is what you're doing with the network and where it's being used.

    Good luck! Don't suck on the wires, especially when you know for sure that they aren't live (they are always 'live').

    -Adam
  • We have many possibilities: ground one end; ground every end; ground every end and install a isolation transformer at the punch-down block; or don't ground any end. What should we do?"

    Ground one end: YES

    Ground every end: no, unless you want noise

    Ground every end and install isolation transformer (on the ethernet TX/RX ckt I assume): No, unless you want an expensive solution to produce noise.

    Don't ground any end: this is the safest if you don't know what you're doing.

    IANAEE, but common sense tells me that you should be running your computer equipment on isolated ground circuits. If you're also using shielded ethernet, either ground the shield at the nics or at the hub, and be consistent. If you're using isolation transformers ON the ethernet TX/RX circuit itself, grounding should definitely be at the hub, bonded to the isolation transformer's neutral or ground.

  • Go wireless.

  • use fiber optic (Score:4, Informative)

    by wotevah ( 620758 ) on Sunday November 24, 2002 @01:57AM (#4741744) Journal
    Grounding both ends of the cable would definitely be a serious problem. The lower the wire gauge is, the worse the effect is going to be (you are creating a low-resistance connection between two points of different potential, this will generate significant current draw through the wire).

    I would suggest to go fiber, it's immune to this kind of problems and I have seen fiber optic cards for decent prices in stores. Or use fiber for long distance connections and regular wired ethernet for LANs.

    If you must do copper, ask someone who has done a copper-based university or large building WAN. They will likely have had to solve the same problems, as the same grounding issues can appear between equipment powered from AC sourced different phase lines. I don't recall exactly but grounding at only one end and leaving the other in the air might have been the solution.

  • DoD grounding recommendations (Score:5, Informative)

    by Anonymous Coward on Sunday November 24, 2002 @02:45AM (#4741876)
    The applicable US DoD standard is MIL-STD-1310G - "Shipboard Bonding, Grounding, and Other Techniques for Electromagnetic Compatibility and Safety". I'm not sure if that document specifically addresses Ethernet, but one of the illustrations should describe how to ground some differential twisted-pair data network. You can get MIL-STD documents from DAPS [dla.mil].

  • Grounding, ethernet, etc. (Score:5, Interesting)

    by AlphaOne ( 209575 ) on Sunday November 24, 2002 @03:12AM (#4741964)
    First things first... lots of people are saying that ethernet doesn't require a ground. That's WRONG... a ground is always required, especially on transmission lines like ethernet cables. However, CAT5 is a balanced transmission line and shielding isn't all that important unless you're running the cable through an environment likely to disrupt this balance.

    Second, the type of boat you're on makes a big difference. If you're on a metal hulled boat (or a boat with a metal mast or other convenient absolute ground), you're ground doesn't float... the water is ground. Salt water is especially good as a ground reference.

    If you're in a fiberglass boat, the whole vessel is floating above ground. See if you can find a metal object (preferably brass or other non-oxidizing metal) always submerged and use it as your ship's ground.

    Either way, the easy answer is don't worry about grounding the shield.

    If you run into problems with data corruption (or want to prevent such problems), you should ground BOTH sides of the shield. On long cable runs (longer than a half-wave at your operating frequency) you should ground only ONE side. The hub-side is the best to ground as it is already the central network point... so making it the central grounding point makes sense as well.

    The BEST solution is to use something called telescoping grounds. In a telescoping ground, both ends are grounded, but to different shields. Both shields are independent and isolated from one another and overlap for the length of the cable. This prevents common-mode currents on both components and avoids ground-loops at the same time. Professional audio cables use this technique often.

    Another option you may not have considered is wireless. With a wireless network, the only ground that's important is that of the base station (for efficiency's sake). The client devices have their own ground reference and its relationship to the base station isn't important.

    At any rate, good luck!

  • go optical and/or don't do ethernet (Score:2, Informative)

    by Anonymous Coward
    ethernet is not really intended for industrial situations. the signal levels are too low, too susceptible to induced noise; the connectors (RJ-45) are not seaworthy; on a ship you ALWAYS use stranded pretinned wire, which is not what cat5 cable is (solid, untinned). the salt and vibration will kill the cat5 cable and connectors.

    first: if you can, put in optical fiber. that way you will avoid the ground loop problems entirely.

    indeed, first time most folks saw optical fiber was in the raised floor computer room, separating the mainframe chassis and peripherals.

    second: electric grounding on boats gets really complicated really quickly. you have several conflicting requirements for "grounding": HF radio counterpoise, electrical noise reduction, shore-side AC ground, perhaps onboard generator AC ground, onboard DC ground, lightning strike diversion, metal conductor bonding (hand rails and engines etc.) ... and electrolysis - don't let stray currents in the water eat your propellors!!!

    go to a marine book store like red tide or armchair sailor or west marine and get some books on marine electrics. or hire a real marine electrician, it's money well spent.

  • Don't ground aboard a ship. (Score:2, Informative)

    by Anonymous Coward
    Grounding? Not until you know everything there's to know about galvanic corrossion.

    YOU COULD SINK THE SHIP IN HOURS.

    Sorry to shout, but current flowing through the hull is about as bad as sailing over reefs or bumping into mines. Check with experts before connecting anything.

  • Noise vs. Electrical (Score:4, Informative)

    by ONOIML8 ( 23262 ) on Sunday November 24, 2002 @10:46AM (#4742980) Homepage
    You've had a lot of input here, some of which seems good to me. But I think you're looking for input on something that isn't YOUR problem.

    Grounding the network cables would be your problem if you were fighting noise. Then grounding one end would be the place to start.

    But if you're worried about a difference in ground potential causing you problems, anything to do with network cable is NOT the solution. That is something that had better be dealt with by a marine electrician. Not any electrician, one who knows ships and knows them well.

    Here's why: Network cable is small signal, low current stuff. If you have a difference in ground potentials you could be talking about high voltage and/or current. You don't want this on your network, it will fry everything. Have those problems solved first by an electrician and know that the enviroment you'll be working in is friendly for your network.....AND FOR YOU!

    As an AC pointed out, you can do harm to the ship by changing the electrical properties of it. When those changes start to happen it will likely also do damage to your network.

    If it were me I would:

    1) Consult the marine electrician most familiar with the ship. Once you have his ok for safety and blessing with your project then...

    2) Check with the radio operator/technician if you have one. He's got equipment that works much like your network. There are things he may know which will save you heartache and speed up your progress.

    3) Install the network, tie all shields to the hub. Then "ground" the hub to whatever the electrician tells you to. Remember that you're only doing this to prevent noise on the network, it has NOTHING to do with ships ground...and you don't want it to.

    4) Ask the electrician and radio op to check your work when you are done. Log this. Then if there is ever a problem you can show the captain the log and he'll know you did your homework. Cover your ass always.

    5) Have a nice day.

  • Oh... (Score:2)

    by joto ( 134244 )
    I would suspect most large ships built of metal which is of course permanently grounded to the saltwater in the sea would be as reasonable to work with as any other building. But since you ask this question, you probably worry about something.

    So, why would you run grounded ethernet? What's wrong with a normal Cat-5? What kind of ship is it? How far is it between the pc's? Have you considered fiber? Unless you can say a bit more about why you worry, it's impossible for anyone to give a qualified answer.

  • On a ship (a vessel at sea) where a floating ground exists...

    haha... Very funny... :)
  • Ok, I've got some experience here- I've been testing and designing Ethernet for one of the major computer manufacturers for a couple years now. Most manufacturers of Ethernet use shielded connectors (jacks). This shield is what connects to your shield on shielded RJ45 cable. On every shielded ethernet port I've seen, the shield connects to chassis ground- which is the same ground as the ground pin on the power cord. The differential pairs themselves (according to the IEEE spec) should be electrically isolated with a transformer to take up to 2250 VDC or 1500 VAC.

    In a situation where you're not sure if you're running the two ends off of a common ground, I definitely would not recommend using shielded cable. You could end up sending that cable up in smoke if your grounds shift.

Slashdot Top Deals

No man is an island if he's on at least one mailing list.

Close