johnson@EA.USL.EDU (06/17/91)
I have 2 buildings about 140 feet from each other which I need to run ethernet between. I have decided to run direct bury thinwire and am aware that there could be a possible ground loop problem between the two buildings. I have not checked this, but know it can be. Is this something about which I should be concerned??? Also, why should a coax only be grounded on 1 side of a segment? Why not put a small resistance at the other end to drain that small amount of current which might be there? Any suggestions would be appreciated. Lee J.
hedrick@athos.rutgers.edu (Charles Hedrick) (06/18/91)
Please do not connect Ethernet directly between buildings. Ground loops are only one of your dangers. A lightning strike can induce enough current in the wire to kill machines on both ends. (We've had this happen.) We strongly recommend using fiber for all runs between buildings. Fiber transceivers are now fairly inexpensive, though depending upon your setup you may also need one or more repeaters.
dan@gacvx2.gac.edu (06/18/91)
In article <0094A40A.0DF70DA0@EA.USL.EDU>, johnson@EA.USL.EDU writes: > I have 2 buildings about 140 feet from each other which I need to run > ethernet between. I have decided to run direct bury thinwire and am > aware that there could be a possible ground loop problem between the > two buildings. I have not checked this, but know it can be. Is this > something about which I should be concerned??? Also, why should > a coax only be grounded on 1 side of a segment? Why not put a small > resistance at the other end to drain that small amount of current which > might be there? > > Any suggestions would be appreciated. > > Lee J. Under normal circumstances the difference in the grounds between buildings is small. However since the cable will also be connected during abnormal conditions, you must plan for the reasonable extream. When an event like a thunderstorm passed over your campus, the ground potential changes rapidly. These changes are part of what makes lightning work. Between two points within just a few hundred feet of each other there can be hundreds to thousands volts difference during a storm. The grounding within a building keeps it at same level through out a building. The difference during a storm has the potential to do damage to electronic ciruits. This is damage due to a direct hit, just from a potential change caused by the charges in the ground and the clouds. If you tie into the grounds on both ends of your cable, then it must be able to take the current it could possibily be subjected to, because you would be tying the grounds in the two buildings together. Optical isolation is the way to go. You could do it with a IR tranmiter and receiver pair, but it is better to go for fiber. Fiber read for ground burial is about $2.00 a foot. It should be about $150 an end for termination with connectors. Ground burial coax cable is expensive too, since it usually has an armor coating, and sometimes a second sheild. When I used to run RS-232 all over the place using various concentrators and line drivers. I found that sometimes a ground difference of only a few volts would make the difference between a terminal working and not. The ground float would sometimes cause 0s to be read a 1s and 1s also would be read as 1s or vice versa. Checking to make sure that the ground on the lines into the MUX ran all the way back to the computer center where the other MUX was located would usually turn up a place where I forgot to tie it together. Balanced line or differential drivers were virtuall immune to this. Gustavus is built on a hillside (well, top of a valley side.) It turns out that the hillside used to me mostly ravine. Over time this has been filled with sand, and whatever else. The sandy soil makes a very good dielectric. Because of this, we attract they type of damage caused by ground differentials, and some argue that we even attract lightning. Some soil conditions are better than others, ours is about the worst. -- Dan Boehlke Internet: dan@gac.edu Campus Network Manager BITNET: dan@gacvax1.bitnet Gustavus Adolphus College St. Peter, MN 56082 USA Phone: (507)933-7596
djdove@hprnd.rose.hp.com (Dan Dove) (06/19/91)
In comp.dcom.lans, johnson@EA.USL.EDU writes: > Also, why should > a coax only be grounded on 1 side of a segment? Why not put a small > resistance at the other end to drain that small amount of current which > might be there? Lee J. Lee, The reason for a single point ground connection IS to prevent ground loops. A problem with your approach would be insuring adequate isolation of the buried cable shield. It is not real difficult, but would add to the task at hand. The Ethernet coax is isolated from ground at each node, and thus the single point connection is required to maintain a safety path for static charge. In addition, each node does have ~1Megohm of DC resistance to provide a charge bleeding capability. I would recommend that you use a fiber link if the co$t is not absolutely out of your budget. FOIRL between two repeaters is a pretty good solution. Good Luck, Dan Dove