padama@batcomputer.tn.cornell.edu (padama) (02/03/88)
Given that the maximum length of an Ethernet segment is 500m, it used to be that the cable lengths used to make up this total had to be of one of three specified lengths..23.3m,70m, 117m. Has this restriction become obsolete, and can one now any arbitrary lengths of enet coax as long as the total does not exceed 500 m? Must the coax only be cut at the 2.5 m markings?
robert@richp1.UUCP (Robert Miller) (02/04/88)
In article <3559@batcomputer.tn.cornell.edu> padama@batcomputer.tn.cornell.edu (padama) writes: > >Given that the maximum length of an Ethernet segment is 500m, it used to This is true. >be that the cable lengths used to make up this total had to be of one of three > >specified lengths..23.3m,70m, 117m. Has this restriction become obsolete, and can one now any arbitrary lengths of enet coax as long as the total does not exceed 500 m? Must the coax only be cut at the 2.5 m markings? Never heard of this one. Each segment must be 500m or less. You may connect several segments via a repeater or router of some sort. Note however, that the max point-to-point distance between any two nodes that you want to link should not exceed 1000m. Also, there should be a max of two repeaters/routers between any two nodes that you want to link. The nodes must be tapped/cut at the 2.5m markings( usually a black ring painted on the yellow coax ) or you risk setting up standing waves along the coax (coax is essentially a transmission line) that will degrade the LAN's performance. ............................................................................... "To open, cut along dotted line." ________________________________ _____\______________________________________________ / | Robert Miller @ ihnp4!richp1!robert | | \_____/
ron@topaz.rutgers.edu (Ron Natalie) (02/05/88)
The following is for IEEE 802.3 10BASE5 cable. This is perhaps the most up to date specification for Ethernet (the yellow kind). The sum length of the cable may not exceed 500m and may be made up of sections joined with two male plugs and a barrel (double female). 802.3 recommends: 1. that you try to make up the entire distance from a single cable piece, 2. that you use the same lot/manufacturer on multiple pieces to minimize impedence mismatches, 3. that if you are going to use mixed lengths that you mimize reflections by using odd integral multiples of a half wavelenth of the cable at 5 MHz to avoid the reflections adding in phase (these yield your 23.4, 70.2, and 117 meter lengths +/- .5m) and that if you use only these lengths you should be safe. 4. If you use random lenghts measure the worst-case signal reflection due to the discontinuities does not exceed 7% of the incident wave when driven by a MAU. The 2.5m rings are for transciever placement and not for cable cutting. Discontinuities in the ring spacing at the ends of the cable are explicitly allowed by the spec. -Ron
eshop@saturn.ucsc.edu (Jim Warner) (02/06/88)
In article <95@richp1.UUCP> robert@richp1.UUCP (Robert Miller) writes: >Note however, that the max point-to-point distance between any two >nodes that you want to link should not exceed 1000m. Also, there >should be a max of two repeaters/routers between any two nodes that >you want to link. > The max distance between two transceivers is 2500 m (plus a little more for transciever cables). The max number of repeaters in the path between any two stations depends on whether you use IEEE 802.3 repeaters or Ethernet version 2 repeaters. For the former, the max is four. This is one area where V.2 and 802.3 differ significantly. The Ethernet and 802.3 specs do not address the situation where multiple Ethernets (with repeaters) are connected by routers or other store and forward packet switches. But "routers" is what the NSFNet backbone is made of, and it spans the whole country.