jas@proteon.com (John A. Shriver) (06/07/89)
Using type N tees like BNC tees is not the same. Your error rate will go up a little. Thin ethernet cables are magic lengths (multiple of a bit long), so all the nodes are in phase. Thick Ethernet puts the connectors in the wrong place for nodes, compared to the stripes. (Out of phase.) The reason is that by tapping into the cable off the magic points, you will get more reflections, and they will be out of phase. What will happen is that those reflections will hit the receiver well after the preamble, after the receiver has locked on it's clock frequency and phase. The reflection will cause a small phase shift of the received signal. There is a possibility that the receiver will have to abort, due to decoding errors. Most chips flag this as a collision fragment. You are making compromises of the same sort that apply to thin Ethernet. If you configure your thick Ethernet by thin Ethernet configuration rules (less nodes, less distance), then you should be OK. You can look at what you are doing to the cable with a Time Domain Reflectometer, and see the reflections. An Ethernet is not just ordinary wire, it is an RF transmission line. One must treat the cabling specifications with respect. (See sections 7.3 and 7.6 of Ethernet V2.0.)