Quoting "Aaron S. Joyner": > Jim Ray wrote: > > >>Do note that a traditional, old school, hub does not "reform" the > >>packets in the manner that you're attempting to do, to extend an > >>Ethernet segment. You need to use a switch in that situation. > >> > >> > >[JR>] signals propagating in active digital transmission circuits whether > >hub or switch go through transistors that snap 0's and 1's back into shape. > > > > > > > Where "shape" is defined as a square wave on an oscilloscope, sure. > Degradation of "shape" or "height" in that respect are all byproducts of > attenuation, and as I mentioned earlier a hub will correct for > attenuation related problems. What it won't correct for is timing > issues. Consider a 100m cable run, operating at 100MBits / second. > Remember that both ends have the potential to attempt to talk at the > same time, this isn't a frame or token based network where everyone > talks in an agreed-upon order. So let's say one end raises the voltage > to 5v (I don't recall the actual voltages on Ethernet, these are for > example purposes) - the other end won't see that voltage raise due to > propagation across the copper for _at_best_ 334 nanoseconds (assuming > the speed of light in a vacuum, in practice it's much slower, thus > longer). If the voltage ever goes to 10v, it means that two stations > tried to talk at once and a collision occurs, and everyone has to try > again. The farther / longer that signal has to travel before being > received by everyone else on that unswitched segment, then the higher > your chances of collisions. As a final note, if you only have two > machines on the segment, and they're operating in full duplex mode, this > problem is really a moot point - but if that's the case it's highly > unlikely that you're also going to have a classic hub involved. :)
100Mbps isn't exactly a square wave, I believe it's 3 level signalling with shaping to prevent high frequency components (of square waves). > I'll freely admit that I'm stretching the limits of my knowledge at this > point, but if someone with a stronger physics or EE background wants to > step in and clarify the velocity (to borrow an RF term) of Cat-V cable, > feel free. I have MSEE and coursework of MSPhysics but generally what you're looking for is the propagation velocity; approximately 0.7c (c = 3x10^8 m/s). Optically it goes as c/n (n=index of refraction) but electrically you have to look at the problem as a waveguide. More details follow, gotta pick up my son from daycare :-) -Rob -- TriLUG mailing list : http://www.trilug.org/mailman/listinfo/trilug TriLUG Organizational FAQ : http://trilug.org/faq/ TriLUG Member Services FAQ : http://members.trilug.org/services_faq/ TriLUG PGP Keyring : http://trilug.org/~chrish/trilug.asc
