On 08/04/13 13:41, Michael NEWBERY wrote:
From memory, it's actually HALF the fundamental frequency. I.e., 3Mbps required 6MHz. That's because bits are signaled by transitions. The advantage is that it's self clocking.
It's the fundamental. A down-up transition signals a one, up-down a zero (or vice-versa, I forget). If you transmit all ones, there's an up-down transition between the two bit times, completing a full 10 MHz (at 10 Mbps) cycle. If you look at that on an oscilloscope, it looks like a 10 MHz sine wave. Every bit re-aligns the clock, avoiding the need for nasty hacks like bit-stuffing. A 1-0 (or 0-1) bit pair has no inter-bit transition, which is why the preamble is 63 bits of 10101010... (with ...10101011 to indicate the start of the actual frame). The transitions are all on actual bits, which sets up the clock. On a 'scope, that looks like a sine wave that gets up to a peak, then dips slightly before returning to a peak, then follows a 10 MHz sine curve through the next transition. The 10 MHz waveform just gets pushed 180 degrees out of phase in that case. (I'm not sure what this looks like on a spectrum analyser, the hardware geeks might be able to answer that.) 100 Mbps uses a 3-level code (Manchester encoding is two-level), and gigabit a 5-level code (and more pairs), which make for much more efficient use of spectrum than the Hz=bps relationship of Manchester.
From memory, 100 Mbps runs at around 33 MHz and gigabit at less than 50 MHz.
dave.green(a)snowline.co.nz wrote:
And some of us drilled holes in said coax for the vampire tap. .
I learned the hard way to be careful handling old 10base-5/Thick Ethernet cables, the ground spikes (connecting the cable's braid to the transceiver's signal ground) often stayed stuck in the cable when you removed a tap (the signal spike was more firmly attached), and would do serious damage to your hands if you ran them along the cable (e.g. to coil it). -- don