4 ms·
In a nutshell: If you have to guide an EM wave without dispersion (e.g. TEM propagation), it takes at least two conductors; those have loss. The smaller they
by madengr 7y ago
In a nutshell:
If you have to guide an EM wave without dispersion (e.g. TEM propagation), it takes at least two conductors; those have loss. The smaller they get (think a thin coax) the more lossy they become.
If you need a wide bandwidth, you need a smaller conductor arrangement to keep it from “over-moding” (becoming non TEM). Once it’s non-TEM, you get dispersion and corrupt your signal.
So there is a fundamental trade-off of bandwidth versus loss. Free space propagation is always TEM, so plenty of bandwidth, but now you must direct it with antennas as opposed to guiding it with conductors.
You can get 110 GHz of bandwidth on a 1 mm coax, but it is very lossy, so much so that the microwave industry/research is looking into non-contact wafer probing for mmWave and THz applications.
Fiber has similar issues. It’s extremely low-loss, and non-TEM, but that dispersion is small enough you can multiplex in multiple channels without much dispersion across a single channel. Over long runs, still the dispersion is large enough that it needs to be compensated with various tricks.
A 10G Ethernet copper cable is at most 3 meters, and hard-wired to the SFP modules. But you can buy a 10G mmWave radio and get 10 km. Now several km of that copper cable would be hundreds dB loss.
- MegaDeKay 7y agoFiber guys are lucky that they can cheat with things like Raman amplification that us poor RF souls have no real equivalent for. https://en.wikipedia.org/wiki/Raman_amplification https://en.wikipedia.org/wiki/Raman_amplification
- madengr 7y agoI suppose parametric amplifiers would be similar in the RF realm. At least we have distributed amplifiers, which I suppose could be an amplifying medium if you could make them fine enough.