Here's my understanding, based on decades-old experience so the numbers may have changed over the years:
The impedance of a wire is determined by its dimensions, what other conductors are nearby, the distance to those conductors, and what material is used for insulation. On PC boards, a wire over a ground plane gives you 50 Ohms with a reasonable thickness of the insulating layer. You could increase the impedance by using a thicker insulating layer, but then you have a thicker board.
For USB and CAN, the dimensions of the twisted pair and insulation type and thickness give you the 90-120 Ohms.
Coaxial cable typically gives you 50-75 Ohms.
Being off by a few percent doesn't matter. An impedance mismatch gives you reflections, but if the mismatch is small the reflections are small compared to the signal.
The trade-off is between dieletric breakdown and power delivered to the load...for a fixed inner conductor diameter one could increase outer diameter to increase breakdown voltage...but the characteristic impedance would increase and that would make power deliverable to the load smaller...so there is a well defined ratio of the of conductor diameters that maximizes power handling capability of the impedance of the coax
The tale I heard is that a 1/4 wave vertical antenna on a right-angled ground plane is about 36 ohms, and a non-folded dipole above the ground effect is about 72 ohms. 50 ohms is a nice round number in the middle.
A 1/4 wave vertical antenna with a ground counterpoise angled downwards (such as might be found on the rooftop of a ham radio operator) is a very close match to 50 ohms. When properly tuned mine usually measured better than 1.1 to 1 standing wave ratio when fed with 50 ohm coax cable.
HI Glen. "The tale I heard is that a 1/4 wave vertical antenna on a right-angled ground plane is about 36 ohms, and a non-folded dipole above the ground effect is about 72 ohms. 50 ohms is a nice round number in the middle."
That's funny! SWRs don't worry about round numbers, and with most antennas you can get a really good SWR with 50Ω cable.
I think the natural resonance of a folded dipole is around 50 ohms, so that's why 50 ohms is so commonly used. But TV antennas seem to use 75Ω, as do satellite dishes.
I remember my comms textbooks had a lot to say on this subject, but they are back in Zimbabwe - possibly.......
@sukeshv - "How can that be related to PCB environment..??"
Good point - we do tend to get off-topic here. Betajet's explanation is good - PCB traces approximate to 50Ω and it is (happily) a commonly used impedance. But it will vary with the thickness of the board, track size, etc as Betajet points out. Also ECL - for years the predominant high-speed logic family - works at 50Ω as well, maybe that has something to do with it?