I've been conducting a long-running personal project, designing and building a custom PC case. I've got round to the front IO panel and figured to design and make the circuitry myself instead of buying something that wouldn't quite look right. The panel will use a PCB with 2 USB A / 3.2 receptacles, power button and HD audio (and power LED).
During designing the PCB i've discovered a few things, one being that the differential signal lines (D+/-, SSTX/RX) require a specific 90 Ohm impedance. I understand some of the theory behind this, but not all, and am hoping someone can fill in my gaps:
I understand that impedance matching is required to prevent signals from "bouncing" at the boundaries, for instance between the usb port and the PCB traces or the traces and the header. I also understand that the traces need to be approx. identical lengths so that the differential remains "in sync". That all makes sense to me.
I've seen in numerous places that to acheive the required impedance, a "Ground plane" is required. This ground plane provides some sort of effect to the traces carrying the signal that affects their impedance. One question I have is: what is the role of the ground plane here?
Take for instance my USB 3 scenario:
Image source: eTechnophiles - 10 USB Pinout Explained- USB A, B, C(Male and Female)
USB 3.0/3.2 Gen 1 has 3 sets of differential signals, V+, and 2 Ground contacts. Likewise, the USB PCB header on the motherboard will have ground contacts that correspond to the USB receptacle. Does the ground plane just connect to any contact named "ground", regardless of what component it is? Are there "types" of ground? Should I only connect the ground plane to a specific set of contacts?
I guess what I'm getting at here, is asking if all components can share a common ground. Will 2 USB ports + HD audio sharing the same ground lead to any complications?
Additionally, I have seen discussion about how the spacing/width of the diff. line traces, as well as their separation from the ground plane, affects the impedance. In these discussions there is reference to online calculators that allow you to calculate the impedance expected for a set of width/spacing/distance parameters, but upon using them, all of these calculators give different results.
Is there a quick cut rule for this sort of thing, in regards to typical width/spacing/distance from ground plane? Can anyone also fill me in on why there is such contention here also? The only part I can kind of understand is the "skin effect", which makes sense, but when I go to online calculators for AC resistance for the relevant factors (100 micron trace diameter, 10-20 mm length, 1GHz frequency), the resistance is shown as negligible. I'm guessing there's other factors at play, but at this point I'm stuck.
Very grateful for any clues or information anyone can give me here. It's been fun learning about all this stuff, but truth be told I am a little overwhelmed.