You don’t always want the shield coupled to the signal ground. If it’s coupled on both sides then you can have all kinds of fun problems. Ideally, the shield should be connected to the ground on one end
> Ideally, the shield should be connected to the ground on one end
Depends on the signal. With low-frequency stuff (like audio) you want to ground the shield on one side to avoid ground loops, but with high-frequency stuff (USB 3, for example) you definitely want to ground the shield on both sides: if you don't, you end up creating an antenna which is really bad for your EMI performance.
USB 1.x happens to be at an in-between point where neither issue is a big deal. The connector spec literally says "use accepted industry practices", so either approach is fine if it manages to pass emissions testing.
That goes against all the guidance I’ve read, but I’m an amateur so you may be right. I’m confused on why it relates to signal speed though. For instance, I’ve read that this guidance also applies to Ethernet which is quite high frequency and based on differential pairs
> Question: “I have been told that you should only ground the Ethernet cable at one end. Is this true?”
>
> Answer: “Sometimes yes, but only if you are seeking to avoid a ground loop. You likely don’t have to worry about this. If you are running shielded Ethernet between two structures that have their own separate AC systems or separate grounding rods installed, then YES, you should worry about what is known as a ground loop. A ground loop occurs when you have actual conflicting AC ground systems. Merely bonding your Ethernet cable at multiple points to the same AC grounding system in a single structure does not create ground loops. Think in terms of ground rods. How many ground rods are involved? If you have more than one with more than one structure involved then you are wise to bond to ground at a single end only.”
