[alexsv]

  I'm curious as to why both housings need grounding. Is the chassis not in parallel with one of the leads? Might actually increase noise with multiple paths if so. I'm not arguing but asking here.

 I'm unfamiliar with USB protocols or cable requirements.

The USB 2.0 standard specifically specifies that the shield needs to be connected to both housings. They don't specify the shield material so I suppose braided aluminum would work.

 

[Speedskater]

How a circuit or interconnect system works at power line or audio signal frequencies is very different from a radio frequency circuit.  A parallel but separated path may have very high impedance/resistance at radio frequencies.  In an RF current caring wire over a ground plane, all the return current through the ground plane is directly under the wire.  The circuit self inductance is lowest when the Send and Return wires are closest to each other,

 

[goodvibes]

  The USB 2.0 standard specifically specifies that the shield needs to be connected to both housings. They don't specify the shield material so I suppose braided aluminum would work.

For some reason I was thinking ferrous instead of copper. Senior moment. Copper should be fine and a lot easier to solder. I wonder of that ground is shared in PCs. If so. it may be advantageous to just tack the pin off the housing connection for the power supply portion.

 

[dvw]

USB signal is differential signal. There need to be a common ground for it to work. In most cases, shield is the common. The key specification is the impedance as well as the the delay time. For audio application the USB port is more forgiving because most DAC chip designs are USB1.1 and not 2.0. With a lower speed and purposedly shorter cable design, the cable will most likely work. However, the danger of DIY if you don't know what you're doing is not simply just bit error. The reflection caused by the mismatched impedance could piled up at the transmitter and blow the driver. It is always better just buy CERTIFIED USB cable.