REV05 USB Shield Termination

[andrewortman]

Version Number

REV05b0 Review Board

Bugfix or Enhancement

Potential Bug

Description

From my understanding, the USB shield should only be connected on the Host (PC) side of the connection, and unconnected (or at most capacitively coupled to ground) on the Device Side (REV05b0 MOBO) - R59 in the board posted on discord (https://discord.com/channels/688723670461448225/739250421893300264/1192530637593264148) appears to be populated with a 0 ohm resistor as mentioned in the schematic.

Suggested Solution

I believe there is a lot of debate on what to do with the shield pin. See this stackexchange conversation: https://electronics.stackexchange.com/questions/661902/usb-shielding-device-or-host-side. Since EMC is an important design goal of this revision, it may be worth at least documenting why/why not the shield pin on the device side is connected to the device ground plane, or at least doing EMC testing with different designs (directly connected, unconnected, ferrite bead, cap, etc)

[warasilapm]

Cutting to the chase here: you're doing exactly what you should be doing for this problem.

Connecting to ground through a zero ohm makes the most sense at first blush here, and it may be "just fine"™ to leave it that way. Ultimately, though, you're going to want to validate this design against what you think are the most hostile expected conditions and how they affect your circuit. It could be HF interference. It could be LF interference (I doubt it, given the motor drivers are attached to their own huge antennae). It could be ESD! I suspect your main issue will be ESD.

So, if it were me, I'd leave two 0805 pads and an SMA or SOD123 package in parallel between the shield and ground, unpopulated with a trace shorted across one of the 0805 pads (like you used to do under the motor drivers). If there are no issues connecting straight to ground, you're good to go. Once you test against your expected environmental hazards or certification requirements, you've left yourself tools to work with.

The only reason I say two pads is that it gives you the possibility to do an RC in parallel (instead of series). This is a better approach for passing HF interference because it allows you to provide a drain for charge that would otherwise build up on the shield side of the capacitors and provides at least a first line defense against ESD on the lines instead of just waiting for the capacitor to fail internally or voltage to short across the pads.