Shop OBEX P1 Docs P2 Docs Learn Events
Designing a DC printed circuit board, ground concerns. — Parallax Forums

Designing a DC printed circuit board, ground concerns.

Zap-oZap-o Posts: 452
edited 2011-10-05 19:43 in General Discussion
Designing a DC printed circuit board…

…Would you tie the outer shell of a USB connector (type B see picture) to the negative or ground of your supply? If so, then would you connect to either the analog ground, or the digital ground to the connectors shell?

I also would ask to only focus on the question and exclude the “chassis” or ‘earth ground” for now. I want to soak it in one topic at a time.
AU-Y1007_sml.jpg

Comments

  • Rick_HRick_H Posts: 116
    edited 2011-10-04 21:56
    It defiantly would go to ground, if someone touched it with a 0V potential it would short out. its carrying a digital signal then it gets a DGND and it would be grounded at the side that is supplying the power over the usb wires.
  • Zap-oZap-o Posts: 452
    edited 2011-10-05 07:23
    Then would you tie earth ground to negative on the supply?
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-05 08:04
    The USB cable's shield should never be grounded at the client ("B") end, but left floating. It's grounded at the host ("A") end. Grounding a shield at both ends negates its effectiveness, because it can then carry return currents. Shields should not carry current.

    -Phil
  • Zap-oZap-o Posts: 452
    edited 2011-10-05 08:39
    Phil

    What if the connector is not connected to a host("A") and someone touches the client end("B"). Should I offer some protection against that type of damage?
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-05 08:42
    I'm not sure what kind of protection you're referring to. Having the shield connected there wouldn't accomplish anything.

    -Phil
  • Zap-oZap-o Posts: 452
    edited 2011-10-05 08:56
    I imagine that a person would approach the circuit board (inside a plastic enclosure) and attempt to connect the USB cable. The person zaps the client ("B") connector and where will the charge flow if I do not connect this to ground and the other side ("A") has not been connected yet?
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-05 09:07
    You do not connect the "B"-end shield to ground, period. See here:

    -Phil
  • Zap-oZap-o Posts: 452
    edited 2011-10-05 09:44
    Ahhh, Well thanks Phil.
  • Mark_TMark_T Posts: 1,981
    edited 2011-10-05 15:38
    Phil Pilgrim (PhiPi) wrote: »
    You do not connect the "B"-end shield to ground, period. See here:

    -Phil

    Surely you ground the client shield if a hand-held battery device but not if a mains powered device? In the mains case you want to prevent a ground loop, but in the hand-held case you want to prevent static-discharge from occuring on the signal lines by arranging the shield to common the device grounds on first connection and take any spark.

    (Actually maybe the correct thing to do is ground the client shield via a 10k or so resistor or an RF-choke to soak-up static discharge while preventing significant ground-loop currents. See "Digital Hardware Design", Catt/Walton/Davidson, ISBN 0333259815)
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-05 15:50
    In my experience, learned the hard way I'm afraid, if a shield is connected at both ends, it fails to function as a shield because it becomes part of the ground return circuit. Any noise coupled to it under those circumstances can cause a ground shift between host and client, which can disrupt normal operation. It doesn't have anything to do with mains or battery operation. Every USB reference I've seen says not to connect the shield at the client end of the cable, so I don't.

    -Phil
  • Zap-oZap-o Posts: 452
    edited 2011-10-05 18:19
    This is exactly why I asked the question. Its a damn if you do and damn if you don't type of situation. I will say that I have never connect the ground to the chassis of the USB connector in any of my PCB designs and I have passes CE testing, not that its the right way or anything, it just gives me a little boost of confidence.

    I decided to ask the forums here mostly because I am reading "The Circuit Designers Companion" and it shows various ground loops and I am very leery about changing things that seem or are working.
  • Peter KG6LSEPeter KG6LSE Posts: 1,383
    edited 2011-10-05 19:43
    I have seen a TON of devices with a inductor to tie V- AND GND .
    Peter
Sign In or Register to comment.