Where to attach cable shielding for portable battery-powered device

• Where to attach cable shielding for portable battery-powered device
  Posted by etaripeka on 30 Apr, 2024 14:08
• Hello,
  
  I'm making a portable device, which mainly consists of the following components:
  1. A LiPo battery
  2. A self-designed PCB - mostly just buck converters to a few different voltages (5V, 12V) to power the SBC, sensors, and a fan
  3. A single-board computer (SBC)
  4. Some sensors (e.g. accelerometer, RS232 @ 115200 baud rate)
  5. A 3D-printed plastic shell (so no metallic chassis to connect the shield to)
  
  I have a shielded power cable between the LiPo battery and the PCB, and some metal cable sheathing covering the wires between the accelerometer, SBC (data) and PCB (power). I'm wondering what is the proper way to deal with the shielding on the cables.
  
  From what I've found reading online, I see:
  1. It's a very complicated topic  :'(
  2. A shielded cable without connecting the shield somewhere is the same as / worse than an unshielded cable. (actually, is there even a point to having a shielded power cable? does it only matter for high speed data cables?)
  3. Connecting the shield on both ends can cause a big ground loop which creates a lot of noise. However, I see some people mention that a single ended connection can just become an antenna which is also not ideal. Connecting it on one end for a portable device is to some the "least worst" option?
  4. Since the shield cannot be connected to a metal enclosure, connecting the shield to the ground plane on the PCB seems to be recommended. And even then, there are a lot of different practices of putting wildly different sizes and combinations of capacitors  / resistors / inductors between the shield and ground.
  
  Some links I've read:
  https://www.eevblog.com/forum/projects/usb-shield-ground-connection/
  https://electronics.stackexchange.com/questions/661902/usb-shielding-device-or-host-side
  https://www.eevblog.com/forum/projects/lets-talk-about-cable-shielding-and-pcb-gnd-layer/
  https://electronics.stackexchange.com/questions/689066/cable-shielding-best-practices
  
  Hence, I'm still a little unsure of what I should be doing, as it seems like the answer is always some kind of "it depends/its a trade off between ...".
  The simplest answer I see for my case is just to connect the shield to the ground plane of my PCB.
  
  Hence, I was thinking of:
  1. (For the data cables) Placing separate screw terminals on my PCB (my board connections are mostly using screw terminals), next to the screw terminals where the cables conductors terminate, that I connect the shielded braid of the cable to. These screw terminals are simply connected to the ground plane. Or should I add some of the capacitors/resistors/inductors in between? Some have recommended a "semi-isolated" ground plane for this "I/O area"?
  2. (For the power cable) Maybe just crimping the metal braid of the shielded power together with the GND conductor into a ferrule, and connecting that to the PCB.
  
  Is this a terrible idea?
  
  
• #1 Reply
  Posted by xvr on 30 Apr, 2024 17:47
• Connect all shields (one end, nearest to SBC/PCB) directly to ground on PSU (I guess it will be negative pole of battery in your case)
  
  There is no ground loops, because only one ground and only one connection.
• #2 Reply
  Posted by etaripeka on 01 May, 2024 07:39
• Oh yea, I'm being dumb; there is no ground loop. Thank you!
• #3 Reply
  Posted by Terry Bites on 01 May, 2024 11:57
• The way I always look at it is: are there potentials that can cause a current to flow in the shield, if so, stop them.
  
  For power/ motor cables where the small CMVs don't matter, its normal to ground both ends to maintain an equipotential bond.
  That also applies to large signal cables like STP. Obviously shield is there to keep the crap in, not to keep it out.
  
  As noted in other replies, for low level signals its good practice to ground only the at the measuring end.
  Tying it to ground via a small resistor adds some CM rejection to s single ended measuring end, say 10-100 ohms.
  
  For EM induced currents, tie the shield to ground through an inductor. Unexplained noise and offset voltages are often casued by RF dancing on the shield.