0

For me shielding is always a difficult question. I'm sometimes really unsure how I should connect the shield. What I know is that the shield should be connected with very low impedance, that for sure, but when does make it sense to connect it to the system GND and when to protective earth?

Let's have a look at two examples:

enter image description here

In this situation, the device is battery driven and is floating with respect to the noise source. I think, when the battery and the device are connected together with low impedance GND, there should really not be any problems, also when there is no shield at all. When a shield is added, I think I would connect it to the GND and NOT to PE also (since no reference,) but I'm not sure if the GND connection to the shield is really necessary, since if this is a Faraday cage, any field penetrating the shield will produce a counter field, that should practically attenuate the fields inside the shield.

enter image description here

This situation I have encountered many times. The device is supplied by a PSU which has a transformer inside and some galvanic isolation to the secondary side. The shield needs to be connected to Earth (electrical safety), but should it also be connected to GND? There are PELV and SELV devices, where one is connected and the other not, but in terms of noise attenuation, what is the best to do in this case?

Edit: About the noise source, assume it to be a high frequency, high voltage source that produces fast voltage transients and a strong E-field in the nearfield. What would happen if the noise source goes more into high frequency H-fields?

JRE
  • 71,321
  • 10
  • 107
  • 188
HansPeterLoft
  • 1,038
  • 1
  • 22
  • 40
  • define the noise source first. – Kartman Apr 02 '22 at 09:58
  • Ok, I edited it. I'm mainly interested in HF HV noise sources, where fast voltage transients occur over larger cables, that act as antennas. An example would be a high power heating element, where the power is controlled with PWM and IGBTs. What me further would interest is also, what if the noise source is both, HF E and H fields? From my understanding, the Absorption and Reflection Loss of the shield will act for both nearly the same, only low frequency magnetic noise sources are more difficult to shield off. – HansPeterLoft Apr 02 '22 at 10:07
  • For a battery powered device why would you need to connect it to PE? How many commercial devices have you seen that require this? Does your mobile phone go berserk when exposed to the ‘noise’ you propose? It would have to be a significant RF field, lightning or large magnetic field methinks. It’s been years since I’ve had to do CE testing so hopefully someone with recent knowlege can quote the required standards that quantify the expected fields devices are expected to withstand. With the mains powered scenario, it probably is not necessary to connect the gnd to PE. – Kartman Apr 02 '22 at 20:00
  • Remember that current flows in a loop, so you need a path that flows through your electronics to create a problem so your examples are somewhat synthetic in that you normally wouldn’t have a magic box with no controls, indicators etc. if all the devices I designed fitted that mold then EMC compliance would be much easier. With switches, indicators etc then they become a path for ESD and RF to enter your system. Once you have external connections is when the problems really begin. Like a Karate master, you have to block and redirect the energy to a more preferable path. – Kartman Apr 02 '22 at 20:08
  • Hi Kartman, I know it looks a little bit synthetic, but my main problem to understand is, if/when it is necessary to GND/PE the shield or if even the shield as a farady cage can be enough? To PE a smartphone makes ofc no sense, but lets assume you have a device near a high voltage pylon in a metal cage, where the pylon is connected to earth. So there the pylon is the noise source with reference to PE. In this case, does it make sense to connect the shield to earth? I would say yes and probably also GND should be connected to PE, since any field inside has then a return path. – HansPeterLoft Apr 02 '22 at 21:11
  • About the box with no apertures, lets nust assume that the apertures are less than 1/20 lambda, such that there is large attenuation for the fields. With field retun path I mean a return path for a current, induced by the fields. – HansPeterLoft Apr 02 '22 at 21:20

0 Answers0