2
\$\begingroup\$

I have a controller board powered by DC that may or may not be isolated from earth ground. The chassis of the device is non-metallic. This board connects through a long (30-40m), shielded cable to a small sensor board. The sensor board does not electrically connect to anything else, and is also housed in a non-metallic enclosure.

I know the usual "best practice" recommendation is to connect the cable shield to ground at only one end, and this recommendation makes sense when that grounding would be to a conductive, earthed chassis. Any noise induced on the shield would be shorted to earth and be kept away from my sensitive circuits.

However, in my case there is no conductive chassis, no separate earth wire connection to the system, and even the ground return of the DC input to the system may be earth-grounded or it may be battery powered (and thus completely isolated from earth). Does it still make sense to connect the shield to the ground of my circuit in this case? that would mean connecting it to the signal-ground layer (as there is nothing else)... I worry that any noise induced on the shield would just couple more effectively to my circuit through the suggested connection to ground. I'm essentially connecting a huge antenna to my circuit...

On the other hand, I heard everywhere that a completely unconnected shield is as bad or even worse than no shield at all.

So, what would be a good grounding scheme in such a situation?

\$\endgroup\$
  • \$\begingroup\$ Decide in context of the system. \$\endgroup\$ – Gregory Kornblum Apr 6 '17 at 19:53
  • \$\begingroup\$ I realize the answer depends on the context, that's why I tried to provide it in the detailed question... what other information is required to decide, would you say? \$\endgroup\$ – Tinkerer Apr 7 '17 at 20:33
  • \$\begingroup\$ Oh... If would have to describe my considerations in my system, this o e page would not be enough. It's standards you have to comply, client's equipment, technicians education... Electrically- just make sure current has where to flow (ok, for short cables). But practically, what kind of capacitor to use may depend on safety requirements of your specific client. \$\endgroup\$ – Gregory Kornblum Apr 7 '17 at 20:35
  • \$\begingroup\$ By the way, sometimes same product has different options for different applications, just because tberenis no single solution for all of them. \$\endgroup\$ – Gregory Kornblum Apr 7 '17 at 21:21
2
\$\begingroup\$

You could use a 1M resistor and 0.1uF capacitor in parallel to connect the shield ground and board ground together. Our board designs at my work do this. It essentially grounds the shield while decoupling the noise that you're concerned with.

enter image description here

\$\endgroup\$
0
\$\begingroup\$

Well the only correct answer will depend from what do you really need to protect your system.

I will try to cover most common situations although it will need a book to do that properly.

First trap - 30-40m cable, you will most definitely have common-mode problem if you are not careful. Be careful about grounding(read Olin decoupling answer), try to use isolated communication transciever and use good twisted pair cable.

1)I want to prevent electric coupling between conductor and noise source Use nonmagnetic shield and connect ground on one side.

2)I want to prevent magnetic coupling between conductor and noise source Connect both shield sides to ground. Yes you will have shield current flowing, but if done carefully you could use that phenomena to cancel the field produced by conductor itself.

More precised answer is possible with better insight regarding the details.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.