4
\$\begingroup\$

I have been doing some research on this question and, while I found a number of questions here asking something similar, I don't quite get the side-by-side comparison I am looking for. Therefore I decided to ask my own version of the question.

I am designing a PCB which will eventually be fitted inside an aluminum enclosure and used outdoors exposed to the elements. A number of other devices are connected to this box through potentially long lengths of cable. Since this device will be exposed, I need to do my best to protect the PCB from nearby lightning strikes and similar events. The enclosure itself has a large copper ground lug to which an external earth ground strap is attached. The ground strap is connected to a large grounding rod pounded deep into the earth. My question pertains to the connection between the PCB and the aluminum enclosure.

There are two options that I would like to compare:

  1. A heavy-duty wire about 3-6 inches long is soldered directly to a large earth ground pad on the PCB which connects to a copper pour covering the top and bottom layers of the board. This wire then connects to the back of one of the ground lug mounting screws with a large ring crimp connector.

  2. Earth ground contact is made between the PCB and the aluminum enclosure using four large copper-plated mounting screws which secure the board in the box via the provided tapped mounting holes. The mounting holes/pads on the PCB are plated through and connect the PCB to earth ground by way of the copper-plated screws.

Which of these two options is preferred to ground the PCB and protect against EMI and/or large surge events? Obviously I am looking for the lowest-impedance option, which intuition would tell me is the one where the ground connection is made through the mounting screws, but there has been some discussion of the opposite being true.

\$\endgroup\$
8
  • \$\begingroup\$ This question is almost impossible to answer. We don't know anything about the diameters (impedances) of the proposed solutions. In my opinion: If you really need the lowest impedance option, why don't you just apply both options? \$\endgroup\$
    – Stefan Wyss
    Commented Apr 22, 2019 at 18:30
  • 1
    \$\begingroup\$ Could you place a lightning rod near to your device instead of using your device as a lightning rod? \$\endgroup\$
    – Andrew Morton
    Commented Apr 22, 2019 at 18:33
  • 1
    \$\begingroup\$ Aluminum shouldn't come in contact with the earth/soil and cannot be in contact with alkalines or embedded directly in concrete. You'll require qualified bimetallic elements to transition between copper and aluminum. Water running off copper will oxidize aluminum, too. And you can still get side-flash/arc-over inside a box. Perhaps more than you want to read, but perhaps look over NFPA 780, UL 96 & 96A, and LPI 175. \$\endgroup\$
    – jonk
    Commented Apr 22, 2019 at 18:48
  • 1
    \$\begingroup\$ @AndrewMorton Nearby lightning strikes, not direct strikes! The long cables coming into this box act as antennas and if there is a lightning strike (even within a couple of miles) it can induce current in the cables, causing a significant voltage spike. I have surge protection on the board (gas discharge tubes, chokes, TVS diodes, etc) that shunt the excess energy to ground. None of that would survive a direct strike though, and I am well aware of that. \$\endgroup\$
    – DerStrom8
    Commented Apr 22, 2019 at 19:11
  • 1
    \$\begingroup\$ @jonk Right, I would not expect this device to see earth ground current very often, and when it does it would be a very brief transient. I imagine direct contact would be my only concern at this point, unless long-term testing suggests otherwise. \$\endgroup\$
    – DerStrom8
    Commented Apr 22, 2019 at 19:27

1 Answer 1

Reset to default
3
\$\begingroup\$

Using method #1 as a basis, the wire is supposed to go to to a large pour that is NOT the ground plane on the PCB, but is split from it yet still connects at one point. All other things that might be hit by transient events must also connect to this plane (like connector backshells and cable shields).

The purpose of this is to provide a path for transient currents to flow to to ground while keeping the PCB at the same potential as the transient, yet not allowing the transient to flow through the PCB on its way to earth.

This automatically rules out #2 because having multiple connections to ground means that the transient current might flow THROUGH the PCB on its way to earth. Even if the earth plane is separated from the ground planes, the fact that they overlap introduces capactive coupling between the planes.

Connecting to just one point on the PCB with no split plane has a similar issue in that current spreads out when on the ground plane and this can flow under components which is why you want the split and connection to the ground plane at only one point.

If you can find a copy of Henry Ott's book it covers this in more detail in Chapter 15.

Here is a table that compares the impedance of soldered vs screw connections. It's not directly applicable to your scenario, but I think it indicates the copper screw and riser method should be fine, as long as it's only connected to one point on the PCB. Since it's only a single-point connection, can connect straight to the ground plane in this case too since transient currents in the enclosure shouldn't be flowing through the riser. You don't need a mini-earth plane on the PCB (that is the conductive enclosure itself).

This connection point to the enclosure should be as close as possible to all cable shield connections to the enclosure, if you have any.

enter image description here

Taken from Electromagnetic Compatibility, Henry Ott 2009

\$\endgroup\$
7
  • \$\begingroup\$ I am well aware the earth ground is not the same as the circuit ground. It is connected to the circuit ground by way of a Kelvin connection right near the earth ground pad. The pour on the top and bottom layers is intended to act as a sort of "shield", while I have my circuit ground on an internal plane. This rules out your ruling out of #2 because the four connections are on the earth ground pour, not on the circuit ground plane. Therefore, my question is still valid. \$\endgroup\$
    – DerStrom8
    Commented Apr 22, 2019 at 19:15
  • 1
    \$\begingroup\$ Is the earth ground pour overlapping with the ground pours in the #2 scenario? Because that would still present a similar issue due to the capacitive coupling between the planes. \$\endgroup\$
    – DKNguyen
    Commented Apr 22, 2019 at 19:19
  • \$\begingroup\$ Hmm, that is a fair point which I failed to take into consideration. They do indeed overlap, which would allow capacitive coupling. There was one other option that may solve this issue - Having a small copper island around one of the mounting holes on the earth ground net. This island would connect to the circuit ground through a Kelvin connection, like before. The remaining three mounting holes would be isolated. This provides a single earth ground point that does not overlap circuit ground. Perhaps this would be better than having a 12AWG wire soldered to the board? \$\endgroup\$
    – DerStrom8
    Commented Apr 22, 2019 at 19:24
  • \$\begingroup\$ Would I be correct to say that boils down the question as to whether a directly soldered wire vs a copper screw would be better for a single-point connection for the PCB to earth? Let me go through my book. I seem to remember it had comparisons for the impedance associated with different types of connections. \$\endgroup\$
    – DKNguyen
    Commented Apr 22, 2019 at 19:26
  • \$\begingroup\$ That is effectively the question, though really the goal is to determine whether or not a screw connection would work well enough to replace the soldered connection, which is more costly and requires more effort to make. \$\endgroup\$
    – DerStrom8
    Commented Apr 22, 2019 at 19:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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