I am building a shield that contains an AC power supply and relay for an Arduino-like board. The problem I am having is that the relay is quite noisy with respect to EMI, and I can see a few jitters or jumps in my system during normal operation even with a zero cross detector.

I have added in another shield between the 'relay' shield and the main board, which has helped the problem by creating a larger gap, but I would like a more permanent and robust solution.

I was thinking of building a simple shield that has nothing much on it and is simply a large ground plane that sits in between the relay shield and the main board. This would be a simple solution that does not involve making any shielding cages.

This ground plane would be connect to the chassis of my device, which is then earthed.

Since the device has both a DC ground voltage (0V DC) and a AC earth (0V AC) I was thinking maybe I should use both as part of my shielding. Since the board has layers on both sides perhaps I could design it to have the side facing the relays to be earthed, while on the other side, facing the CPU, would be DC ground. I could even couple the two planes with a capacitor, if it helps.

On the one hand I think the earthed option should be fine, but on the other hand I would like to know more about using the DC ground as an EMI shield. Would the simple earth only option work? Would even a DC ground only option work? Does it help to use the DC ground as an EMI shield for the DC components or am I over complicating it?

  • 2
    How are you detecting "a few jitters or jumps in my system"? Why do you think it is EMI rather than a voltage dip causing them? How are you powering the relay? Is it from the Arduino's 5V supply? The simplest explanation for most "jitters" around an Arduino is electrical rather than EMI. – gbulmer Aug 30 '16 at 16:06
  • It's a number of things, but mainly if I take away the AC I don't get the issues. Also as mentioned in the question if I move the shield further away I also don't see the issues. I can detect them as I have a sensor that goes offline when it encounters bad signals over the SPI bus. – Brad R Aug 30 '16 at 18:31
  • Unfortunately your question has been migrated to a site where, IMHO, you might be less likely to get the kind of help you'd like. – gbulmer Aug 30 '16 at 19:09

What would be helpful would be explanation of the type of AC you are using and then possibly a schematic or board layout.

That aside, there does seem to be noise between the two boards. The Idea of placing a PCB with just a large copper pour would work.

DC Ground

If you connect the DC ground from the main board you should use a ferrite bead(FB) between the two grounds, this will ensure that you don't get noise getting back into the ground of the main board after you have tried to shield it. You should try to match the ferrite bead to the noise frequency you experiencing (use a scope if you have one).

Using the DC ground as a shield is often done, if you look at the EMI cages on cellular phones, WiFi modules, and other noise generating devices/susceptible-to-noise devices, they are connected to ground. It gives the noise somewhere to go, instead of your signal lines. Thus using the DC ground is defiantly an option


Main ground>--------[__FB__]-------<SHIELD

AC Ground

You could use the AC earth line on the top part of the shielding board. The coupling part might be a bit over kill, leave space for it to test to see if it helps when you make the board. Also AC earth can be used for shielding but is used for the safety aspect of earth leakage.

Don't do a direct DC ground to AC ground unless you know what you are doing as this will create a ground loop, this will change the way stuff behaves and will make your circuit no longer a floating device.

Other stuff

I would also make sure how you drive the relay is isolated from the control circuitry, although from what you describe:

if I move the shield further away I also don't see the issues.

If you share power rails between the relay board and the main board, try filtering the power going to the CPU section. Also the control lines can be isolated between the controller and the relay with opto-couplers, this might stop noise coming back into the circuit from the relay part.

| improve this answer | |

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.