What're design rules to consider in order to have the best chance of complying with EMC requirement?

Thread Starter

Younes Thabet

Joined Jan 9, 2019
144
Hello guys,

I am designing a 2 layer PCB that contains a power part (PSU, Relays) and digital one (MCU, digital & analog I/Os, ..)...
I looked up some design rules that should minimize EMI but i am not sure if those rules I applied are sufficient for my board to be EMC compliant!!
here are the rules I considered:

  1. Routing signals/pwr on top layer and ground bottom.
  2. Adding decoupling caps close to ICs.
  3. using ground plane stitching.
  4. Avoid long traces on I/O and digital signal lines.
  5. Avoid putting high-speed circuitry between external mix-signal connectors.
  6. following layout recomendations for ICs.
  7. and some obvious layout rules here and there....

what other rules that I could consider to make my board complying with EMC requirement?
How I can know for sure that my board will comply to EMC requirement?

Thanks,
 

Deleted member 115935

Joined Dec 31, 1969
0
How can you be sure you ask,
Answer you can't
I'm afraid it all comes down to experience,
the tools that exist to predict, are both VERY expensive, and very difficult and time consuming to get realistic numbers out of them,

So ,
to add to the confusion,
there are many different types of interference to worry about,

EMC / ESD / EMI are all part of the same problem as far as a system designer is concerned.

This might help as a start

https://emcfastpass.com/rightfirsttime/
https://en.wikibooks.org/wiki/Practical_Electronics/PCB_Layout
http://learnemc.com/


On board, one signal interfering with another.
answer is to keep signals that are sensitive further from possible noise sources than they are from a ground.

On board, signals radiating off the board,
answer is to keep possible radiators in a short loop., Think of an aerial, the bigger it is the more it can radiate.
Also consider an earther box.

Signals coming out of connectors.
Those dammed cables you connect to your board make great aerials, ( see above )
Answer is real pain, you either slow the signal down by a low pass filter on the output,
but that limits the speed you can work at
You use differential signals, which also tend to have small signal swings,
but if your interface is single ended, you can't
or you use a shielded cable, where the shield does not conduct any signal

signals coming into connectors
ESD is the killer here, use input ESD diodes on or in the connector,

Generally,
look at the transmitting loop area,
put decoupling right on the IC pins,
put as much gnd over the board as you can, ( no point etching it off and giving the Cu to the makers )
if you have a switch mode PSU, the manufacturers provide great tools to help you design them,
 

DickCappels

Joined Aug 21, 2008
10,180
The first thing you need are the EMC regulations for the region of the world in which the product will be sold and/or used. Those will spell out exactly what needs to be taken care of and in many (every?) case describes how to measure the phenomena.
 

PadMasterson

Joined Jan 19, 2021
63
Unless you are really pinching pennies for this board, I would use a flooded ground on both top and bottom of the board and route your powers if possible. (like "crutschow" mentioned above) I never like to say this, but you might consider a 4 layer board with ground on both internal layers and route powers or even flood them on the surface layers. You may find that the cost difference is not all that much for a better design. You also didn't mention the edge rate of the signals on the board or it's size. These are factors also since if the board is very small, you may not have many or no issues, but that's dependent on the rise times of the signal edges. Be sure that any connectors leaving the board have filtering on the signals is important since as was mentioned, cables are really good antennas and typically not what you want them to be. I didn't look at the links mentioned above, but it looks like they would be a good place to start. There are a multitude of answers to your question, just like there are a lot of ways to solve them. The good news is, your first step was asking... :) Good luck.
 
Top