Author Topic: Stripline Stackup for EMI  (Read 356 times)

0 Members and 1 Guest are viewing this topic.

Offline KasperTopic starter

  • Frequent Contributor
  • **
  • Posts: 748
  • Country: ca
Stripline Stackup for EMI
« on: May 06, 2020, 08:31:08 pm »
Hi Everyone, thanks in advance for any help.

I have heard that running clock traces internally reduces EMI since outer planes act as shields. 

Sounds simple but is it always true?  What about a 4 layer PCB with relatively thick middle as shown in stackup 1:

Stackup 1:
1 - GND
d1/2
2 - SIG
d2/3
d2/3
d2/3
d2/3
d2/3
3 - PWR
d3/4
4 - GND

Assumptions:
1 - Majority of return current (from signal on layer 2) travels through layer 1 since it is much closer to layer 2 than layer 3 is.
2 - Current travelling through an external plane emits as much EMI as an equivalent amount of current travelling through an external trace would.
   
Conclusion:
Stackup 1 would radiate as much EMI as if layers 1 and 2 were swapped (signal on external layer).

Questions:
Are my assumptions correct and my conclusion valid? 

Is it fair to say a relatively thin middle as shown in stackup 2 would be best for EMI?
   
Stackup 2:
1 - GND
d1/2
d1/2
d1/2
2 - SIG
d2/3
3 - PWR
d3/4
d3/4
d3/4
4 - GND

Is it ok to use PWR as the return plane?  Would stackup 3 be better?

Stackup 3:
1 - GND / PWR
d1/2
d1/2
d1/2
2 - SIG
d2/3
3 - GND
d3/4
d3/4
d3/4
4 - GND / PWR
   
Would an even stackup (all dielectrics having same thickness) be sufficient?

Looks like stackup 1 is more common than 2/3, is that right?

Would stackup 2 or 3 cost more than stackup 1?

Are there other things I should consider?

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf