Your PCB designers and engineers need to understand basic antennas and have a
few formulas at their disposal. Have them think like an EM wave. Where would
you go in the circuit if you were an EM wave? What are the current loops and
what is the area? What potential is on the daughter card and how much
radiation will this produce. I find that doing this and then modeling in NEC
gives me a feel for what is important and what isn't.
 
   Dave Cuthbert
   Micron Technology

From: [email protected]
[mailto:[email protected]] On Behalf Of C N
Sent: Thursday, March 11, 2004 10:46 PM
To: EMC-PSTC
Subject: Re: EMC Checklist


Well, first of all, I had the oppurtunity of having Doug Smith as a boss. 
I learned invaluable lessons from him before he finally decided to break 
free from my former company to start his own consulting business. 
 
The second important thing was having a chamber in which to literally 
live with a product to get to know it intimately. Or have an area for 
somewhat reliable scans. You have to do a detailed analysis of the 
board before you can begin to start recommending things. 
 
Also, please understand that there are obviously some that will remain 
trade secrets and I make no apologies about not sharing those with anyone. 
 
The understanding was that "all things being equal" change one thing. 
This took a lot of work, a lot of scans, of one product. None of this 
may have any correlation to your particular product. But it turns out 
that much of the results were based on some very simple concepts 
and methods. 
 
1. Interference from one interface card to the next as they are lined 
   up next to each other could be reduced by as much as 15dB by 
   simply changing the position of the offending card so that the 
   component side was not facing any other card. In other words, 
   the board itself could be as much as a 15dB barrier to interference. 
 
2. Starting from a 2-layer board and increasing layers by 2, emissions 
    from the board could be reduced by approx 10dB with each succesive 
    2 layers. I.e., if a 2-layer board was emitting at a certain level, 
    the 4-layer version could be 10dB lower. The 6-layer version could 
    be 10dB than the 4-layer. Maxing out somewhere around 8 layers. 
 
3. Moating and bridging some output interface on a board could have 
    as much as 15dB drop in emissions. 
 
4. Ground studs thru the board to the metal chassis in close proximity to 
    interface connectors could have as much as 30dB drop in emissions. 
 
If you are interested in this type of thing, there's no general checklist 
that can be dropped out of the air to fit your particular product to 
produce the specific results above. A thorough knowledge of your 
product's emissions and sources is absolutely necessary. 
 
Regards, Doug McKean 
 

  _____  

Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com



Reply via email to