EMC design question

[terryg]

I am designing a digital device that is to be tested to meet FCC Class B. This device is connected to 4 inputs (contact closures) and 4 temperature sensors (analog). The device is powered from a switching wal-wart supply and is mounted in an extruded metal enclosure. I am concerned about the long wires that connect to the switches and sensors. Should I put common-mode filtering on these lines? I plan to use LC EMI filters on each conductor. My concern is that these lines are large antennas which will radiate and conduct.
My second concern is to do with the metal enclosure and circuit board grounding. The enclosure will not be connected to earth and will be floating. The PCB will be 4 layer with a single micro-controller (no external memory bus). Do I connect the circuit ground to the enclosure? I was planning to break the ground plane, connecting it to the enclosure at the input connector with the filters and then connect that plane to the digital plane with a ferrite.
Comments? Thanks in advance.

 

[55pilot]

Designing for EMI is far too complex a subject to tackle here. There are a lot of rules of thumb, but each of them comes with enough caveats, presumptions and exceptions to make them useless as general rules that can be used without further insight and understanding.

Designing so you can pass FCC Class B 100% of the time is not hard to do. You need to start with a few courses that go into the subject matter beyond the rules of thumb. Then you need to find yourself a mentor that can teach you with real life problems in YOUR life and an employer that is willing to allow you to experiment in the early stages of development so you can push the boundary, fall flat on your face (where it does not hurt) and learn from it.

Good luck!

---55p

P.S. All of that assumes that you have the brain power to learn and UNDERSTAND the material. Not everyone does and hence the reputation of EMI as "Black magic". Black Magic is the code word for "I am not smart enough to understand it".

 

[NickS]

Hey Terry, sounds like a fun task (this is the sort of project I have enjoyed for the past 2 years).

It sounds like you have made some good strides in learning some of the precautionary measures you can take. I like to send floating signal lines as twisted differential pairs when possible(it makes a big difference). I also agree with your instinct on grounding for the micro.

Unfortunately all your prep work will not likely bring the finished product(on first try) but that is ok. My RF mentor has 25 years of radio under his belt and he still always demands a three spin design stage. Often he comes in done at two tries but he understands the nature of the work well enough to know that you must expect un-expected problems(if that makes sense). So give yourself time to try, fail and fix.