As I said in my last post, sometimes I' am contacted by companies to help them pass EMC. I like the challenge and I know that I will learn something new.
In this post, I'll share some things that I do at the beginning of an investigation.
When you have a product that doesn't pass EMC, most of the time the radiated emission will give you enough tips to guide you on where to start the investigation. It is like a photo on a crime scene, you have some tips and your job is to put all of them together to resolve the puzzle.
Another reason why you should start fixing the radiated emission first is that once you have fixed it most of the time the radiated immunity is also improved.
What I do I divide the emission into 2 or 3 ranges.
Frequency below 100 - 200 MHz, the emission is usually coming from the cables... while frequency above let says 300 MHz up (usually there is something wrong at PCB level).
Tor frequency below 100-200 MHz the usual suspects are DCDC converter, AC/DC converter, dimming circuits, etc.
Then by looking at the emission shape you can create a list of suspects...
For example, look at the photo attached to this post. you will notice that in the range 35MHz - 70 MHz there is a typical shape of the emission of a DCDC converter.... (they always look like a bell spread over a range in this case of 30 - 40 MHz ). Sometimes the emission can be even at a lower frequency range... the point is that it looks like a bell.
Then you need to prove it!, and there are several ways... depending on the kind of DCDC converter the cure could be different...
(maybe next time I may discuss this)
What I want to say here is that EMC is not "black magic" and if you have enough experience you can identify the most likely cause of emission by just looking at the radiated emission graph and the frequency.
Therefore you should not be intimidated by visiting an EMC lab instead you should go and learn something new. You will start to see your design, in the frequency domain rather than in the time domain and your skill will grow exponentially!
Comments