A few years ago I was doing EMC testing for a product I designed myself. The emissions were so low that at some point the EMC house doubted the equipment was on... I had my phone with me and I showed them that was fully functional. :-)
I toke this comment as a compliment, and I started to get a good degree of respect from this EMC house which made me very happy. By the way... these kinds of results may not be as rare as you could think these days, since there is a boom of low-power products that use technologies like LORA, BLE, or other similar ISM technologies.
Today I want to discuss what kind of consideration I do when I'm testing a product and I see bunches of odd and even harmonics.
This is just an introduction to "how to fight odd harmonics".... and "how to fight even harmonics". The example will be the simplest I can think of.
For example, let's say you are doing EMC testing on a product with just one microcontroller and one clock (as the main sources of emission). Let's say this microcontroller has just one clock (for example 20 MHz clock) and let's assume for now that some errors were made in the project, which is causing some emissions to be above the limits. For example, we have the 11th,12th,13th, and 14th harmonics above the limits.
My question to you is do you know how to fight even order harmonics and how to fight odd-order harmonics?
If the answer is yes, good! excellent! but if you don't know it (yet) then I suggest you keep reading.
From the Theory of Signal, we all know that a squarewave is formed predominately by odd-order harmonics! (3rd, 5th, 7th, etc.) So in an ideal world, I would have expected emission on the 11th 13th order harmonic only. By then why do I see the 12th and 14th harmonic as well?
Let's start by fighting the odd-order harmonics.
Odd harmonics are due to the fact the squarewave has a rising/falling edge too sharp (in other words we have a good squarewave), so if we could make the edge less "sharped" we could reduce drastically the emission on the odd harmonics.
In the example above we could add a small capacitor (e.g. 18pF ) at the output of the clock and I would expect the radiated emission (11th and 13th) to get better.
Obviously, this is not always the case, but you now know what to look for when you see odd harmonics!
What about Even harmonics?
Those are usually due to impedance mismatching ( remember from the theory of transmission line? )
So how to fix this?
What I do is I add a small resistor (e.g. 10 ohms) in series to the clock. (e.g. cut the trace if possible as near as possible to the clock source and insert a resistor on the trace.)
So, EMC troubleshooting is extremely fascinating. By looking at the emission you get hints on the error hidden in the design. Your job is to join theory and practice...
Bye now!
Comments