Question on EMC

[Freshman]

I have a question on how does EMC issue arise in a PCB board.





Suppose I have a PCB board that has switching DC-DC converters and other high speed interfaces such as Ethernet, and USB.



If my board says, radiates say 75MHz, and it goes above the threshold limit, I will find out the source of the 75MHz.

While debugging, I find that I don't have a 75MHz source, but rather a 25MHz clock.



Now my question is:



1. Why does the 75MHz signal radiate and cause the issue where the 25MHz does not cause the failure in EMC?

Or in other words, how does or which factor (say PCB parasitics, layout, RC component values etc) is responsible for making the 3rd harmonic cause the problem and not the fundamental?

[CaptDon]

A six inch long PCB trace makes a far better radiator of 75MHz than of 25MHz. Being a 25MHz squarewave clock it could have measurable harmonics out to 250MHz. Every unbalanced trace with squarewaves riding on it makes a great radiator and above 100MHz even 10pf of capacitive coupling begins to look like a wired connection.

[Freshman]

Thank you @CaptDon.

Can you please explain how does the 6inch long trace radiates 75MHz better than 25MHz? I am finding difficult to understand this point. What factor determines that 75MHz will radiate better than 25MHz?

Also, when you say, "Unbalanced trace" - do you mean "unterminated trace" ?

[wraper]

Lower frequency (higher wavelength) needs longer antenna to radiate efficiently

[radiolistener]

1. Why does the 75MHz signal radiate and cause the issue where the 25MHz does not cause the failure in EMC?

DC/DC don't use sine wave which consists only one frequency, it uses square wave which consists of a sum of a large number of odd harmonics which are multiply of square wave frequency. It means that if you're using 25 MHz square wave it consists of 25 MHz sine + 25*3=75 MHz sine + 25*5=125 MHz sine + 25*7=175 MHz sine + 25*9=225 MHz sine + ...

And since square wave duty cycle may not be precise, it also may consists a lot of even harmonics (in your case 25*2=50 MHz + 25*4=100 MHz + 25*6=150 MHz + 25*8=200 MHz...)

Higher frequency has shorten wavelength, it means that it needs shorten wire to radiate and as result higher frequency is radiated more efficiently on wires.

Also, when you say, "Unbalanced trace" - do you mean "unterminated trace" ?

No. Unbalanced means that transmission line wires do not have symmetrical electrical characteristics relative to each other. For example, when one wire has different geometry in 3d space than second wire, another example when transmission line has two ground wires and one signal wire or vice versa. Coax cable for AC currents also unbalanced, because AC current flows on conductor surface, so coax cable actually has three wires for AC current - central conductor surface, inner braid surface and outer braid surface.

Such unbalanced transmission line may leads to common mode currents and as result may receive or radiate interference from/to environment, it may work like large coil of transformer, where all bodies around such line will be used as transformer core which will be used to couple your transmission line with other nearby electronic circuits.


Regarding to PCB, such PCB needs to use as small as possible high frequency loops and small conductor surface area used for RF currents. It needs to use proper ground geometry, blocking capacitors and inductors for filtering to block RF current flow through ground and other path.

[Freshman]

Thank you.

Just for clarification:

"Higher frequency has shorten wavelength, it means that it needs shorten wire to radiate and as result higher frequency is radiated more efficiently on wires."

When you say, higher frequency signals will radiate with short traces,

does it mean:

1. These short traces that radiate, should they be terminated? Or is it applicable for any traces with or without controlled impedance traces?

2. Does the radiation happen because of "Resonance" phenomenon? Or can you just shortly explain how does the radiation happen in short traces?

Thanks a lot!

[radiolistener]

1. These short traces that radiate, should they be terminated? Or is it applicable for any traces with or without controlled impedance traces?

it depends on the goal. Technically varying impedance along transmission line leads to high radiation loss, so if you want to transfer signal on some distance you're needs to keep stable impedance across transmission line. But if there is no such goal, you're needs to avoid RF current flow on long distance, you're needs to block and filter it as close to source as possible.

2. Does the radiation happen because of "Resonance" phenomenon? Or can you just shortly explain how does the radiation happen in short traces?

No, radiation just happens when current flow changes direction like it happens for AC current in a wire. Technically it happens when charged particle acceleration is changed. But radiation process is a bit more complex, because some radiated waves are consumed by wire again, how much will fly away and how much will remains around the conductor as oscillated waves depends on wire geometry and frequency, also wave may be reflected from nearby conductors.

Resonance effect just allows to accumulate wave energy in some space, as result it leads to rise AC current flow (due to higher wave amplitude in conductor) and higher radiation loss. It helps to radiate EM wave with a small size conductor but at the same time it also increase heating loss, because higher current from accumulated waves also leads to a higher conductor heating.

It's like soldiers marching on a bridge, when their steps resonate with the bridge, the energy of their step waves will be accumulated in the bridge and when it reaches a certain level it can destroy it. But if they will walk at odds, wave energy will not be accumulated and bridge destruction will not occur.