Radiation Emission Test Problem

[zeallin]

Hello my friends,

Thank you for the amazing Olimex A20 board! We are developing product base on it.

Currently are working on a EMI test but it is not passed (both vertical and horizontal).
We found that strong radiation is detected around the ethernet PHY (Port Physical Layer) chipset, and It should be the cause of EMI test failure. The emitted radiation is distributed based on odd harmonic relation (195.87, 250.18, 375.31, etc)

Do you guys have any suggestion on removing the radiation noise so that the EMI test can be passed ?

Here is the radiation test report from TUV lab:



Thank you very much for your support!

Best regards,
Zeal

[LubOlimex]

The only way to reduce it would be shield the board in some suitable cover. Maybe metallic box. Since you care if the whole product passes the certification you don't need each of the components to pass it, only the product as a whole.

[zeallin]

Hello LubOlimex,

Thanks for your suggestion. However we tried to cover the A20 board with metallic box but no luck.
Do you have any other suggestion on how to lower the radiation?

Thank a lot for your help!

Best regards,
Zeal

[zeallin]

Hello my friends,



We found that the 125Mhz reference clock (as marked on the image) should be the source of radiation noise.
We would like to ask if there is any configuration we can try to disable this clock though system kernel of build settings?

Thanks in advance!

Best regards,
Zeal

[soenke]

That clock signal goes into the A20 CPU so you would have to disable it on the RTL-Chip which i think is not possible.

Anyways i think ethernet wont work anymore if you disable that clock signal.

If it is made correct a metal case shields this frequency at 100%. Maybe you forgot a proper grounding of the metal case? If grounding is not sufficient the metal case will work as an antenna instead of a shield.

[zeallin]

Hello Soenke,

Thank you very much for your advice! We will try again.

Best regards,
Zeal

[zeallin]

Hello soenke,

Sorry we are not sure about how to proper ground the metal case. We tried all the methods we know but still no luck.
Would you kindly please give us an example of how to ground the metal case to make it 100%  shield?

Thank you very much!

Best regards,
Zeal

[soenke]

I would simply connect a cable from a screw connector of the metal case to PE (building ground) like it is done in any metal case covering electronics. Also you should connect the GND of the A20 to this PE so you have a common ground. The metal covers of the RJ-45-Connectors should have direct contact with the metal case, probably with the metals springs on the RJ-45.

You should mesaure a significant difference with and without PE connection.

The wavelength of the ethernet clock is by far too long to get through small air gaps in the case.

[soenke]

Also try to cover the ethernet ports with some metal and check if that makes a difference.

[zeallin]

Hello soenke,

Thank you very much for your suggestion, we will follow it and try.

Best regards,
Zeal