2
\$\begingroup\$

An FCC unintentional radiators test is failing at 550MHz, after probing with a near field probe the source was located to these areas of the PCB (shown below in black circles). The RMII is running at 25MHz, so that would be the 20th harmonic. Ther are three sources of possible noise. Rise times from digital signaling, or conducted emissions from other areas of the processor. Or (one that is less likely) the PHY (KSZ8051RNLU)itself. The other option is there is some kind of conducted emissions from the processor.

The series resistors for the RMII have also been changed from 22Ω to 33Ω, with no reduction in the near field emissions. If I increase the impedance and the near field emissions do not change, then rise times should be affected, due to the RC time constant being increased. Is that a true statement?

It was my belief that after the 10th harmonic of rise times, radiation would pretty much be dead due to roll off is that true?

I also changed the power with a resistor instead of a ferrite as suggested.

EDIT: Interestingly enough we have matched the drive strength and gone from 22Ω to 50Ω with no reduction in the near field radiation.

enter image description here

enter image description here

enter image description here

\$\endgroup\$
16
  • \$\begingroup\$ Have you got a continual ground plane under the resistors? \$\endgroup\$
    – colintd
    Commented Nov 16 at 8:16
  • \$\begingroup\$ Are the RMII series resistors on PHY side for both inputs and outputs - should the resistors always be on outputa? Also the schematic appears to have same net names on both sides of resistors, so are the resistors bypassed so changing them has no effect? \$\endgroup\$
    – Justme
    Commented Nov 16 at 10:05
  • \$\begingroup\$ @Justme Yes, inputs and outputs. The name is actually different so they aren't being bypassed. There is another name in front that I had to remove so it wasn't identifying the CPU we are using. \$\endgroup\$
    – Voltage Spike ♦
    Commented Nov 16 at 15:03
  • \$\begingroup\$ @VoltageSpike OK, resistors are inline, but are they physically located right at the transmit pins on each chip, instead of just placed near one chip, so the resistors at chip inputs are not really doing impedance matching? \$\endgroup\$
    – Justme
    Commented Nov 16 at 15:46
  • \$\begingroup\$ @Justme Do I need to impedance match for 25MHz? Also my understanding was that the resistors only need to be on one side of the interface to slow down the rise times. \$\endgroup\$
    – Voltage Spike ♦
    Commented Nov 18 at 16:49

1 Answer 1

Reset to default
1
\$\begingroup\$

The drive strength of the RMII pins on the CPU and the series resistors were changed to 50Ω. Even changing just the CPU drive strength to 50Ω to reduce rise times cut the near field emissions and also passed FCC unintentional radiators.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.