# Eliminate 2.4GHz (Wifi) signal on power lines

We a designing a small pcb where a temperature sensor is placed close to a add-on wifi module.

The sensor has four traces connected to it power and i2c (3.3V, gnd, SDA, SCL).

The sensor needs to have some distance from other components (MCU,PSU, ect.) due to heat transfer and the pcb power planes are not routed below the sensor.

It is therefore nessacery the use relative long traces from the MCU and PSU to the sensor (about 30mm).

To minimize the 2.4GHz signal coupling onto the power traces going to the sensor, I have added four capacitors close to the sensor (22pF, 100pF, 1nF, 100nF). I was thinking about adding a small chip inductor, to make the filter even better.

But is this the way to go?

Or are there more efficient ways of eliminating the 2.4GHz signal?

-
Out of curiosity, how badly does this noise affect measurements? How close and powerful is the WiFi module? –  tyblu Mar 1 '12 at 5:10
Year, this is where it gets really funny. Because we are in need of time, I'm just assuming that this would cause a problem. I have not measured that it actually is a real problem. But I figured that since this is a prototype, but it should be working, I would design for worst-case, and then trim the final version. –  JakobJ Mar 2 '12 at 7:02

22pF at 2.4GHz is 3$\Omega$, which should be sufficiently low to suppress any coupling. Use an SMD, not a wired part. An inductor will improve the behavior, but isn't required. If you want to use one you may consider the BLM15G series chip ferrite bead from Murata (warning: full catalog, so long document). For instance the BLM15GA750SN1 has an impedance of 2000$\Omega$ at 2.5GHz.

I don't think the 100pF cap will do much (not adding to the 22pF for the HF suppression, nor power supply decoupling), so I'd leave that out. Since the sensor will not require much current a 10nF decoupling cap instead of the 100nF will do.

If you want to avoid thermal coupling between the sensor and the rest of the board you may consider milling the PCB around the sensor, so that it's on a peninsula, with only a connection to allow wiring.

edit
One would expect that the 100pF capacitor has an even lower impedance than the 22pF. At high frequencies like 2.4GHz ESL (Equivalent Series Inductance) plays an important role, and the larger the capacitor the larger the ESL. In practice the impedance of the 100pF at 2.4GHz often will be more than 1/5 of the impedance of the 22pF.

-
+1: ferrite beads are the answer –  Jason S Feb 29 '12 at 19:30
Thank you very much. :-) –  JakobJ Mar 1 '12 at 13:47