Failing radiated emmisions at 292MHz

[Jester]

This board is battery powered and has numerous "noisy" components:

• 2MHz Buck regulator
• ARM processor
• Bluetooth
• NFC

After a bit of tweaking it now passes everywhere except 292MHz, however it fails by 15dB!

Thoughts on adding a 100pF, 3nH series LC circuit across the battery leads?

 

[mtwieg]

You at least need to determine which of those systems is radiating the 292MHz. Such a high frequency is not characteristic of switching power supplies...

 

[treez]

all your current loops are small in area as possible?

 

[Jester]

You at least need to determine which of those systems is radiating the 292MHz. Such a high frequency is not characteristic of switching power supplies...

I tried disabling the 7.2 to 5V buck regulator and powered the board with an external supply and the 292 MHz noise dropped significantly, I'm not sure what to make of this?

 

[Warpspeed]

You need a spectrum analyser and a very small "sniffer" loop to carefully go over the entire board and find out which part of the board is radiating the 292 Mhz energy.

There may not actually be anything obvious running at 292 Mhz, it could be some unsuspecting resonant part of the layout that is being shocked into ringing.
(an unintentional frequency multiplier).
Either some damping or detuning of the offending resonant structure should make it go away.

I once had something similar, and it was rectifier reverse recovery spikes exciting an unfortunately long PCB track. A small capacitor and resistor in series placed directly across the diode damped out the ringing completely without any adverse effects.

 

[FvM]

Bad layout and insufficient on-board filtering is the most likely reason for spreading fast switching signals and possibly radiating it.

I don't think we'll get far without discussing circuit and layout details.

Regarding original question, you should be at least able to determine the interference source. Is it clear that the buck converter is the exclusive interference source?

Multi 100 MHz interference sounds like schottky diode.

 

[mtwieg]

Seconding the sniffer loop idea. Typical FETs don't switch fast enough to produce such high frequency interference. Diode recovery is plausible, but only if the FETs are fast enough to force that recovery.

A picture of the board would help a lot. Or at least some part numbers for the suspicious components.