3
\$\begingroup\$

I am working on a circuit with a AC to DC power supply module and wireless MCU module. The MCU operates at 240MHz and on the power input to the module there are 2 x 0.1uF capacitors and a 22uF capacitor used for bypass. These have been found through trial and error to most limit the noise passed on to the rails. To further reduce noise, a ferrite bead is located on the power side of these capacitors and has worked to reduce the propagated MCU clock noise to acceptable levels.

simplified representation of circuit

The remaining problem that I am hitting is that there is a strong "travelling" transient spike (as shown in the spectrum plot below) that starts at about 1MHz and makes its way to about 25MHz before disappearing into the noise. At any given time there is only one spike a specific frequency and it travels from low to higher frequency while reducing in amplitude. The travel time of this transient is about 2 seconds. This transient is strongest near the MCU module but is getting onto power rail (Test Point) and out through the AC/DC module to the power cord.

enter image description here

I have attempted to add more capacitance (1uF, 10uF, 22uF, 47uF) near the MCU module but have seen no measurable effect on this travelling transient spike by doing so. Ignoring the presence of this spike, the noise level is near the floor as seen by the spectrum analyzer. The presence of the ferrite bead mentioned previously has no effect on the amplitude of these lower frequency spikes as its peak impedance is in the 200-400MHz range. If the MCU module is replaced with low value resistor, the spikes are not present.

Based on this, I have a few questions:

  1. For a travelling transient like this, can anyone suggest how this would affected the conducted emissions values versus a fixed frequency spike?
  2. What would the quasi peak of this sort of transient spike look like?
  3. Are there any suggestions on what might be possible to reduce the amplitude of these spikes given that various additional capacitances on the MCU module power input didn't seem to work? I will be trying a ferrite core on the power cord, as well a low frequency ferrite bead on the input to the MCU module and have already tried an inductor on the input to the AC/DC module.

Thanks.

\$\endgroup\$
  • 1
    \$\begingroup\$ Could you provide a simple schematic so we can visualize easier what is going on? \$\endgroup\$ – C K Aug 10 '18 at 18:51
  • 1
    \$\begingroup\$ Hopefully the updates make the question more clear. The amount of time for the spike to travel is based solely on watching the spectrum analyzer output with the transient seeming to "travel" from lower to higher frequency. \$\endgroup\$ – Jason K. Aug 10 '18 at 19:54
  • 1
    \$\begingroup\$ It is the output of a spectrum analyzer as stated in the preceding paragraph. It is only there to demonstrate how the transient spikes are "travelling" \$\endgroup\$ – Jason K. Aug 10 '18 at 20:19
  • 2
    \$\begingroup\$ How are you observing these "travelling peaks"? Is it possible that they're a measurement artifact of some sort? \$\endgroup\$ – duskwuff Aug 10 '18 at 20:55
  • 1
    \$\begingroup\$ It may be synchronous broadband spikes. Change sweep speed to sync with line frequency. \$\endgroup\$ – Sunnyskyguy EE75 Aug 10 '18 at 21:04
1
\$\begingroup\$

Normally "traveling" peaks do not contribute meaningfully to quasi-peak or average measurements, which are (usually) what counts when it comes to EMC regulatory standards.

Frequency dithering can induce traveling peaks - this is a standard EMC "cheat" for power supplies and is a feature of some PWM controllers to specifically beat the QP and average emissions limits.

A QP detector is sort of like a peak detector followed by a lossy integrator. Spurious peaks (ones that aren't always there) tend to get filtered out. Average detectors are even less sensitive to spurious peaks. These travelling peaks only show up in the QP and average measurement frequency windows for short periods of time, hence their minimal impact.

\$\endgroup\$
  • \$\begingroup\$ I was wondering if the module might be doing this shifting intentionally but just wasn't sure what the effect might be. Are you saying that the EMI measurement equipment in a test lab is typically going to be an averaging detector of some sort? \$\endgroup\$ – Jason K. Aug 10 '18 at 20:53
  • \$\begingroup\$ Yes. Both QP and average emission limits for IEC EN 61000 EMC testing are forms of averaging. There is no formal 'peak' regulatory limit that I'm aware of. \$\endgroup\$ – Adam Lawrence Aug 13 '18 at 13:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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