Switching frequency of SMPS (SSFD)

Dithering the switching frequency of a SMPS (SSFD) introduces sideband energy that by the conservation of energy must result in a reduction in harmonic-peaks. There are two important characteristics of a frequency-modulated waveform bandwidth rule.

The first is that the total power of a signal is unaffected by the frequency modulation, i.e. SSFD does not reduce noise energy. The second is that 98% of the total power of a frequency modulated signal is contained inside the bandwidth 2(Δf + fm).

Hence the energy of the fundamental component of a trapezoidal waveform is spread into a band given by: BT = 2(Δf + fm) and subsequent harmonics are spread into an increased bandwidth BTn = 2fm(1+n × Δf/fm) where n, is the harmonic order, fm is the modulating frequency and Δf is the depth of modulation. As the power spectral density decreases the narrowband noise of a switching regulator becomes broadband noise increasing the noise floor. This is not cheating the quasi-peak detector as previously suggested, it’s just spreading out the noise energy to reduce the probability of inference. In fact a quasi-peak detector responds to noise signals in a linear fashion - high-amplitude, low-repetition-rate signals can produce an output of the same spectral magnitude as a low-amplitude, high-repetition-rate signals, so I say the detector is inherently deficient in assessing EMI. If you’re not keen on frequency modulation try phase modulation, it has the same effect.