UCC28070: What's the suggestion dither rate for UCC28070

Hello Shaquille,

That's a good question that is not fully addressed in the UCC28070 datasheet (DS). The DS dithering discussion on page 15 uses 1kHz as an example for the dither rate on a 100kHz nominal switching frequency. However the dither rate can be much higher, but not too high.

The purpose is to shift the switching frequency (fSW) above and below the nominal frequency quickly enough to reduce the amount of time that fSW dwells within the 9kHz measurement bandwidth of an EMI receiver. The fSW is intended to shift out of the measurement window long enough to reduce the peak EMI noise measured. It is not eliminated, but it can be reduced by some dBuV.

I think setting the dither rate (fDR) to 10% or less of the nominal fSW is a reasonable target. Using 100kHz as an example, fDR should be </= 10kHz. It shouldn't be too slow either, or fSW will remain in the 9kHz measurement window so long that the reduction of EMI peak noise is defeated. Note: the recommendation of fDR </= 10% of fSW is my opinion, not a scientific fact.

One should review the conducted-EMI data and adjust fDR up and down (as well as the dither magnitude fDM) to see if different settings can improve the results. Also note: the dithering affects the harmonics of fSW in direct proportion, so the peak-to-peak magnitude frequency swings of the harmonics are wider. (If fSW swings 10kHz, its 2nd harmonic swings 20kHz, etc...) When the nominal fSW is < 150kHz, the harmonics of fSW are measured but the fundamental is not measured, so less dither magnitude may be used. When fSW is >150kHz, the noise of the fundamental frequency is also measured so more dither magnitude may be needed.

I hope this clarifies the dithering function.

Regards,
Ulrich