UCC28063A: UCC28063A

Hello David,

Thank you for your interest in the UCC28063A interleaving-PFC controller.

The most likely cause of the "notches" in the input current waveform (blue) that you show is that the "Cin" value is too big for the power level being drawn at the PFC output.  Cin is the input capacitor located immediately after the diode bridge.  Whenever line current is zero around the line crossings, it means that the Cin is holding up the internal voltage higher than the line voltage, so no current gets pulled from the line.   A smaller value of Cin will reduce this.  

Choose a value for Cin to obtain an acceptable waveform at full load.  Since Cin is a fixed value optimized at full load, be aware that such notches in the current are likely to reappear at lighter loads, for the same reason.

Note, however, that smaller Cin may also affect the differential-mode conducted-EMI noise level.  The switching-frequency ripple voltage across Cin will be higher with a smaller value, so be sure to recheck your EMI-filter performance after reducing Cin. 

Regards,
Ulrich 

Hello David,

I'm sorry that the value of Cin did not improve things.  I did not recognize it earlier, but now I see that the switching waveform (yellow trace) is indicative of a semi-bridgeless configuration, rather than the usual interleaved phases.  Cin does not act the same way in this case.

So looking further, I see that the first notch occurring in the first negative half-cycle (blue) corresponds to a brief lack of switching after the zero crossing (yellow).  Most of the other notches are in the positive halves which do not have the corresponding switching waveform shown.

But since it appears (in yellow) that switching is stopped and then suddenly starts, this can be caused by a few kinds of faults.
Since the notches appear on both half-cycles (blue), I guess that it is not a Phase-B-only issue. Therefore, it is a fault that will stop both Phase A and Phase B switching. Since it does not take long to recover, I assume that it is not one of the lathed-faults that discharge COMP to 20mV to restart. Therefore, it must be either overcurrent (OC) or high overvoltage (HIGH_OV). Both of these faults can shut off switching to both phases, and recovery to switching again quickly.  (See UCC28063A datasheet block diagram page 14, and Figure 33 page 26.)

Since the negative notch shows no switching I’d expect that the positive notch to show current spikes if the fault were an overcurrent. Since I don’t see any OC in the yellow waveform, I suspect that your output voltage is triggering the HIGH_OV fault. Please check your system operation to see if that is the case.

OV can be caused by having too fast of voltage loop response to transient load steps and/or insufficient output capacitance. This is especially true if your load has periodic steps that repeat close to the line frequency or a low harmonic of the line frequency.

Regards,
Ulrich