UCC2810: Sharpness of harmonic current waveform

Hello Conor-san,

I agree that the probable cause of higher THDi in sample X7 board is due to the spike on the rising edge of the AC input current. 
Such a spike may be caused by a saturated amplifier output on the ICOMP pin of the UCC28180 PFC controller. 

The UCC28180 uses leading-edge PWM, where the turn-on edge of the GATE signal occurs when the internal M2 ramp crosses the ICOMP voltage. 
At the AC-line zero-crossings, especially at light loads, the ICOMP signal is near zero V and duty-cycle is maximum.   
As AC voltage rises, ICOMP should also rise and reduce duty cycle to keep the input current tracking the voltage.
If ICOMP does not rise as fast as the AC voltage rises, the high duty cycle will begin to generate excess input current.
When ICOMP jumps up to the normal level, the duty cycle becomes normal and current drops to normal, so there is a brief spike which generates higher-order harmonics.  

I suggest to compare the ICOMP signals of X7 and X1 boards to verify a difference coincident with the spike on X7 input.  
If verified, then the cause of the ICOMP difference must be found.

On these boards, a 10nF capacitor (C22) is used to filter ICOMP.  This is an unusually high value (unusual to me, anyway; I haven't seen any previous applications higher than 4700pF), and it is possible that the actual value of C22 is significantly different between the boards.  
High value capacitance in a small case-size results in significant DC-Bias effect in MLCCs.  Although the ICOMP voltage should be very low at the zero-crossing, there may be enough difference in capacitance even at low voltage to result in a difference in ICOMP response.  
Changing C22 on X7 board may show a difference in THDi.  

Alternatively, swapping IC2 between boards X1 and X7 can identify if the THDi issue follows the controller IC or follows the board. 

Output capacitors C19 and C21 add up to 660uF which indicates a fairly high-power PFC; probably over 750W.
This application must have quite stringent THDi requirements for a "FAIL" result at a 75W load.  

Regards,
Ulrich

Hello Conor, 

Thank you for providing the result of your testing. 

Conor said:

We performed an A-B-A swap. I replaced the IC on the defective board X7 with the chip on the X1, but the same problem occurred.

Based on this result, it is clear that the problem remains on the X7 board regardless of whether the original X7 controller or the X1 controller is installed.
Therefore, the "FAIL" result for THDi is due to something different in the power train of the X7 board, not in the controller IC. 
I suggest that you can compare the switching waveforms near the zero-crossings between the X7 and X1 boards to search for differences that may point to the cause of the spikes on X7.  If possible, swapping power components (one at a time) between boards may indicate which part makes the spike move from X7 to X1.  Swapping boost inductor, for example, and/or input capacitor, and other components one by one.

For Q1: The UCC28180 Excel calculator tool (https://www.ti.com/tool/download/SLUC506 ) will calculate the correct value for Cicomp, based on datasheet equation (100) and precursor results.  

For Q2:  The PFC converter will operate stably at all loads, down to 0W.  The PF becomes poor at light loads for 2 reasons: (a) input current phase-shift from X-caps in the line filter and (b) increased THDi because of DCM in the inductor.   At light load, the displacement current (90-degree phase shift) in X-caps forms a major portion of the total input current. In the inductor, light loads tend to operate in DCM, but that introduces harmonics.  Low distortion comes from operating in CCM.   But at light loads, offsets and other factors also increase THDi.

Both of these factors happen in any PFC, so PF always falls at light load, regardless of controller. 

Regards,
Ulrich 

Hello Conor, 

By stable converter operation, I mean that the PFC output voltage remains in regulation.  Distortion of current at light loads is an expected situation, especially when the controller is designed for CCM operation (as the UCC28180 is) but the inductor current is in DCM.  But distorted current is not unstable current.  

If possible, please post some waveforms that show what you mean by unstable current waveform. 

With your new information about the IC swap, I agree that it is more likely that the IC originally from X7 causes the higher distortion with the leading-edge spike.  I can only guess at the reason for this.  It may be that the amplifier at the ICOMP pin has less current drive capability than normal and cannot respond to the ISENSE signal as quickly as other ICs can.  If ICOMP voltage is slow to rise, then the duty cycle will be wider than it should be leading to higher inductor current, until the ICOMP recovers and delivers the correct voltage for the operating conditions.

I don't know what would cause such a situation.  If this is a one-time occurrence, then perhaps the IC was damaged somehow, by ESD or other overstress.
If this happens with multiple controllers, then there is a pattern and the cause can be found.  There is always variability in device parameters. Maybe the 10nF or 100nF cap on ICOMP can overload ICs with the lower levels of ICOMP output current.  Changing Cicomp to a lower value may solve this. 

Regards,
Ulrich