UCC28070:Power management forum

Hello Riki. 

Are you able to discuss this issue in English?  If not, I may have to re-assign this to one of my counterparts in Asia.  Please let me know.

If so, please provide some schematics and load conditions.

Regards,

Ray

symptoms

・In products with the UCC28070, the output voltage of the PFC became unstable. (The PFC SWFET repeats the switching operation in a period of about 2 seconds (waveform 1)

　Its operation is divided into A-phase and B phases, and the way the pulse is removed is different in the two phases, and one phase is in a stopped state (waveform 2, waveform 3))

　This symptom improves when the IC (UCC28070) is changed. I would like you to tell us the cause and solution to which these symptoms occur.

condition

・Input voltage: 200V

・output voltage of the PFC: 390V

・No load condition (symptoms disappear when a few% load is pulled)

Waveforms and schematics.pdf

Hi Riki-San,

The 0.4Hz pulse grouping behavior is very unusual.

Normal no-load behavior should be that both phases A & B should have relatively random, narrow switching cycles scattered over the line cycle.  Maybe clustered more toward the peaks or the zero-crossings, but still relatively scattered.

I can't tell if by replacing the part you mean that the issue was completely solved or just better or different.  If it is completely corrected, then perhaps you had a bad part or solder issue. 

If the issue is still present but occurs less often (or is somehow less severe), then the difference in the parts may be just a matter of different amplifier offsets causing one part to have the issue more/less often.  In this case, one suggestion I have is to check your VSENSE capacitor (actual value).  This type of pulsing may occur if you have a very large value in that location.  In waveform 1, with ~2sec dead time between switching groups, the situation may be that the VAO output is also oscillating at this frequency.  When VAO goes above the threshold to allow switching but VSENSE is too slow to react, it overdrives PFC Vout until VSENSE exceeds the VREF.  VAO then drops down below the threshold and suppresses all switching until the bleeder loading (from VSENSE and other small loads, if any) finally pulls it below VREF again.  It's just one idea and gives you an idea of more nodes to probe if the issue persists after checking Cvsense.

I hope this helps.

Ray

Mr. Ray.

Thank you for your answer.

I changed the Cvsense capacity to about 2nF, but there was no improvement in the symptoms.

This symptom occurs in about 1 in 200 units, and in all of them, changing the IC eliminates the symptom. Is this due to variations in IC characteristics? I would like to know.

Riki

Hi Riki-san,

do you mean during a lot of units for mass production, the failed rate is 1/200=0.5%? or you just test 200units, and found one unit?

if this phenomenon is captured only at no load condition,  because of offset in amp op on the CSA/CSB circuit inside the UCC28070 controller is different, it is normal that each phase controls each phase output pin(GDA or GDB) and GDA/GDB can not have the same pulse output .

once you increase the load, the influence of the offset decreases, so the phenomenon remove.

if possible, you can send a full schematic in the post also for me to review it.

thanks.

There is a 1 in 200 chance of this happening, and it has happened in about 4 units so far.

The problem is not that phase A and B do not pulse at the same time, but that they stop and operate in bursts with a period of about 2.5 seconds.

I will also attach the waveform when the IC was changed and the symptoms improved.

I can send you the schematic if it is private.

waveform④⑤.pdf

Hi Riki-san,

please make sure the vsense should have the same resistor divider with VINAC pin.

you can send the schematic to me through wenbin-huang@ti.com.