UCC28019A: UCC28019A

Hello Ashok, 

Thank you for your interest in the UCC28019A PFC controller. 

No, the negative clamp diode on the ISENSE pin is not recommended on UCC28019 and -019A because the PCL threshold may be as low as -1.15V (per the datasheet) and a diode in that location would clamp the voltage and defeat the PCL function and possibly the SOC threshold, too.  (In the UCC28180, the PCL threshold is much lower, so a P/N diode drop will not affect it.)

The UCC28019/A devices are made on an older semiconductor process than the UCC28180, and are less sensitive to negative voltages on the ISENSE pin. 

Regards,
Ulrich

Hi Ashok, 

No, that level of negative voltage will not cause failure of the controller, whether VCC is applied or not.  
This voltage comes from current being pulled out of the pin from the negative voltage across the Rsense resistor, through a series resistor probably at 220ohm.  

If you wish to reduce the negative magnitude at ISENSE, the current may be reduced by increasing the value of the series resistor to 1kR or so. 
There is a ~2uA bias current (for open-pin protection) which would add ~2mV to the ISENSE voltage, so increasing the series-R has little impact.
However, if you have a noise-filter capacitor from ISENSE to GND, that cap value must be reduced by the same factor that series-R is increased to keep the same filter time delay.  You don't want to increase ISENSE time delay.   

Regards,
Ulrich

Hello Ashok, 

For other viewers of this forum, I will reply here without violating any confidentiality of your design. 

The current-sense resistors definitely do need to be specifically rated to handle the worst-case inrush current at first start-up.
Based on the 68A peak inrush surge, peak power in a 25mR resistor = ~116W in the first surge.  Subsequent peaks are much lower, but the first one causes all the damage in the resistors.

Since the first lasts for about 4ms, I estimate that surge energy to be about 116W *0.004s = 0.4624J.  It is possible that the surge shown may not be the worst-case and the maximum energy may be higher.  So the current sense resistor should be selected for 0.5J or higher for long-term margin.

In addition, the I^2t product of the 16A fuse is listed as 1332A^2s (despite being a "fast-acting" fuse).  On the other hand, the sense resistor fails open at 68A^2*4ms/3 = 6.2A^2s  (1/3 taken due to roughly triangular shape of current peak).  In this case, Rsense makes a much "better" and faster fuse than the actual fuse does. 

When the inrush surge current is strong enough to blow open the sense resistor,  the only remaining path for the current that is already flowing would be through the GND pin of the UCC28019A and out of the ISENSE pin.  Inductance in the surge path will raise the voltage to keep the current flowing, if the line voltage itself does not already over-stress the UCC28019A.  Nothing can protect the UCC28019A in this case (not even the diode on ISENSE) and prevent it from failing.

But such a failure is a consequence of another component failure, not due to a flaw within the IC itself.  Massive surge damage to the controller can in turn damage the gate of the boost MOSFET and cause yet another follow-on failure (such as a shorted-MOSFET) to further increase the total current energy until even the PCB metal track may fuse open, all before the main fuse opens.   

The ultimate purpose of the input fuse is to prevent a fire from uncontrolled line current in the event of some other failure.  It is not intended to protect semiconductors from current overload because it simply is not fast enough.  Be sure that the fuse rating is suitable to the application power level and follow manufacturer guidelines and app-notes on fuse selection and derating.

Regards,
Ulrich