Electrical Overstress (EOS) Damage in Power Supply Component

[DoctorProf]

eos damage

Switching power supplies are bad for creating switching tranients. These transients can generate overstress events in the control/drive circuits. EOS is typically one of the highest causes for field returns. I've talked to several companies that state EOS accounts for about 20 to 30% of the total returns from the field.

I've been tasked at looking how to make our parts more robust to EOS events. I have evaluated several of the highest offenders and found the most common aspect of EOS is related to the high current switchin node. Pins driving this node or monitoring the voltage on this node are most susceptable to EOS conditions.

I would like to get other peoples opinions on what is the source of EOS damage. It has been my experience that most of the EOS damage is related to a poor board layout or poor selection of components connected to the controller/driver IC in power supplies. The added capacitance/inductance produces switching noise that causes breakdown in conponents and ends up destroying components.

Thanks,

Dr.Prof

 

[Btrend]

eos ic damage

1. shoot through of high side & low side MOS will burn down both MOS
2. dead time too large will burn down low side MOS
3. if high side MOS and low side MOS have different turn on and turn off delays, the above two points should be checked carefully.

 

[leo_o2]

electrical over stress

What's kind of switching power supply? What's applications?
I guess the circuit with more power output will be easier to be damaged.

 

[DoctorProf]

eos ic

Thanks for the replies but I'm not as concerned about he MOSFETS. I'm more concerned about the controller circuits/MOSFET drivers getting damaged by the transients generated during switching.

The reference design does not cause failures but not all customers use the reference design for thier supply. As an example, the power controller for an Intel microprocessor uses a 4 phase PWM controller driving 4 sets of MOSFETS. The output is programmed to be between 1.3 and 2.5 volts The chip is powered by the 5 volt ATX supply. The pin that monitors the uP supply voltage gets EOS damaged. It should not see more than the processor core voltage but yet it gets blown up indicating that it the voltage on the inputs exceed 5 volts or goes well below ground.

This is the type of problem I'm trying to track down.