LNK304 - Feedback capacitor voltage rating

5 posts / 0 new

In a design using the LNK304 in a traditional none-isolated buck converter, we've seen the feedback capacitors fail in some cases.

We have 12V output and use a standard X6S 0805 16V ceramic capacitor. The voltage will indeed get fairly close to the rating at switching, but should still be within the voltage-rating.

I can't seem to find much information on recommended voltage-rating of the feedback capacitor.
So long story short, what's the recommended voltage-rating for the feedback capacitors?

Adding to the above, when measuring the voltage on Cfb, this is the ripple seen when switching. Nothing that seems to indicate a voltage-spike that would cause a component failure as seen.
Increasing capacitor voltage rating might help, but we would really like to know what's causing the failures seen, rather than just trying higher voltage-rated cap, hoping for the best.

DS4_QuickPrint19.png28.71 KB

Thank you Dennis for the question and clarification.

The voltage rating of the feedback capacitor is mention in Section 9 of AN70 as ≥1.25*Vo.

The capacitor you selected is in the specified range and it should be OK. There is no indication of existence of a similar problem before.

Two factors have the potential to cause the issue, separate or in combination.

- Capacitors voltage derating with temperature
Derating usually starts at 60 Deg.C if no different specified in the datasheet.

- Transient input and load condition.
The steady-state ripple from your picture obviously cannot be responsible for causing 16V capacitor to fail. But those ripples will be very different in transient conditions for input and load. There, the design point and CCM/DCM operation gets involved.

If you manage to find the worse condition and look inside the transient pulses you most likely will see the voltage going over the limit.
Increasing the voltage rating of the capacitor will help, but it may help to try another diode as well. In CCM the capacitor will “see” the whole inductor voltage for the recovery time of the diode. The inductor voltage is Vin-Vo.

Best Regards,

Hi PI-Yoda
Thank you very much for the information.

I will do some further thermal testing and try to provoke higher voltage ripple/spikes, when doing transient response etc.. It just might be the temperature derating and a narrow voltage spike that has caused the problem.

The diode used is an ES1J, which should have very low recovery time already.

Great, Dennis,

You are correct, this diode is pretty fast.

To separate both issues during the investigation a zener 1 to 2V higher than Vo attached to the output will reduce the output transient during the starting process.

Best Regards,