Post Go back to editing

When is a snubber needed on a buck regulator?

Category: Hardware
Product Number: LTC3313

All switches ring. If the ringing is severe enough, it justifies an RC snubber to reduce it. My question is, how much ringing is "too much"?

I'm looking at a 5V in, 1.2V @ 8A out buck regulator such as an LTC3313 or similar. If I put placeholders on the layout for an RC snubber (to GND) on the SW terminal, and after the PWB is assembled, measure the ringing to decide if the snubber should be populated. I have found lots of good guidance on how to select the correct R and C values, but I can't find any guidance on how much ringing is "too much" and when the snubber should be populated. I realize this is a trade-off question, not a hard and fast rule, but what is the general consensus of when this snubber should be populated? More than 25% overshoot? More than 50%? More than 100%? Obviously at some point, it can approach the breakdown voltage of the switch transistor, but assuming we are well below that, when is it too small to care about?

I know the actual amount of ringing depends on the PWB layout and the parasitics, which are hard to predict. Hence the need to measure the actual board. And I know that a snubber will always reduce (but not completely stop) the ringing. So at what level is it worth populating these components? 

Thanks.

  • Hi,
    as you can see in all the demo boards, buck regulators don't have a snubber, because they don't need it.
    When the power switches are inside the IC, there is neglectible ringing. When there are external power switches and you obey the layout guidelines, there is no need for snubber either.
    As you know already, a snubber would be needed if maximum voltage is violated because of ringing OR if you have an EMI-problem.
    There is always the first strategy to have good (best) layout and therefore need no snubber.
    PS:
    With step-up converters (with high in-out ratio) / isolated converters it is a different case. Here you often / always need a snubber.

  • Thank you for the reply.  I was hoping for something a little more quantified, but I guess I can say as long as the over/undershoot is less than 90% of the VDS limit, we are OK.