From: John Larkin
Subject: Re: Synchronous buck switch node waveform question
Date: Mon, 23 Sep 2002 20:14:15 -0700
Organization: Posted via Supernews, http://www.supernews.com
X-Newsreader: Forte Agent 1.91/32.564
On Tue, 24 Sep 2002 02:57:01 GMT, wrote:
>I'm looking at a buck regulator switch node waveforms and see something I''d
>like comments on.
>First, let me explain that I'm controlling the delays between drive signals
>to try to minimize body diode conduction in the synchronous rectifier (SR).
>As I reduce delay between bottom switch (SR) turning OFF, and top switch
>turning ON, I see a negative voltage of approx. -0.6V, which appears to be
>the body diode voltage. After 30-50ns, the voltage heads back to ground or
>a volt or two above ground, and hovers there for 20-50ns, with the longer
>times coincident with heavier load. Then, the voltage snaps up in a few ns.
>Is this hover period the diode recovery time? Why does it not stay at the
>negative diode voltage until its cleared? Any thoughts on this switch node
>rising would be appreciated.
Sounds maybe like the body diode in the lower fet is acting like a
step-recovery diode, ie storing charge when foreward biased and
reverse-conducting until the charge is swept away. This is very bad
for efficiency. One fix is to use a bottom fet with a fast-recovery
substrate diode, or parallel it with an external schottky.
Possible gotcha: some power schottkies themselves have reverse
recovery, due to P-N guard rings or something.