Re: MOSFET driver killed mysteriously and need help

On Apr 19, 5:11 am, e c kern <eck...@xxxxxxxxx> wrote:
1. Connecting the input manually to vcc actually should not work. The
driver you describe is a bootstrap-only type (as opposed to hybrid
bootstrap+chargepump) and so the output line must periodically go to
ground to recharge the bootstrap capacitor.
yes I know. and that's why I first played around with low side output,
then pump in a low frequency square wave to test high side with
multimeter. Because of low frequency I used bootstrap cap 100uF. At
12V supply I checked HG output voltage is 20V when is high justfying
bootstrap was working, but then somehow it failed.

2. Are you just connecting HIN and /LIN? Many IRF datasheets for
single-input drivers discuss having specific circuitry to prevent
shoot-through. You should install a protective resistor on the rail
and use an oscilloscope to see if there are current spikes during
switching. If there are, you must redesign your drive circuit to have
a deadband where both mosfets are open between high and low.
I thought the internal anti shoot through circuit which generates
500uS dead time is a bullet proof. Even if cross conduction occurs the
one who dies should be MOSFET, not driver.

3. High-side drivers usually have a Vbs threshold to prevent
destroying mosfets by applying an insufficient drive voltage. To some
extent this is more of a sanity check -- depending on your mosfets and
load, this threshold may not be high enough to fully protect your
circuit. Verify that Vbs never goes below the (practical) "full-on"
Vds of your mosfet.

-chris.
If Vbs goes low, at most the chip shut down and deactivate MOSFET, but
this should not kill driver itself.

Thanks for your helping. This is the circuit I test with breadboard:
http://www.geocities.com/w2kwong/MOSFET.jpg

.