UCC23513: High dv/dt causes SiC MOSFET to turn on incorrectly

Hi Allen,

The voltage spike is proportional to both the injected Miller current (I=C d(Vds)/dt) and the impedance between the gate and source (Lgate +Rgate+Rol+Rg). You should try to reduce both of these.

One trick to reduce the dV/dt across the low-side switch is to add a little distance between the phase and the low-side drain. This will add a tiny amount of inductance, which will absorb some of the dV/dt from the high-side closing, and divide down the dV/dt across the Miller Cdg. Another possibility is to add a snubber to the High Voltage rail. This can be tuned to cause a high frequency transient droop when the switch turns on, which will lower the dV/dt a small amount also.

On the gate driver side, I do not see the UCC23513 in your schematic. Based on your labels, I surmise that you have 39 ohms of turn-off resistance. Also, you have 2.2 ohms in series with your external Cgs. This is a lot of resistance.

If you have a ground plane, you likely have very low turn-off ringing. You can probably use 1-2 ohms of turn-off resistance and not generate significant undershoot. For R102 and R100, You should probably use 0 ohms, since these resistors prevent the capacitors from instantaneously absorbing the miller injection, and integrating the current down to a lower voltage.

There is a version of this device called the UCC23514, which has a Miller clamp pin added. This lets you use large turn off resistors, and then the 0 ohm Miller clamp pin closes in parallel with these resistors to provide a low impedance bypass for the injected Miller current.

One other possibility is if you do not need 55A in this application, you can use a cheaper FET with higher Rdson and lower Cgd. The larger, more expensive FETs have higher capacitance and have more Miller injection than right-sized FETs.

Best regards,

Sean