H-Bridges and MOSFET / IGBT Gate Drive

by Bryan A. Thompson

Last Updated 01/12/2007

 

High Side Gate Drive

At anything over 20-30V, this sucks.  I tried dedicated FET Driver ICs (IR2184) and learned that this IC doesn't work at PWM duty cycles under about 10% or over about 90%.  This made it impractical for BLDC control. 

I developed a driver circuit based on the HCNW3120 Optocoupler/FET driver IC and an Isolated DC-DC converter by C&D Technologies. This proved very effective up to 100KHz and may be good up to 1MHz with snubbers around the FET gate. I did a board (FET Driver board v3.6) for 4 of these drive ckts and four TO-247AC FETs/IGBTs.  I forgot to include Extra Flyback diodes and a current measuring shunt.  International Rectifier sells something called DirectFETs (or DirecFETs) that have the high side driver built in, allowing TTL control.  IPMs also have FET drivers built-in, but this is the same technology as the IR2184 and not good for low or high PWM duty cycles. 

Low Side Driver

The Low Side FETs are easy to drive since they're ground-referenced. Note: May not be ground referenced if there is a shunt resistor between the source of the bottom FETs and the HV (-) terminal. This may not matter if not much voltage is dropped across the shunt resistor even at max current. My philosophy for low side drive is to use the same drive circuitry as for the high side drive, so that timing issues will be easier to calculate. May want to reconsider this, as the drive circuit that I have designed has issues related to very long propagation delay and relatively slow switching. I'm thinking of changing this to a TC4421/22 FET driver from Microchip due to higher drive current, faster switching, lower propagation delay and the fact that the two low side driver ckts won't require the Isolated power supply at a cost savings of $16-28.

Flyback Diodes

If the load is slightly inductive, the flyback diodes in the MOSFET body is not enough. Use Ultrafast diodes to protect the MOSFETs from voltage spikes at turn-on or turn-off. No idea how much current to rate them at? Some suggestion that MUR120 or STM BYW99W-200.  These are just paralleled with the body diodes in the MOSFETs (4 per H-bridge).



FET Switching

I tried switching both top and bottom gates by placing the inputs in parallel. Everything I've read says that the dissipation in the internal body diodes will be much higher if this switchmode is used. Instead try turning on the top FET and leaving it on, and just applying PWM to the bottom FETs. I think this topic is called Synchronous Rectification.



FET Heat Dissipation

They don't seem to dissipate much heat in proper switchmode with the optcoupler drive method. I used an IRFP264NPBF FET, which has 60mOhm On resistance. At 10A, this is 6W dissipation, confirmed by experimentation. I attached it to about a 2lb heatsink and it stayed cold. At 44A (its max), this is 116W and I did not test this due to lack of a sufficient power supply.



FET Failures

They almost always seemed to fail immediately if they didn't have a TVS (or back-to-back Zeners) between the gate and source to protect it from overvoltage.   When MOSFETs fail, they always seemed to fail shorted between Drain and Source (On), which has the potential to power the load at full power at 100% duty cycle, or to cause a shoot-through situation shorting the HV power supply to ground.