I chose to use the L6202 H-bridge IC because it has built in dead time as well a thermal shutdown protection. However I've already broken multiple of these from what appears to be current spikes when switching motor directions.
During operation the DC motor will be powered with 48V, and draws around 200-400 mA when moving. Right now we are only giving it full speed either direction, no PWM.
I've set up the circuit as basic as I can following the datasheet reccomendations.
- Vs: connected to 48V
- SENSE: connected to ground
- Bootstrapping capacitors: 47nF/50V
- VREF: 10 nF capacitor to ground
- IN1/IN2: 3.3V high / 0V low
- EN: connected direct to 3.3V
I powered the motors with a power supply (24V, 1A current limit) for testing and it worked fine. The chip has a built in 40ns dead time which I thought should be more than enough considering the switching is happening on the scale of nanoseconds. However, I started adding a delay because I saw current spikes on the power supply exceeding 1A (power supply was entering constant current mode). That made me realize the 40ns dead time was probably not enough. I added a 50ms delay between one direction and the other. When we tried running the motors in operation with 48V power supply (unsure of the current limit), a 500ms delay between switching directions worked, but with a 250 ms delay, the chip was destroyed.
I also thought maybe we needed to add freewheeling diodes to help current flow when we make sudden changes to the motor speed, but it seems to me that this is already built into the chip as part of the MOSFET structure.
When the chip is destroyed it appears to short the two outputs, Vs and gronud all together which is very strange considering that the chip has built in thermal protection that is supposed to shut it down at 150C junction temperature. When the chip is cooled down again Vs and ground are still shorted together along with the outputs and the chip is no longer usable.
Am I doing something wrong and/or is there a way to mitigate this problem and continue using this chip? 250 ms seems like a ridiculously long time to wait for a motor to switch direction.