How to protect my mosfet h-bridge?

[chris414]

I've made an H-bridge to control the power to a DC motor (24V, drawing about 7amps) using PWM. My h-bridge uses the SFP30N06 mosfets driven by an IR2113 high/low side driver. The PWM signal is generated with a microprocessor.

While testing some code, the h-bridge mosfets blew. The only reason for the short that I can see is that in my code I accidently turned the both sides of the h-bridge on at the same time when changing directions - ie. something like this:

Code used to go in reverse:

reverse = 1;
forward = 0;

and then when changing directions the code executed is:

forward = 1;
reverse = 0;

In other words, for one tiny fraction of a second both the "reverse" and "forward" sides of the h-bridge were both on, causing a short circuit. The 15A fuse I had blew immediately, but so did my mosfets. If my fuse could not break the circuit quick enough to prevent damage, how am I supposed to protect against similar careless mistakes in the future?

 

[smanches]

There really is no easy way to prevent what happened. You would have to have current sensing for each mosfet that will shutdown the gate during an over current event. Some mosfet drivers have over current fault protection built in, and use either a current sense resistor, or current sensing mosfets.

 

[crutschow]

The old rule is "the transistor is always faster than the fuse". You could add a transistor type current limiter in series with the power to the bridge such as this **broken link removed** which would be fast enough. This, of course, will add a small drop to the voltage due to the series current-sense resistor (the current starts to limit at about a 0.6V drop).

 

[MikeMl]

I have a lab power supply which has independently adjustable output voltage and output current. I use it while debugging untested circuits/software. The output voltage can be set to say 12V, while the current limit trip point can be set to say 5A. If the circuit draws more than 5A, the supply goes into current-limiting mode, so it would put a few mV into a dead-short (~ 0 Power)

 

[crutschow]

I have a lab power supply which has independently adjustable output voltage and output current. I use it while debugging untested circuits/software. The output voltage can be set to say 12V, while the current limit trip point can be set to say 5A. If the circuit draws more than 5A, the supply goes into current-limiting mode, so it would put a few mV into a dead-short (~ 0 Power)

There could be a problem using the lab supply current limit to protect a transistor if the supply has a large output filter capacitor. The energy in the capacitance could be enough to blow the transistor even if the current limit is operational.

 

[Sceadwian]

Chris, the solution is simple, don't make careless mistakes =) Slow down and double check what you're doing, remember that every time you have to replace the FET's in your bridge.. There is unfortunatly no circuit capable of protecting itself from misuse. Everyone that experiments with electronics has done something like that, it's easier to simply tread carefully, considering if you have the knowledge to build a completly fool proof circuit in the first place you probably have enough knowledge to not make the mistake that requires it.

 

[MikeMl]

There could be a problem using the lab supply current limit to protect a transistor if the supply has a large output filter capacitor. The energy in the capacitance could be enough to blow the transistor even if the current limit is operational.

This particular supply (made by HP) has very small output filter capacitor used only for stiffening the output at high frequencies; the filter capacitor on the regulator input is conventional. The current limiter seems to react fast.

 

[oleone]

This might be a belated reply, but in my PWM project, I've been having a similar issue with my super H-Bridge of Evil. You have to create a delay circuit, either by using logic gates upon the output of your microcontroller, or program in a particular delay into your program.

How much of a delay? Have to read the datasheets of your mosfets to see how long it takes to turn on and off. I also have a .007 ohm resistor in series with my load to connect to an op-amp based over current protection. You can then use that as an analog input into your microcontroller to shut off/send warnings/ shock the user.
Here's a mighty good website that incorporates a RC circuit into a logic to create a sure "dead Time"

**broken link removed**