Hi everyone,

Happy new year!

I am trying my hand at an overcurrent / short-circuit protection circuit. This is a small portion of a larger circuit that has been omitted, but I'll be using an ESP32 as the microcontroller. The uC output will be driving a P-channel mosfet through an NPN transistor. I have a 1 mohm 2W shunt for measurement and ground for reference on a rail to rail differential opamp. My question is about the resistor below the zener diode (2.2V for 22A trigger). I just picked a lowish value, not wanting to interfere with the ground reference on the zener. I'm not really sure on how this will interract/interfere with the current trigger point.

Also this is for automobile purposes and want to make sure the non-stable 12 / 13 volts plays no role in this. the 3.3V is coming from a switch-mode power circuit.

The uC input is for current monitoring.

Any advice on this would be helpful as this is my first attempt on this. also would this be best practice as a mass produced product, considering safety of failed/ing components?

Thanks in advance.

G

 

The uC output will be driving a P-channel mosfet through an NPN transistor. I have a 1 mohm 2W shunt for measurement

And you can stop right there. Monitor the voltage across the resistor with the micro’s ADC and cut power when it exceeds the limit. No additional circuitry is needed.

Best practice for a commercial product is simpler still, it is called a fuse, as found in every power circuit in a car.

 

And you can stop right there. Monitor the voltage across the resistor with the micro’s ADC and cut power when it exceeds the limit. No additional circuitry is needed.

Best practice for a commercial product is simpler still, it is called a fuse, as found in every power circuit in a car.

Hey Bob,

Thanks for the relpy.

I should have clarified.

The over current protection is actually further protection to the mosfets in case the current is close to the fuse rating. I should have mentioned there will be 4 of these mosfets driving different high-ish power devices, so with a reasonably high 100 A fuse for the main board, I thought better to protect the 4 mosfets with their own circuitry.

As for the micros ADC, its only as fast as the time taken to run through the code before getting to the analog reads. which could be some time given the complexity of the firmware, running bluetooth and a bunch of protocols. Both cores are actually quite full and a few microseconds could be all it takes.

As for the ADC, i figured, I'm using a diff-opamp config anyway for the SC protection, i might as well use the increased resolution of the opamp output for the uC to map the current.

 

The micro would still be far faster than a fuse.

The way you do this is to set the ADC to sample continuously and interrupt when done. The interrupt handler could disable the output is in well under 1 microsecond.

But do whatever you want, just giving you my opinion.

 

Is the MOSFET so under-rated that it is running
on the ragged-edge of releasing the infamous Blue-Smoke
so that You have to have virtually instantaneous disconnection for it to survive ?

A Current-Limiting-Circuit may be in order,
then it will become a matter of adequate Heat-Sinking,
and the Lock-Out-Protection-Time can be much longer.

There are already quite a few Automotive-Aftermarket-Manufacturers that
have worked all of this out by popping lots of FETs so You don't have to.

N-Channel FETs are cheaper and more rugged.
.
.
.

 

Is the MOSFET so under-rated that it is running
on the ragged-edge of releasing the infamous Blue-Smoke
so that You have to have virtually instantaneous disconnection for it to survive ?

A Current-Limiting-Circuit may be in order,
then it will become a matter of adequate Heat-Sinking,
and the Lock-Out-Protection-Time can be much longer.

There are already quite a few Automotive-Aftermarket-Manufacturers that
have worked all of this out by popping lots of FETs so You don't have to.

N-Channel FETs are cheaper and more rugged.
.
.
.

I considered bootstrapping an N channel but im not good enough with the math to get it switching at 20khz pwm.

The typical power would be between 50-120W, 4-10 Amps, but you never know what people will wire in. With the packages and dissipation im using it should be safe up to 15-20 amps but I'll impiracly test once I have the boards made.

The micro would still be far faster than a fuse.

The way you do this is to set the ADC to sample continuously and interrupt when done. The interrupt handler could disable the output is in well under 1 microsecond.

But do whatever you want, just giving you my opinion.

There will be a fuse in place to protect the wires to the board, im just trying to make this thing as safe as is practicable. How can you get the ADC to sample continuously, then interrupt? I've never heard of any micro being able to do that, that wasn't a digital interrupt.

But to circle back to the OP, should my schematic / values work?

 

Im suppose this short circuit protection is because I'm not really sure of the burn profile of all 60A fuses, and these mosfets have a few (2.7) milliohms of resistance when fully on, its easier to not gamble with a dead short which will let out the smoke first.

Maybe you guys know something I don't. Which is why I'm posting on here before designing it in.

 

The led won’t ever light up, nor the positive feedback be achieved, because Q3’s collector is clamped to 0.7v by Q6’s BE junction.

It is a fun and a learning experience to attempt to reinvent the wheel. If this is a one-off project, by all means continue your quest.
But if this is something that will end up in somebody’s else’s hands, then use the automotive high side drivers available from STMicro or Infineon. They have products which have all sorts of protection features, including the main Fet’s die temperature and SOA, which would be from difficult to impossible to implement on a discrete circuit.

Search automotive high side switches.

 

The led won’t ever light up, nor the positive feedback be achieved, because Q3’s collector is clamped to 0.7v by Q6’s BE junction.

Thanks for catching that.

Would adding a resistor between the collector of Q3 and the base of Q6 work to solve that problem?