I am trying to come up with a system for pulse-by-pulse current monitoring for my controller. I am considering using a shunt resistor like this (.0005Ω)for each mosfet. For example, if I wanted to limit my current to 50A through the mosfet, this resistor would output 25mV for 50A through it. I would use a comparator to compare this voltage output to a fixed reference voltage (25mV) so that I can shut off the Mosfet in the middle of PWM duty cycle (via the driver enable pin). So I need a super fast comparator that has a very low (1mV?) resolution. I am looking at this one. It is super fast but I can't figure out what the resolution is. it lists Input Offset Voltage @ 1-3mV but I'm not sure if that's the number I should be looking at. I'm just assuming that there's such a thing as a resolution rating; is there? I don't see one here.

 

The resolution is determined by the voltage gain of the comparator. For the LM161 the gain is typically 3V/mV (3000) so 1mV of differential drive (ignoring any offset) will cause the output to change 3V.

But I think your bigger problem will be noise in your circuit. It will be difficult to keep transient ground noise voltages below 25mV if you are switching 50A through a MOSFET. The common for the 25mV reference and the common for the shunt resistor must be essentially the same point for you to have a chance of reliably detecting the current limit point without spurious tripping.

 

One way to reduce noise sensitivity is to lower gain. If you are using low side switches, an easy way to do that is by sampling the voltage drop across the MOSFET instead of adding a series current sense resistor. Since MOSFETs usually have resistances in the tens of mΩ, you should get a much larger signal to sample from which will allow you to run lower gain. Of course, there are variations due to temperature and load, but it should be good enough for current limiting.

It's a little more tricky to sample the voltage drop across high side switches or high side current sense resistors, but there are current sense amplifiers designed for that. Here is the product line from Maxim.

 

One way to reduce noise sensitivity is to lower gain. If you are using low side switches, an easy way to do that is by sampling the voltage drop across the MOSFET instead of adding a series current sense resistor. Since MOSFETs usually have resistances in the tens of mΩ, you should get a much larger signal to sample from which will allow you to run lower gain. Of course, there are variations due to temperature and load, but it should be good enough for current limiting.

You know I agree that would be a superior way to do it, if I could make it work, because rather than a simple current measurement, it would be a dynamic "power dissipated in the MOSFET" measurement. As the mosfet heats up, it's resistance will increase, causing a larger voltage drop, causing more dissipation. The only problem is that when the MOSFET is OFF, there will be 72V across it so the comparator will never let it turn on(this also necessitates a comparator that can withstand a high voltage). I would need some clever way of having it only look at the voltage across the MOSFET during the ON portion of the duty cycle (actually, part-ways into the ON cycle, after enough time for full turn-on to occur). Do you have any ideas for that?

It's a little more tricky to sample the voltage drop across high side switches or high side current sense resistors, but there are current sense amplifiers designed for that. Here is the product line from Maxim.

Yes, for the time being, I'm only doing low side switching. Later on, after I get this controller going, I will be working on a brushless motor controller, so I will be doing lots of high side switching. I'm sure this link will be invaluable to me at that time.
Thank you

 

.................
The only problem is that when the MOSFET is OFF, there will be 72V across it so the comparator will never let it turn on(this also necessitates a comparator that can withstand a high voltage). I would need some clever way of having it only look at the voltage across the MOSFET during the ON portion of the duty cycle (actually, part-ways into the ON cycle, after enough time for full turn-on to occur). Do you have any ideas for that?
..............

You could use another small MOSFET to switch the voltage to the input of the comparator when the power MOSFET is on. You could add a small delay to the small MOSFET signal by using an RC delay into a Schmidt trigger driver circuit such as the CD4584, for example. You might want to use a one-shot also so the small MOSFET is off before the voltage starts to rise.

 

You could use another small MOSFET to switch the voltage to the input of the comparator when the power MOSFET is on. You could add a small delay to the small MOSFET signal by using an RC delay into a Schmidt trigger driver circuit such as the CD4584, for example. You might want to use a one-shot also so the small MOSFET is off before the voltage starts to rise.

ok, sounds good. I will look into it. I appreciate the help.

 

It's a little more tricky to sample the voltage drop across high side switches or high side current sense resistors, but there are current sense amplifiers designed for that. Here is the product line from Maxim.

I've just been thinking about this; why is harder for high side? In my limited understanding, a certain resistance should have a certain voltage drop across it for a certain current through it, no matter where it is in the circuit. I don't understand

 

Only because you typcially have to deal with an offest voltage differential in your signal. For low side sensing, your signal has a ground reference. For high side sensing, you're using input voltage as a reference. It's not a major problem, but can be tricky when measuring differences in voltages with a much higher reference than the operating voltage of your sensing electronics.

For example, high side current sense amplifiers typically offer an output referenced to ground, but can output signal levels close to input voltage if currents are high enough. That can happen if a current spike occurs. You need to use a clamp of some kind to protect your sensing electronics from those conditions. With low side sensing, the rail to rail limitation of the amplifier clamps output to acceptable levels. No additional consideration is required. Though, high side sensing does have its advantages.