INA240: CURRENT SENSE RANGE

While INA240 is called a "current sense amplifier", this doesn't mean that it literally measures current. It is an application-specific instrumentation amplifier that measures the difference voltage between its two inputs, and outputs this difference amplified. Thus, whether it can measure 150A of current depends entirely on what you use to convert said 150A of current into the voltage the amplifier expects. In other words: the question about "what TI part would work for 150A" could be asked if TI made current sensors. As far as I know, TI doesn't make them.

Thus, it's up to you to provide the current transducer: a device that converts current to voltage that the INA240 could amplify. Of course now the obvious question is: the amplifier is the second in the signal chain. What is your transducer? That device is often a shunt resistor. But that's not the only option, so starting the design from the amplifier out is not always the best approach.

INA240 is available in one of four fixed gains: 20, 50, 100 and 200. The output swings essentially rail-to-rail, but beware: it can't swing fully to GND, so the current measurement range doesn't extend to 0.000A, but a bit above it.

Suppose now we selected a 200 gain part, and suppose the 150A should cover 4.5V of output range - that will depend on the ADC you're using. We'd need to have 4.5V/200 = 23mV on the INA's input. Since the gain of 200 is "similar" to the full-scale current of 150, we can instantly approximate the power dissipated in the shunt to be a bit less than the full-scale amplified output voltage, i.e. ~4W, or more precisely 23mV*150A=3.5W. If you choose a part with lower gain, this power dissipated in the shunt will go up, since there'll be a larger voltage across the shunt.

It is a typical requirement in modern designs to minimize volume and power dissipation, thus it's likely you'd have to use the highest gain option. To maintain gain accuracy of the tranducer (the shunt), you can make it equithermal with something else that has known temperature. Sometimes a power stage module has a dedicated heatsink temperature sensor, or perhaps your fan controller or system monitor has a sensor attached to the heatsink - that could be then used to calibrate the transducer's gain-vs-temperature dependency in software. Otherwise, you'd use a temperature sensor attached to the shunt itself - that's often problematic. You may find out that the shunt's temperature coefficient is acceptable - but this can't be ignored, since shunts inherently warm up during normal operation. You'd have to calculate the change in full-scale reading across the entire temperature range you expect the shunt to be at (i.e. lowest ambient temperature to highest ambient+shunt temperature rise due to full load).

The shunt for the 150A, 4.5V full-scale output would have 23mV/150A=0.15mOhm resistance. A quick check on DigiKey yields a 150uOhm, 37.5W rated shunt - perhaps too big or too costly for your application. A search for a wider range of shunt resistances, and a broader range of suppliers, would yield other suitable parts.

And let's not forget that using shunt resistors as current transducers isn't the only solution. Hall sensors with built-in signal conditioning can be used as well. For AC currents you can use Rogowski coils, which can often be attached "directly" to ADC inputs via a transient protection network - no need for preamplification. There are plenty of options, and a lot depends on cost, size and time to market constraints.