This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

INA301-Q1: Low current delay / compression

Part Number: INA301-Q1
Other Parts Discussed in Thread: INA301

I have read several questions posted regarding low current "delay" or compression. (Is there a correct name for this behavior?) I have observed the same behavior with an inductive load.

If you can see the terrible scope picture below, the pink line is a Tektronix current probe, and the washed-out blue line is the output from the INA301. There is a 300 uS delay, and then the output jumps to about 60 mV. I do not understand why it behaves this way. If I connect a resistive load, it delays about 10 uS and rises quicklly. I'm using a 50 mOhm Kelvin shunt.

My real question is this: Is this behavior inherent to all / most of the current sensing amplifiers? I'm just curious if this is something I can avoid by selecting a different device, or if this is just the nature of current sense amps.  I use these with common modes of up to 24V, usually automotive or industrial engine applications. Current rise from 0 to 1.5 A is in the 15 mS range - quite slow.

Thank for your time and efforts.

  • Michael, 

    What is more than likely happening here is not due the output response of the device (which is shown in figure 24 of the datasheet and copied below, but rather the saturation recovery response. 

    Like all op amps, our devices have swing limitations, and operating in single supply mode, the lower rail is GND. As shown in the datasheet, the worst case swing to GND is 30mV, and the typical for this device is 20mV.

    What this means is that you must ensure the output remains above 30mV to design for worst case (for the given test conditions. For other load conditions, see figure 16) for the device to remain in linear operation. If the device is driven below 30mV (again, you may be able to go as low as 20mV typically), the device will saturate against the rail, and enter a non-linear state. Based on your scope shot, it appears as if you are beginning the test at roughly 0A (or possibly even a few mA's below based on the curve appearing slightly below the reference point of CH3) which with no differential signal being placed across the shunt. This would attempt to drive the device to 0V and therefore saturate the device. When this happens, and the device is brought back into the linear region, there will typically be some time associated with this recovery, commonly known as saturation recovery time, which is most likely the several hundred μs you are seeing. I would recommend trying your test again, but try a base current of a few mA's or so to bias the device into the linear region and you should see the response time of the device improve. 

    A way to easily avoid this is to use a "bidirectional" device, where you simply reference the output stage so that it sits above the swing to GND limitation, and then can be driven to true zero without fear of saturation. This comes at the tradeoff's of a small sacrifice of full scale range, and the need to produce a reference voltage capable of driving the REF pin to the desired level. 

  • Thank you for the clear explanation. I was unaware of the saturation issue on the single supply devices, but that makes sense. I didn’t understand the 20 / 30 mV to mean a saturated value, but the scope trace says it is. 

    Yes, the scope trace starts at zero current, so the output is driven to the negative rail. 

    The “bidirectional” solution will work out fine. There’s a couple of avenues to take in that direction.