Part Number: INA901-SP
Tool/software:
Hi Team,
Is there a way to calibrate voltage error for the shunt of 2.5mV?
-Jared
Hi Carolus,
I believe the concern is centered around the spec below:
The goal would be to try and calibrate this aspect of the device to reduce the offset voltage. Is there a known good way to do this?
-Jared
Jared,
Your math is sound here, and calibration is ideal if the customer can do it. Typically, the easiest way to do this is to utilize logic. One point and two point calibrations are typical in higher end systems to remove offset and gain error. A one point calibration removes offset error, where a two point calibration removes both offset AND gain error from the response.
To perform a one point calibration on the operating point you mention above, the customer would capture the input value of their shunt voltage (measure the true input, don't just assume 20mV), as well as the actual output value of the sensor at this point, and subtract the ideal transfer function value from it (for the INA901, the only GAIN option is 20V/V, so it would always be 20*Vshunt). The remaining value would be the curve offset (which can be positive or negative), and would need to be stored in memory. This value is then subtracted from every subsequent measurement moving forward to correct the part for the offset. An example curve is shown below. Note that the curve below does not show gain error, which would still very much be present in the part, and would require a two point calibration to adjust.
I will be travelling first thing Monday morning to Dallas, but I'd be happy to get on the phone and talk through the opportunity early next week if you think that would be beneficial here. We can discuss what design options the customer has available to them and how we can best support the customer here.
Jared,
Let me tackle these one at a time, as each answer is quite different.
On the question of time, as far as I am aware, there are no long term drift issues for the INA901. We put these devices through accelerated life testing and they remain within datasheet spec.
On the question of radiation, the answer is yes, but the magnitude is not so severe that it should matter. Here is the result of the offset extracted from the TID report of the device:
As you can see, even after 50krad exposure, the average change for a typical device is less than a 1mV in the extreme cases. This would be challenging to build into the calibration, because we do not know at any given moment what level of radiation exposure the device has been subjected to.
Finally, with regard to temperature, the answer is yes, temperature drift will exhibit additional offset in the device, and the drift coefficient is listed in the datasheet (shown below):
This additional offset can be calibrated out, but unfortunately, using the method I describe above you can only calibrate this for one temperature. If the customer plans to operate at a specific ambient operating temperature in their design, I would recommend they calibrate there to achieve optimal accuracy at the normal operating condition. Alternatively, if they have access to a temperature measurement in proximity to the device, they could attempt to remove the drift entirely via temperature correction (capture the temperature in the environment and subtract 25°C [TI's datasheet condition for ambient]. Multiply this result by the dVos/dT. Subtract this result from the measured output value along with the offset value from our previous conversation).
