Hi Team,
My customer is working on a design that uses the DRV425 and they have a couple questions that I’d like to get your help with.
It is worth noting that in their design, they take the single ended measurement of each opposed sensor and perform the differential operation elsewhere. They deviated from the reference design on this in order to increase dynamic range. The sensing current is in the ballpark of -400A to 400A.
Here is a summary of the problems that they are seeing:
1. Offset Deviation:
We run two different calibration steps on our board: once when it is mounted to the busbar, and once when it is married to the other board with the rest of the signal conditioning circuitry/ADC. We are running a linear regression at many points; however, we see that the offset (i.e. the correction factor at zero current) does not remain persistent. We have seen this on multiple revisions of hardware on both the horizontal and vertical mounting orientations. The gain appears to remain correct, but the offset changes (due to an unknown factor after an unknown amount of time) by a sizable amount (on the order of 6 to 12 amps). This is definitely not within the realm of acceptable for their application. This is the biggest problem that we are facing. If we can't come up with a solution for this, we will have to investigate other options for our application. We would like to hear your thoughts for what might be the cause of offset error in this product, and ways to mitigate it.
2. Linearity at Low Currents:
When we run the calibration, we take a linear regression based on the difference between the output of the two sensors, and current from a lab calibrated shunt (accuracy of the shunt is 0.02%). We then apply the linear regression back to the data and take the absolute and percent error relative to the reference shunt readings. We find that we get very non-linear at the lower end of our range. We figure we are approaching some sensitivity limits of the sensor. We are considering running a interpolation/look up table type solution at the lower end of our spectrum, but I don't explicitly see anything in the datasheet for "repeatability of error" if that makes sense.
3. Linearity and Supply Voltage:
In the datasheet, there is a plot of linearity versus supply voltage. Currently, we are running at 5V, but we could potentially modify our design to run at 3.3V (which, according to the plot, will give us better linearity by about a factor of two). Can you give us details on what this exactly means? Do you think this could help with the problem mentioned in 2, or do you think that is more of an issue with the general sensitivity of the sensor?
If you would like to access the schematic or calibration data, please let me know and I will share that with you over email.
Thanks,
Mitchell
