3 Calibration-Enabled Near-Zero V_sense Measurement

In some applications, relying on the data sheet specification of an otherwise well-fitting CSA is not enough to satisfy the accuracy requirement. If this is the case, calibration can provide a path forward by improving system accuracy.

Offset calibration can be adopted to improve the accuracy of a current sensing system at near-zero V_sense During offset calibration, individual system’s offset is measured and stored, which is then subtracted from future measurements. A block diagram of CSA offset calibration is shown in Figure 3-1. The voltage source V_os stands for the input referred offset voltage of the device.

One of the key considerations in offset calibration is to keep the calibration path and the normal signal path overlap as much as possible. By including the complete signal measurement path in the calibration path, all error contributors are taken into account. The cumulative effect of all the error contributors can be subtracted with one calibration operation. When additional components are introduced to enable calibration, the tolerance of such components must be taken into consideration; the effect on normal signal measurement must be considered as well, so that any adverse effect is minimized.

Another consideration is to keep the system in linear operation range. For example, if the CSA output is clipped to either power supply rails, the calibration results are invalid.

In Figure 3-1, the inputs are shorted together to perform offset calibration. However, the swing limitations of the CSA very likely causes the output of the CSA to stuck to ground, masking the effect of V_os. The correct setup is to configure the CSA so that the output can move freely with zero input.

Biasing the output to higher than ground gets away from swing limitation. In the setup shown in Figure 3-2, a bidirectional CSA is configured with a reference voltage equal to half supply. With inputs shorted, the CSA output is compared with the reference voltage by the PGA. The difference equals to the output offset. To calculate input referred offset, use Equation 2.

To build upon offset calibration, gain error can be calibrated with at least one additional data point. Figure 3-3 shows the CSA driven with a nonzero input V_in. The corresponding output is PGA_Out2, and the CSA gain can be calculated as shown in Equation 3.

The nonzero input is assumed to be a known accurate value. In practice, the input is either measured with a precision meter, or is provided by a precision source. In either case the true magnitude of the input must be available, otherwise uncertainties in the input value negatively impact the accuracy and defeat the purpose of calculation.

These are some of the basic calibration schemes. More elaborate ones are possible to achieve higher levels of accuracy.