Disclaimer

The algorithm explained here is determined during the tests. It can be different from the original undocumented algorithm used by Witmotion in their own sensor control tool.

Data layout

In witmotion_config_packet the setting field uses inverse-byte data layout:

• setting.raw[0] is the fine tuning part of bias from 0.0 to -0.25, divided into 256 grades, uint8_t
• setting.raw[1] is the rough tuning part of bias, imposed as signed integer, 127 bidirectional grades, int8_t

Calculation

The sensor built-in calibration coefficient is 4, so the original formula looks as the following:

\begin{eqnarray*} B &=& bias \\ B_a^{rough} &=& trunc(4 \cdot B) \; \mbox{if} \left| B \right| < 127 \\ \left| B_a^{rough} \right| &=& 127 \; \mbox{otherwise} \\ B_a^{fine} &=& \frac{255}{0.25} \left( 0.25 - \left| bias \bmod 0.25 \right| \right) \mbox{if}\; bias < 0 \\ &=& \frac{255}{0.25} \left| bias \bmod 0.25 \right| \; \mbox{otherwise} \end{eqnarray*}

Then the fractional parts of \( B_a^{rough} \) and \( B_a^{fine} \) should be discarded and the values exposed as signed integers into setting.raw subfields.