• Six-Position Method (Multi-position Evaluation): The sensor is placed in six static positions (each axis facing up and down) to measure +1g and -1g, allowing for the calculation of zero-g bias and scale factor.
• Two-Position Method : A faster, less precise method often used, involving measuring the +1g and -1g output to calculate offset and gain.
• Temperature Compensation : Sensors are placed in a thermal chamber (e.g., -40°C to +85°C) to record data for calculating thermal drift of bias and scale factor, ensuring stability across temperature ranges.
• Laser Trimming : During the manufacturing process, silicon structures are laser-trimmed to mechanically adjust the sensor's sensitivity and zero-g offset to within tight tolerances.
• Sensor Fusion/Kalman Filters : In high-precision applications, calibration algorithms like the Kalman filter are used to minimize errors and continuously update the sensor state based on gravitational input.

• Zero-g Offset/Bias: The output when no acceleration is applied, calculated as
  

  $0.5\times (\mathrm{+1}g+-1g)$
  .
• Scale Factor/Sensitivity: The ratio of change in input to output, calculated as
  

  $0.5\times (\mathrm{+1}g--1g)$
  .
• Non-orthogonality: Correcting for misalignments where axes are not perfectly at 90 degrees to each other.

• Precision Test Fixtures: Non-magnetic, precision-machined fixtures for consistent, accurate positioning of the sensor in 3D space.
• Data Logging & Post-Processing: Raw data is captured at each position and processed, often in software, to generate calibration constants stored in the device's non-volatile memory.
• Stable Environment: Maintaining constant temperature and reducing external vibrations during the process is critical.