The most common mistake when comparing MEMS gyroscopes is ranking them by a single bias-stability figure. That figure is measured under controlled laboratory temperature. On a real airframe, temperature moves, vibration is present and mounting stress changes — and those factors drive the drift you actually see.
What bias stability describes
Bias stability measures the fluctuation of angular-rate output while stationary, usually expressed in degrees per second or per hour. It reflects device noise and matters, but only under ideal conditions.
The critical point: **bias itself drifts slowly over time and is temperature-sensitive.** A part with excellent laboratory bias stability can produce far more error than its noise figure suggests if thermal compensation is poor.
Three parameters that must be read together
- **Bias stability**: governs short-term angle random walk and attitude jitter.
- **Temperature coefficient / full-temperature bias drift**: governs how far attitude shifts as the environment changes. This predicts field behaviour better than the noise figure.
- **Scale-factor error and its thermal drift**: a proportional error in rate measurement that also integrates into attitude error over time.
Optimising only the first while ignoring the other two is a routine selection error.
Thermal compensation is not a cure-all
Most modules implement compensation, but its quality depends on the calibrated temperature range and the rate of temperature change:
- Compensation works well inside the calibrated range and can degrade sharply outside it.
- Rapid temperature change creates internal gradients that defeat compensation based on a single temperature sensor.
- Compensation coefficients are stored on the device; a different post-assembly thermal environment — near a hot ESC, for example — can invalidate the original calibration.
Ask for measured curves across the full temperature range, not a statement that compensation exists.
Mounting matters more than most teams expect
Gyroscopes are sensitive to mechanical stress and vibration:
- Improper PCB fixation introduces stress that varies with temperature cycles, masquerading as bias drift.
- High-frequency vibration rectifies into false angular-rate output, especially when the gyro shares a structure with motors.
- Mount close to the vehicle's centre of rotation to reduce error coupling from linear acceleration.
Many apparent device-drift problems turn out to be installation problems.
Relationship with INS fusion
A gyroscope alone outputs angular rate; integrating it yields attitude, with error accumulating over time. An INS constrains that drift using accelerometers, magnetometers or GNSS. Ask:
- Does the fusion algorithm estimate internal parameter drift online?
- After losing external observations such as GNSS, how long does attitude remain usable?
- Does the module output a confidence measure so the host can degrade gracefully?
Acceptance checks
- Run a chamber test across the target temperature range and record the actual bias-versus-temperature curve.
- Compare static and vibrating conditions on the assembled airframe to confirm mounting adds no error.
- Perform a long-duration attitude-hold test and check the drift rate against the system budget.
- Require suppliers to state calibration conditions and compensation range, not just room-temperature figures.
Conclusion
Bias stability sets the device floor; thermal drift and installation quality set the field ceiling. Verify against the real temperature range and mounting environment rather than ranking on one parameter.