How to reduce the influence of quadrature error by using industrial grade IMU

I’ve been thinking about the MEMS Inertial Measurement Unit (IMU) used in the self-balancing guidance systems of personal vehicles. There’s always curiosity about whether consumer-grade IMUs could eventually become advanced enough to eliminate all alignment errors between sensors and integrate all core components onto a single chip. While that sounds like an appealing idea, it's not something we should expect anytime soon, especially when it comes to industrial-grade solutions.

Industrial-grade IMUs, with their robust discrete sensors, optimized packaging, and factory calibration processes, offer significantly better alignment accuracy than most consumer-grade IMUs. Consumer and industrial IMUs tend to describe their shaft alignment characteristics differently. Consumer IMUs typically lump all alignment errors into one cross-axis sensitivity specification. On the other hand, industrial IMUs like the ADIS16490 break this down further, using two distinct metrics: axis-to-axis alignment error and axis-to-package alignment error. The axis-to-package alignment error assesses how well each axis aligns with the physical structure inside the IMU package, while the axis-to-axis alignment error evaluates how closely the accelerometers and gyroscopes align orthogonally. This latter specification is sometimes called quadrature error.

The mathematical relationship between cross-axis sensitivity (CAS) and axis-to-axis alignment error (A2A_MAE) is straightforward: CAS = sin(A2A_MAE) and A2A_MAE = asin(CAS). Non-orthogonality between sensor axes can arise due to manufacturing variances, even if the sensors are located on the same chip. Industrial-grade IMUs address these issues through rigorous factory calibration, providing precise details in their datasheets. For discrete devices, the cross-axis sensitivity specification doesn’t account for assembly deviations relative to the PCB.

It’s a common misconception that because multi-axis accelerometers or gyroscopes are fabricated on the same MEMS chip, their axes are perfectly orthogonal. While all inertial sensors might be on one chip, manufacturing variations can still introduce significant quadrature errors. Compared to fully calibrated industrial-grade IMUs, the alignment accuracy isn’t nearly as impressive. Take, for instance, consumer IMUs, where cross-axis sensitivity often ranges from 1% to 5%. Using the formula above, this translates to an equivalent axis-to-axis alignment error of 0.57° to 2.87°, or 0.057° in milliradians. Industrial-grade IMUs are far more precise; for example, an industrial-grade IMU with an axis-to-axis alignment error of 0.018° has an equivalent cross-axis sensitivity of 0.031%.

Even though the industrial-grade IMU ADIS16489 doesn’t house all its inertial sensors on a single chip, its performance is still approximately 32 times better than the best consumer-grade devices. Quadrature error, a stable component of total accelerometer error, can be corrected through a one-time calibration process. By rotating the device across all possible 90° directional spaces and measuring the static response of each axis to gravity, you can determine and correct this error. However, achieving zero quadrature error post-mounting on a PCB under all conditions is challenging, requiring specialized calibration techniques and equipment.

In contrast, industrial-grade IMUs like the ADIS16489 are pre-calibrated at the factory to minimize alignment errors, including those caused by assembly, cross-axis sensitivity, and temperature effects. This results in exceptional performance in applications like platform stabilization, navigation, and robotics. The ADIS16489 boasts a gyroscope axis-to-axis alignment error of just ±0.018° and an accelerometer axis-to-axis error of ±0.035°. Beyond its excellent sensor performance, the ADIS16489 features a parylene coating to protect its internal circuits from moisture.

While the idea of a single-chip consumer IMU with perfect alignment is enticing, industrial-grade IMUs currently offer unmatched precision and reliability. For now, the gap between consumer and industrial-grade IMUs remains substantial, and it will likely stay that way until manufacturing processes improve significantly.