Over the past decade, we have seen tremendous investment in new processes, ASICs, sensor structures and packaging from MEMS suppliers. Nearly every manufacturer is focusing on reducing the cost, size and power of their IMU’s core.
What’s the result?
Six degrees of freedom (6DOF) IMUs are now available with sub 1 mA of power consumption during normal operation in LGA/QFN packages that are smaller than 16 mm2.
This is a great achievement for wearable and handset applications, but at the same time, the proliferation of the Internet of Things (IoT) is driving innovation at a faster pace than sensor manufacturers can keep up with and new professional consumer applications are emerging, such as drones, gimbals, autonomous robots, action cameras and VR/AR headsets.
All these applications are demanding 3D orientation tracking with very high accuracy, low latency, large bandwidth, and low drift over time and temperature. Key design concerns are:
- Immunity to vibration
- Gyro bias stability over time
- Gyro and accelerometer bias and sensitivity stability over temperature
- Wide bandwidth with low latency
- Signal pipeline synchronization
- Low noise
- Embedded on-chip SDI computation
- Kalman filter output at 1 kHz+
- No board calibration
These applications are typically not space- and power-constrained.
It’s all about performance
MEMS IMUs optimized for lowest power consumption typically don’t have an on-chip SDI or wide bandwidth and good signal pipeline, as they will feature one MUX/DEMUX with a single SAR ADC for all six channels.
Optimizing for a tiny package offers better fit for small devices and PCB boards, but it compromises the mechanical performance of the MEMS sensor, which requires additional post-calibration, as the MEMS sensors are susceptible to thermal and mechanical stress.
Put simply, products designed for mobile phones are not the best fit for prosumer applications.
At Fairchild, we are targeting high-performance IMUs.
The FIS1100 + XKF3 is a system solution comprised of hardware and software providing orientation specification to our customers (AHRS and VRU). The architecture of the FIS1100 has been purposely designed to address such prosumer applications. The AttitudeEngine (AE) computes the high frequency SDI (Strap Down Integral) reducing the coning and sculling effect due to vibration. Furthermore, the FIS1100 synchronizes data coming from the accelerometer, gyroscope and magnetometer, allowing the sensor fusion on the host to compute the most accurate orientation estimation. The AE can be enabled or disabled as users require.
For applications that require accurate orientation, the FIS1100 shows great advantages:
- The AE integrates angular velocity and acceleration at 1 kHz, reducing the accumulation of error offloading the main application processor. It also synchronizes the entire sensor pipeline to reduce error in the prediction stage of the sensor fusion.
- Yaw axis behavior shows amazing stability over time of the gyro bias (~6dph) due to the mechanical design of the sensor (Allan variance is available to our customers), enabling outstanding heading tracking for applications such as robotics and navigation.
- XKF3 corrects the gyro bias and provides auto calibration on the go. The sensor fusion algorithm works both as VRU or AHRS and also works at 6 or 9 DOF.
- Systems with small electric motors experience vibration. Using the AE mitigates this effect and sensor data will not be compromised.
Optical/Electronic Image Stabilization applications (close loop configuration and mechanical actuators) could be very challenging without the FIS1100. Common concerns addressed by the FIS1100:
- Low latency is important for applications to identify the orientation and immediately correct and compensate
- Higher image sensor resolutions demand lower gyro noise, small movements could compromise multiple pixels
- Higher bandwidth helps the system respond accurately to vibration and movement induced by aerodynamics, wind, motors
- Gyro bias stability over time and temperature
The FIS1100 offers a high performance OIS mode, which reduces gyro noise (6 mdps) with 1.6 KHz bandwidth at 8 kHz ODR with low latency. The accelerometer features 50uG/√Hz, which is ideal for dead reckoning applications when used in conjunction with an ultra-low noise gyro.
Additionally, for customers who need a fast easy-to-integrate, turnkey motion tracking solution, Fairchild provides the FMT1000 module series. This family of IMU/VRU/AHRS modules is system-on-board (12×12 mm) and is comprised of a µC (with pre-loaded sensor fusion) along with the FIS1100 and a 3-axis magnetometer. Each module is individually calibrated and tested.
With both product options, customers have access to the XKF3, the extended Kalman filter library developed by Xsens along with extensive evaluation kits and SDKs with source code and example applications for a variety of target platforms. The XKF3 library is available for multiple processor ARM® cores: M0, M3, M4, M4F. The FMT1000 does not require any software integration by the customer and is supplied with extensive driver source code and ARM® mbed™ application examples.
The Fairchild motion tracking portfolio provides our customers with accurate solutions, choice and flexibility at design time.
Learn more about Fairchild’s motion tracking solutions and resources:
Motion tracking modules: https://www.fairchildsemi.com/product-technology/mems-module/
Understanding the AttitudeEngine application note: https://www.fairchildsemi.com/application-notes/AN/AN-5083.pdf
Understanding XKF3 (Sensor Fusion) application note: https://www.fairchildsemi.com/application-notes/AN/AN-5084.pdf
This post is also available in: Chinese (Simplified)