Our motion sensor group continues to shrink north finders, now we are at approximately 2x2x4 cm size. The next goal is north finder system on a chip (NFSoC) which would allow smartphone applications. North finders measure the earth rate vector and projection of this vector to locally level frame is directly what we call direction of ‘north’. Traditional compass does something different. For example, in smartphone applications, after awkward calibration procedure to remove phone’s own magnetic field (Figure 1), the application aligns the compass needle based on magnetic field vector (Figure 2).
Figure 1. Typical magnetometer calibration procedure
This vector does not point towards true north, there is declination correction on the way and local magnetic anomalies to be handled (bring a magnetic screwdriver close to the phone to see what we mean). When we have finished our work, you don’t need to care about magnetic distortion or declination!
Figure 2. Compass applications often derive true heading from magnetic heading and location, and this relation is not known indoors
There are two vectors we need to measure in the process: (plumb-bob) gravity and Earth rate. In principle, those should be observable by smartphone inertial sensors when the phone is stationary. You can try yourself, using for example AndorSensor application. Place the phone on a relatively flat surface and check what the inertial sensors show. In the example shown in Figure 3. linear acceleration is off about 0.6 m/s^2 that is about 6 % of the full vector norm 9.81 m/s^2. But how strong is the earth rate? It is approximately 15.04 deg/h (full revolution relative to the Sun per day + one around the Sun per year) , e.g. 0.0042 deg/s e.g. 0.0000729 rad/s. The snapshot sample of gyro shows 0.003 rad/s, 4100 % of the actual rate! Earth rate signal is so weak that we have serious challenges ahead *. You can read about our approach to tackle this challenge from our latest journal publication:
J. Bojja, J. Collin, M. Kirkko-Jaakkola, M. Payne, R. Griffiths and J. Takala, “Compact North Finding System,” in IEEE Sensors Journal, vol. 16, no. 8, pp. 2554-2563, April15, 2016.
doi: 10.1109/JSEN.2016.2518860 (link)
Open access post-print version is available from TUTCRIS.
Figure 3. Screenshot of AndroSensor readings
*)If Earth would spin faster we would have less gravity, and total SNR would not change?
Blog post by: Jussi Collin