Accelerometer Calibration and Digital Sensors

If you’ve used analog accelerometers before, then mV/g or mV/(m/s²) are familiar units for sensitivity. But what if the sensor itself is digital? The oncoming wave of digital accelerometers will have the digitization built right inside the sensor. This means that rather than outputting an analog signal proportional to acceleration, the sensor outputs digital samples of the acceleration sensing element’s analog output. Before the signal reaches any external equipment, the measurement is already digitized.

Consider a typical analog sensor with a calibrated sensitivity in mV/g. To process this signal for use, the sensor is connected over a cable to an external acquisition system, which performs the analog to digital conversion. This step introduces more error – noise and the uncertainty of the digital conversion. Remember, the sensor calibration sheet is just that – a calibration of the sensor only. To obtain precise calibrated engineering units, you need to account for the sensor calibration, the digitization process, and any added uncertainty during the process. This is where the benefits of digital accelerometers shine – by calibrating the digital output, we calibrate the entire system at once, encompassing the sensor, internal wiring, and acquisition. Additionally, the transmitted signal is digital, eliminating added analog noise effects of the cable.

We’ve chosen calibration units of (digital counts)/(m/s²), in SI units. The maximum digital count depends on the bit depth of the ADC. A 24-bit conversion has 2²⁴  possible values, or ±2²³. This unit is the most precise, directly describing the results of the digitization process. A resolution independent representation of this is (% of full scale)/(m/s² ) or in English units (% of full scale)/g. A full scale value of 100% is the maximum digitizable value before clipping, the maximum amplitude that the sensor can register. In counts terms, this would be 223 out of 223 for a 24-bit resolution. Full scale representation also more clearly hints at the sensor’s amplitude range. Consider that 5% (full scale)/g  means that every additional 1 g of acceleration uses another 5% of the total digitization range. This example implies that the maximum measurable amplitude without clipping is 20 g. 

By using the USB standard to transmit digital samples, a user gains the additional benefit of having easily accessible TEDS-like self-identifying and calibration information via USB descriptors. Furthermore, adapting the USB audio device standard allows plug and play integration without custom drivers. Fully featured software can automatically detect the sensitivity and output calibrated engineering units.