Understanding the AX3 Settings

Accelerometer

The accelerometer has configurable sample rates, adjustable sensitivity and a low power mode. The sample rate affects both the maximum sampling period and the battery life. Higher sample rates reduce the amount of time the device can record for and use extra battery power during recording. The accelerometer has a low power mode to reduce the current consumption and increase battery life, this setting increases the amount of noise present on the accelerometer output and may not be suitable for high precision applications. The device is designed to sample for at least 2 weeks on a full battery charge at 100 Hz in the full power mode. The table below shows how adjusting the sample rate will affect the battery.

Sample Rate (Hz) Typ Battery Life (days) Low Power Battery Life (days)
400 5 5
200 9 10
100 14 18
50 21 31
25 34 47

The accelerometer has a variable sensitivity to allow it to be used in many applications. The selectable ranges are +/- 2 g, +/- 4 g, +/- 8 g and +/- 16 g (g is the acceleration due to gravity or ~9.81 ms-2). The user should experiment with these ranges to trade off sensitivity against dynamic range. Accelerations outside the selected dynamic range result in saturation (“clipping”) of the recorded acceleration. The dynamic range of the accelerometer has no effect on the battery life or memory constraints. Table 2 shows a typical use case for each of the available sensitivities.

Sensitivity Example Use Case
±2g Fine movements such as hand writing or painting
±4g Mild activities such as walking
±8g Moderate activities such as sprinting, jumping
±16g Severe activities such as boxing

Memory

The AX3 has a built in 512 MB NAND Flash memory chip to store the accelerometer data, this type of memory has a limited lifespan of 100,000 read/write cycles and over time will begin to develop “wear”. The AX3 has an inbuilt firmware algorithm to ensure the memory is “wear levelled”, this ensures the general condition of the memory wear remains constant over the entire range. The memory is typically used to store the raw binary data from the sensors as this offers the optimal data compression, error detection and error recovery. Most errors will be single bit errors (affecting a single accelerometer sample) and can in some cases can be ignored. There are two other possible means of removing these errors: error removal and error correction. Error removal discards the chunk of data that contains the error, this is very fast and can be performed on the final binary files after downloading them from the device. The chunk size in seconds depends on the sample rate, at 100 Hz this is 1.2s. The error correction method may be required by some applications where no data loss is acceptable. With this feature turned on the download time will be increased as every sample is checked for errors during the downloading step. Additionally, a battery life penalty of ~10% will be observed. Users are recommended to only use the error correction mode if data integrity must be guaranteed or longer download times are acceptable.

  • Download time of 512MB WITHOUT error correction enabled ~ 6mins
  • Download time of 512MB WITH error correction enabled ~ 11mins

Additional Sensors

The AX3 has built in ambient light level and temperature sensors. These can be added to the data stream to enrich the gathered data. The effect of adding these sensor outputs on battery life is low and, in most cases, there are no additional memory requirements. In the typical usage scenario, the sampled light level and temperature are added to every chunk of data written to the device memory. This allows ~1.2 samples per second at 100Hz accelerometer sample rate (this figure scales with the chosen accelerometer sample rate).

To use the raw temperature values the user must convert the ADC values into ˚C using the following equation:

T = (counts – 171) / 3.142

The light sensor on the AX3 is a logarithmic lux sensor that has a wavelength characteristic close to that of the human eye. Without calibration the sensor can only be used qualitatively due to part variations; That is, comparison between devices is not possible. This is mainly because the sensor output is effected by the light attenuation of its particular enclosure, but also due to the variability of the sensor component. Some level of calibration is essential if the output is to be used quantitavely and to allow comparison between devices. The calibration must take into account the non-linear aspect of the logarithmic sensor and this should use the equation below which approximates the lux level at the sensor surface. A reasonable calibration can then be achieved by applying a coefficient to the approximated lux to take into account the linear attenuation of the enclosure. When measuring this coefficient the user should ensure that they choose a calibration point in the middle of the device operating range and use an incandescent light source. A suggested operating point would be 1k lux. To use the light sensor values, the user must scale the raw ADC values according to the following equation:

Lux = 10(counts/341)

Battery

The AX3 has a high energy density lithium polymer cell which is used as the power source whilst gathering data. The lithium polymer battery is rated at 500 recharge cycles; battery empty to battery full. Battery charging is via a high performance charge controller that permits a full recharge of the battery within ~90 minutes. Once charged, the battery will keep the internal clock running for ~200 days, it is important to note that the battery is being slowly depleted during this time. Many applications will wish to use the scheduled sampling mode to automatically start the device recording once a specific time and date has been reached. If the start date is ~100 days in the future then the battery may only have sufficient capacity to log 1 week of data when it finally starts recording. An even lower power mode is available for device storage where the clock value is not retained. This mode is suited to applications where the devices are fully charged and then stored for a long period before deployment, in this mode the device will retain 90% of its battery capacity for ~100 days.

NOTE: It is always recommended to only deploy devices with a battery charged over 85%.

Real time clock

The AX3 has a built in, real-time clock (RTC) and calendar which provides the time base for the recorded acceleration data. The clock is typically reset when the device is configured (but the precision at the time of reset cannot be guaranteed). Over time this internal clock will drift slightly. For single device applications, that can tolerate a drift of 50 parts-per-million (0.18 seconds per hour) no action is needed. However, when using multiple devices for a single capture session, aligning the data sets from several sources can become difficult. To overcome this, you can add an easily identifiable signal (e.g. claps at the start and end of capture), and tools are available to help synchronize this to an external clock or between devices.