›

Data buffering - temperature accuracy - timestep

ErikaErika
in General

I have been testing a MetaMotionR and I would like to use it in different projects but I stumbled upon some problems. Can you please help me with the following questions?

It seems to me that data streaming immediately quits whenever connection is lost. Is there a possibility to buffer data? If I would write an app that would automatically look for reconnection, would it be possible to buffer data in the gaps of connection?

In the datasheets, it states that the accuracy of the temperature sensor is ±5°C. Is it possibly to improve this accuracy with another temperature sensor or will it be limited by the self-heating of the sensor? Is there an offset temperature known?

I also noticed that the timestep between measurements changes inconsistently, resulting in missing timestamps. For example, I made a measurement of ambient light at 10 Hz which counted 5564 samples, so you would expect the last sample to be at 556.4 s. However, the elapsed time was 556.554 s. Does this indicate that we should have measured 5565 samples? In longer time periods, this will result in missing lots of data. Can you explain me why the timestep is inconsistent and how we should deal with this? How would you define a running time window to calculate features if there is no consistent sampling rate?

Comments

  • EricEric

    @Erika said:
    I have been testing a MetaMotionR and I would like to use it in different projects but I stumbled upon some problems. Can you please help me with the following questions?

    It seems to me that data streaming immediately quits whenever connection is lost. Is there a possibility to buffer data? If I would write an app that would automatically look for reconnection, would it be possible to buffer data in the gaps of connection?

    You can program the device to log data to the on-board flash memory

    In the datasheets, it states that the accuracy of the temperature sensor is ±5°C. Is it possibly to improve this accuracy with another temperature sensor or will it be limited by the self-heating of the sensor? Is there an offset temperature known?

    Those values are the extreme temperature differences. Compare the data gathered from the MetaMotion's thermistor and Bosch env sensor with another temperature measuring device. I highly doubt you will see much variance, if at all.

    I also noticed that the timestep between measurements changes inconsistently, resulting in missing timestamps. For example, I made a measurement of ambient light at 10 Hz which counted 5564 samples, so you would expect the last sample to be at 556.4 s. However, the elapsed time was 556.554 s. Does this indicate that we should have measured 5565 samples? In longer time periods, this will result in missing lots of data. Can you explain me why the timestep is inconsistent and how we should deal with this? How would you define a running time window to calculate features if there is no consistent sampling rate?

    If you want accurate timestamps, you will need to either log the data or use the packet accounting feature. Streamed timestamps as is are always inconsistent as BT packets do not arrive at consistent intervals.

    If you are worried about losing data, then compare "# of samples / elapsed time" with the selected data rate. In your example, you would have a -0.2% error from 10Hz.

This discussion has been closed.