No announcement yet.

Battery burn down. What does it do?

  • Filter
  • Time
  • Show
Clear All
new posts

  • Battery burn down. What does it do?

    There has been some discussion on the forum about the relationship between battery power and robot performance. Running the robot a few times after changing batteries or charging to burn down the peak power has been used in FLL since the beginning. My teams have never done it, but it is fairly common. I think it unlikely that battery power affects anything other than motor performance, but opinions without data are just that, opinions. This is the first in a series of experiments that will try to find any relationship between battery power levels and sensor readings or program operation.

    Experiment 1 is a static sensor test. I started with a static test because it directly addresses some power/calibration claims and it is easy to perform. I have a robot with an NXT light sensor, EV3 light sensor, gyro sensor, and EV3 ultrasonic sensor. I wrote a program that recorded sensor data for 1 minute then ran all 4 robot motors for 5 minutes. I charged up the batteries to full charge and ran the program until the EV3 shut down. It took 46 tests to fully drain the batteries (4 1/2 hours of continuous operation). If the "burn down" practice helps deliver more predictable sensor readings I would expect to see the static readings change during the first few tests.

    NXT Light Sensor

    The NXT light sensor was the only sensor that showed and sensitivity to battery power.


    From the first trial (fully charged) to the second (after 5 minutes of motor operation), the reflected light intensity increased 1%. ONE PERCENT! After the initial "battery burn down" all remaining tests had the same sensor mean. Even the very last run that was completed before the brick powered off when the motors were turned on. Of all the sensors this is the one I thought most susceptible to drift because it is a purely analog device.

    EV3 Color Sensor

    The EV3 sensor is more stable than the NXT. The intensity ready didn't change at all from the first test to the last. This could be because of the built in ambient light compensation or maybe just the reduced resolution of the sensor value returned by the EV3 color sensor block.

    The EV3 sensor bounces up and down 1%, but the frequency of the bounces doesn't have any relationship to battery power. Trial 45 had no bounces and trial 7 had four bounces. If your light sensor readings are changing run to run look for a cause other than battery power.

    EV3 Ultrasonic sensor

    The US sensor is the most stable. This makes sense since the sensor uses a clock to measure the time between sensor ping and the echo. Of the 46 trials only 3 displayed a bump like Trial 23 in the plot above.

    I think the test results indicate that battery power has little effect on sensor readings. Future experiments will look at the US and gyro sensor in dynamic situations, and verify that light sensor range is unaffected by battery power.
    Attached Files
    Last edited by Dean Hystad; 10-25-2018, 09:01 AM.

  • #2
    So, would using 6 Lithium Iron Disulfide batteries (10v, 3500mAh) -- twice the capacity of alkaline and 50% more than LEGO battery packs -- be of no utility, if legal?
    Last edited by winklestork; 11-30-2018, 05:22 PM. Reason: Copy and paste failed.


    • #3
      If they are AA format they allowed by the rules. The challenge document doesn't specify any restrictions on voltage or chemistry, just format. With 10 volts your robot could go really fast, faster than 100% speed if you used the unregulated motor blocks. Your attachment motors would be a little stronger.