Recursion Labs - Sub 250g Motor and Propeller Testing

There is very little community information available on sub 250g Quadcopter propulsion systems, and choosing the correct motor/propeller (and corresponding battery that can keep up) can make or break a build. I've decided to put a considerable amount of my free time into testing, gathering detailed metrics, and reporting on different motor and propeller combinations. I've started testing with 3" class quads, but will venture into 3.5", 2", 4" and eventually see if I can find suitable motors and propellers for sub 250g 5" builds. So far my testing has tremendously helped me make my quads fly better, and I'd like to help others do the same. I invite you to join me in discussing these tests, what you'd love to see tested, and assist with analyzing the results.

Test Methodology

I've built code to fully automate each test, so that the results would be the exact same (as close as practically possible) if I run the same test over and over again. In addition to the automation, in order to achieve my goal I had to eliminate major variables common to motor testing. These variables are input voltage, and heat. When most people run these types of tests, they power the motor with a fully charged lipo, and slowly ramp up the propeller from 0 to 100% throttle, gathering data. The problem with this methodology is that their is no practical way to ensure the voltage is the same for each throttle position between tests, due to the dynamic (based on remaining capacity) sag of the voltage from the lipo. As the motor spins up it generates heat, which adds resistance, significantly impacting performance. With smaller motors, heat becomes even more important, because they can heat up extremely quickly during static thrust tests. To eliminate the voltage sag, I power the thrust stand with bench power supplies, calibrated to the correct voltage (within 0.01v), which maintains a constant voltage regardless of load. The only voltage sag is from the wire to the stand (same size for each test), and the motor itself. Heat is eliminated as the automation bursts the motor to the correct throttle point being tested, in increments of 3% for the entire throttle range being tested. Between these tests, the motors are idled long enough for them to cool off (verified using a thermal camera). Before these variables were eliminated the tests results were extremely inconsistent, but are now extremely repeatable.

I made a video that goes into the methodology in more detail here:


If you prefer to read more about it instead of the video, I wrote a fairly detailed article for Tyto Robotics where you can learn about the methodology, and what problems I'm solving by doing it this way:

Feel free to ask any questions.

Results

Once the testing is conducted, all of the results are compiled into a CSV/spreadsheet which has statistics for each throttle position tested. These statistics include grams of thrust, voltage, current, measured motor RPM, watts, and grams of force per watt. I take this data and build charts to help visualize the data, and formulate my own observations, conclusions, and at times, come up with even more questions that require further testing.

Completed

Builds based on test results

Currently Working on

• 4" testing on 1408 2800Kv motors
  

There is very little community information available on sub 250g Quadcopter propulsion systems, and choosing the correct motor/propeller (and corresponding battery that can keep up) can make or break a build. I've decided to put a considerable amount of my free time into testing, gathering detailed metrics, and reporting on different motor and propeller combinations. I've started testing with 3" class quads, but will venture into 3.5", 2", 4" and eventually see if I can find suitable motors and propellers for sub 250g 5" builds. So far my testing has tremendously helped me make my quads fly better, and I'd like to help others do the same. I invite you to join me in discussing these tests, what you'd love to see tested, and assist with analyzing the results.I've built code to fully automate each test, so that the results would be the exact same (as close as practically possible) if I run the same test over and over again. In addition to the automation, in order to achieve my goal I had to eliminate major variables common to motor testing. These variables are input voltage, and heat. When most people run these types of tests, they power the motor with a fully charged lipo, and slowly ramp up the propeller from 0 to 100% throttle, gathering data. The problem with this methodology is that their is no practical way to ensure the voltage is the same for each throttle position between tests, due to the dynamic (based on remaining capacity) sag of the voltage from the lipo. As the motor spins up it generates heat, which adds resistance, significantly impacting performance. With smaller motors, heat becomes even more important, because they can heat up extremely quickly during static thrust tests. To eliminate the voltage sag, I power the thrust stand with bench power supplies, calibrated to the correct voltage (within 0.01v), which maintains a constant voltage regardless of load. The only voltage sag is from the wire to the stand (same size for each test), and the motor itself. Heat is eliminated as the automation bursts the motor to the correct throttle point being tested, in increments of 3% for the entire throttle range being tested. Between these tests, the motors are idled long enough for them to cool off (verified using a thermal camera). Before these variables were eliminated the tests results were extremely inconsistent, but are now extremely repeatable.I made a video that goes into the methodology in more detail here:If you prefer to read more about it instead of the video, I wrote a fairly detailed article for Tyto Robotics where you can learn about the methodology, and what problems I'm solving by doing it this way: https://www.tytorobotics.com/blogs/custo...ethodology Feel free to ask any questions.Once the testing is conducted, all of the results are compiled into a CSV/spreadsheet which has statistics for each throttle position tested. These statistics include grams of thrust, voltage, current, measured motor RPM, watts, and grams of force per watt. I take this data and build charts to help visualize the data, and formulate my own observations, conclusions, and at times, come up with even more questions that require further testing.