A couple of weeks ago a fascinating video showed up on YouTube where a clever pilot flew his quadcopter with 4 different propeller blades attached to it. Amazingly, not only did it fly – it flew great! You can check out this video here:
Let’s Try It!
For my own version of this test, I equipped my Bolt 210 with a few different props. They were:
- KingKong 5040 Tri-blades
- RaceKraft 5051 Tri-blades
- HQProp 5040 Quad-blades
- Gemfan BN5045
Even after seeing the video, I was skeptical. The first time I lifted the quadcopter off the ground with 4 different props installed, I held it in a hover for a minute with only minor drifts to make sure it wasn’t going to “flip out” on me. Just like in the video – I had no problems. After that, I started doing flips and rolls. I was amazed with how smooth the quadcopter still flew – I could not notice any difference between this set-up and any other tri-blade configuration (except for the distinct noise the quad blades make).
I then attempted a full power climb out and started doing laps around my yard – still no difference. It’s pretty shocking how well this thing flew. No vibrations, no kickbacks, no unexpected deviations from the commanded flight attitude. I did notice that when I really hammered the throttle in a turn, the quadcopter woult tilt ever so much in one direction for a split second before correcting itself. It felt like what I’ve experienced in the past when a motor was going bad. I think it’s safe to say I could fly a race like this without any problems. That is pretty cool to know – if you’re ever at a race and run out of props, installing any old prop from your neighbor is always an option in a pinch.
Once I set the miniquad down I eagerly took it inside to check out the Blackbox logs.
Looking at the PID error tab you can see that the flight controller isn’t struggling in the slightest to keep the quadcopter under control with the mixed props. The throttle values are rarely the same among all four motors but the PID error, which measures pitch, roll and yaw deviations from command, is no worse than on any “normally configured” quadcopter I’ve ever flown. It is the flight controller’s responsibility to set the throttle level of each individual motor so that the thrust it puts out matches the thrust of the other motors in level flight. It does this by a form of dead reckoning encoded in the logic of the PID controller. This mechanism does not care when different props, motors or ESCs are used on different arms – the dead reckoning figures out the throttle level required and sets that throttle.
What’s really fascinating to me is the prospect to use logs like this to measure the performance of a set of props or motors relative to each other. For example, in a level full power climbout, the quadcopter will be limited by the propeller with the lowest thrust. The flight controller will automatically throttle back the motors driving the props generating less thrust to keep the quadcopter level. From the magnitude of this “throttle back” you can get a sense for how one prop / motor compares to the others in flight. Here are two punchouts from the blackbox log we recorded with the Lisam 210:
Let’s clean that up a bit in a table, lowest throttle value to highest:
|Motor 2||RaceKraft 5051×3|
|Motor 3||HQProp 5040×4|
|Motor 4||Gemfan 5045BN|
|Motor 1||KingKong 5040×3|
Look familiar? Our prop speed test for the RaceKraft 5051 review showed remarkably similar results – albeit for speed instead of thrust. It once again reflects the surprising effect blade pitch has on thrust when a quadcopter is actually flying. I found it remarkable that throughout the log, the RaceKraft props were always using the lowest throttle value by a large margin. I really think there is something here to measuring performance in this manner.
The same test could be performed, albeit less accurately, in fast forward flight. This could potentially be game-changing because until now we have only been able to do static thrust tests with propellers – which do a horrible job of testing how props actually work. I will definitely be looking into this more in future comparison articles.
If you are interested in checking out the logs I analyzed for this article, you can download them here.