Idea - Autotune for MOT_THST_EXPO

ArduCopter Copter 3.4
1

Following the recent changes & refactoring of the motor libraries I’ve decided to measure my thrust curve to correctly find the value, instead of using the default - which by the way was very close to the value I got ( 0.6801).

Since measurement it is a time-consuming task and ( i believe ) the expo variable a very valuable parameter for FC, I would kindly submit a suggestion of implementing an autotune routine for that.

Also, esc / esc fw / battery and cable impedance could play a role on the results ( drone vs bench ) , so If we could rely on the measurement to be done in the uav itself would be the best of both words.

For that, the expo autotune could do something like:

Parameters: Height min/max for the test ( ex. from 15 to 100m )
descend phase floor: ( ex. 30m )
Max. curr drown ( same as battery? )

  1. check if mot_thst_hover ( and all related ) values are sane

  2. check battery

  3. lock the vertical position in loiter

  4. Climb to height_min and start the test:

  5. test mot_thst_hover if it is really hovering, if so, define 7 point above to be tested, returning to its original position after every test.

  6. Based on the results, extrapolate the climb rate to a descent rate and then test three points below hover, stopping the test if crossing the height-min floor or an user-defined floor for that could be an “safety value”.

  7. Calculate the curve.

What do you think? Just an idea…

1 Like
2

Sounds interesting. I would suggest creating an issue on the issues list about this. With luck it may get taken up by the devs there

3

Thanks , done!

4

HI, I am interested in it also. I want to know how to measure the thrust values from the vehicle ? Any suggestion ?

5

I think I can test only one motor, Is that OK ?

6

Yes, assuming that your setup is the very same in all motors, that should work.
I’m afraid the devs didn’t liked my idea, no one has posted anything there… Anyway that was just an Idea.

7

Well, I had looked into your idea, mostly I assume that you measure the thrust by the climb rate, If the climb rate is precise in the flight(the vibration will influence that), it will work good. But I don’t think so. That is only my opinion.
By the way, How did you measure the thrust, some devices?

8

Yeah, is not that simple, but with a little bit of math i believe it could be done, just like Autotune. I have measured mine by drawing a curve of thrust vs pwm percentage. I installed the motor in a detached drone arm, held it with a hinge and below the motor i held it with a string attached to a scale. The scale was calibrated and was able to measure lift.

It wasn’t that accurate, but since the setup was the same, the curve should be so accurate as it would be with a professional thrust stand.

In the end, you should export the results to mathlab and it will do the math to give you the expo value.

9

Thank you, I will try it. Best wishes.