Delay between throttle input and altitude reached

ArduCopter Copter 4.5
,
1

Hello,
We experience “delay” when we have big throttle input on our quad.
We move up the throttle for exemple and release it, the drone still go up for about 1 sec before it stops.

UAV Log Viewer plot (2)
UAV Log Viewer plot (2)1022×502 79.3 KB

Logs : log_64_2024-9-9-15-05-00.bin - Google Drive

I know my PSC is not very well tune but I do not find the parameter responsible for this behaviour.

Therefore, MOT_HOVER_LEARN should be 2 by default if our payload change ?

The documentation specify as well that :
PSC_ACCZ_I to 2 x MOT_THST_HOVER
PSC_ACCZ_P to MOT_THST_HOVER

and actually we have almost :
PSC_ACCZ_I to 4 x MOT_THST_HOVER
PSC_ACCZ_P to 2x MOT_THST_HOVER

Is-it suspect ?

Thanks !

2

Yes.

How well has this vehicle been tuned?

3

It follows every step of your awesome guide :wink: and to be transparent, we are starting the “System identification for analytical PID optimization” step.

We have been doing test flights for 3 months now with.

4

Nice, you choose the fastest and safest way :slight_smile:

But the configurator automatically sets

  • PSC_ACCZ_I = 2 x MOT_THST_HOVER
  • PSC_ACCZ_P = MOT_THST_HOVER

How come you have double that? Did you changed it by hand?

5

One might think RC inputs translate directly in velocity commands, but in fact they go through a input shaping process, to make sure the provided commands respect the acceleration and jerk limits in the controller.

Now, this is probably an issue when there’s an unwanted delay when braking from a horizontal velocity, but I’m not sure if that affects vertical velocity as well.

Nonetheless, you might want to play with LOIT_BRK_ACCEL, LOIT_BRK_JERK, LOIT_BRK_DELAY and PSC_JERK_Z. But these will also affect braking on the horizontal plane.

1 Like
6

Looking at your graph, the Desired Climb Rate does start to flatten out as soon as you lower your throttle input (ThI). The desired climb rate doesn’t instantly go to 0 because that require a very large deceleration and jerk. If you go in the PSCD log structure, you can see the desired and actual velocity and acceleration.

I took the liberty of measuring the slope of the PSCD.DAD (Desired Acceleration in the Down direction) during the moment ThI reduces. The slope was 4.95, and PSC_JERK_Z is set to 5. So the rate of change of your acceleration (jerk) is currently limited by that parameter. If you increase the jerk limit, you’ll start decelerating faster. However, your next limit might be reaching the LOIT_BRK_ACCEL limit too fast (which limits the max deceleration you can achieve).

3 Likes
7

Check out this thread and the link in it
Position Z

2 Likes
8

any reason why you use Thrust Expo 0? T-Motor Flame ESC also no longer needed super low value.
Am I correct you are using a 15" propeller? if yes, your ATC_ACCEL_x_MAXs are not the default for 15".
My guess you have done a quiktune.

ATC_ACCEL_P_MAX,44000
ATC_ACCEL_R_MAX,46000
ATC_ACCEL_Y_MAX,10000
MOT_THST_EXPO,0
MOT_THST_HOVER,0.281947
9

Yes I did change this parameter myself, the drone wasn’t locked on the Z axe, nothing dramatic but slow stabilization at each perturbation.

10

Yes I saw my Desired Climb Rate had a good correlation with ThI. The issue seems to be on the “breaking part”

Oh I didn’t know about PSC_JERK_Z, good point. Thank, i’ll try. My LOIT_BRK_ACCEL is good for now so we will see.

11

That’s the value we found on our test bench. We have volonterly understimate it a bit, it could be 0.1.
We have 22" props, so our ATC_ACCEL_x_MAXs are not out of range, a bit lower than recommed by choice.
But you are right, these value could have an impact.

12

Thanks ! I didn’t read this topics.

13

These parameters are interrelated so some trial is usually required to get it the way you want. But, that’s the fun part!

1 Like
14

Then your ATC_RAT_xxx_FLTD, ATC_RAT_xxx_FLTT are not recommended values for 22" propeller.

The recommended MOT_THST_EXPO value for 22" propeller is 0.72. that’s a huge diff.

image
image1160×1073 38.4 KB

15

The recommended MOT_THST_EXPO value for 22" propeller is 0.72. that’s a huge diff.

For common ESC/Motors/Props yes but that’s only a recommendation base on experience. We know for sure that’s not the case for us. If your ESC have already a correction build in, no need for this value.

Then your ATC_RAT_xxx_FLTD, ATC_RAT_xxx_FLTT are not recommended values for 22" propeller.

Should be half of your INS_GYRO_FILTER which is 30, yes higher than the 22 recommended but our frame doesn’t seem to be noisy between 22 and 30 so no need to go that low.

But you are right, that’s not the default recommended configuration.
I could have missed something but for now we have good result with these params.