PILOT_SPEED_UP having adverse effect in 4.1.5

ArduCopter Copter 4.1
1

I found some possibly undesirable behavior associated with PILOT_SPEED_UP in version 4.1.5 in loiter mode.
At least, it feels undesirable to me on my 7" freestyle quad. I was looking to increase the vertical climb rate, so I changed PILOT_SPEED_UP from 500 to 2000. It did not seem to have a large effect on the climb rate—and the quad is nowhere near its max climb rate, I can attest to that. It did, however, cause a massive overshoot on BOTH the ascent and descent to occur, after the throttle stick was released. At PILOT_SPEED_UP, 500 there was some minor overshoot, not particularly bad. At PILOT_SPEED_UP, 2000 however there was a huge amount of overshoot that would easily have caused a crash if I had been closer to the ground. Any advice on what’s going on here? Is this expected behavior?

I am attaching links to the params and logs for a flight where the only change is PILOT_SPEED_UP from 500 to 2000. Side note, probably irrelevant, I was flying in heavy wind (deliberately, as I was checking stability effects of various things during the session).
params for PILOT_SPEED_UP, 2000 flight, with the undesirable ascent/descent overshoot
log for 2000 flight, with the undesirable ascent/descent overshoot
params for PILOT_SPEED_UP, 500 flight, with fine ascent/descent behavior
log for 500 flight with fine ascent/descent behavior

2

Try increasing PILOT_ACCEL_Z. Think about what happens when UP speed is increased with the same acceleration…

3

Ah, I think I understand. Though it still seems like maybe not the best implementation–if the craft doesn’t keep up with my desired velocity due to the acceleration setting while I have the throttle away from neutral, that’s fine. But if I release the stick back to neutral, the craft should decelerate to zero velocity as rapidly as possible–as it does for the roll/pitch directions… At least that would be my desired behavior.

4

It decelerates at the same rate. As I have you will have to find a happy medium between Speed Up/Down and acceleration. I actually have hit the ground while playing around with these parameters. But I always test with a beater quad :slight_smile:

5

two questions then-- Is there an equivalent acceleration setting for roll and pitch while in loiter, that I haven’t noticed? And is there a problem with me simply setting PILOT_ACCEL_Z to something unachievably high, in order to get the maximum response physically possible from the quad?

6

Yea, there are Accel max, brake and jerk parameters for loiter. For accel Z I just tuned it like any other parameter, not sure what very high values will do.

7

On my 7" I had to increase PSC_JERK_Z to 40 to get decent climb/descent response.

1 Like
8

Looking at my 4" tuned for fast auto flight only increasing PILOT_ACCEL_Z worked for me but good point about PSC_JERK_Z.

9

well the problem is persisting. (I collected these data before seeing @juzzle1 comments so I can try that.) However it is still strange behavior, and can’t possibly be the desired behavior; changing PILOT_SPEED_UP and PILOT_ACC_Z from 500 to 2000 (both of them) caused the wild overshoot. At 500 there is slight but… normal? overshoot, but at 2000 there is HUGE overshoot. Maybe ~40-50 feet overshoot? With the more rapid acceleration, it ought to correct faster not slower. Any ideas?

PILOT_SPEED_UP, PILOT_ACC_Z = 2000 and way overshooting log
params
PILOT_SPEED_UP, PILOT_ACC_Z = 500, very slight (ok) overshoot but slow… log
params

1 Like
10

chrikies I just upped the PSC_JERK_Z and that solved the problem. How many freaking derivatives do I have to mess with to get it to behave!? It’s not even in the same subset of parameters (PSC_ vs PILOT_, not to mention LOIT_). I genuinely appreciate the hard work of the developers, but this is a great example of some really non-user friendly organization just to configure some basic loiter up and down behavior. Anyway thanks for the assistance, you all hit the nail on the head regarding my issues

2 Likes