Problems in PosHold and AltHold with ArduCopter 3.2.1

[Luca Angelo Micheletti]

Hello Randy and Tridge.

Yesterday we had a problem with our drone and ArduCopter 3.2.1.
We flew mainly in AltHold and PosHold.
Towards the end of the flight we had a crash: not a real crash, but a quick descent to the ground with a progressive decrease of the motors.

From the logs we could see that, when the altitude began to decrease, the Throttle OUT has successfully begun to rise but not the RC OUT and then the motors.
Even increasing the Throttle IN we were able to correct this behavior.

You can see this from the logs and the image attached.

Do you have any idea about this kind of behavior? There may be a problem with the Throttle OUT and then the saturation of PIDs?

Thanks a lot for your help!

Best Regards

[Robert Lefebvre]

Hi Luca, your vehicle is quite underpowered.  Hover throttle was around 1800 to start, and then gets higher as the battery voltage drops.  At the end, one motor is pegged at full throttle.  The other 3 throttle output is being held back by the stability patch as needed to provide some reserve capacity for the other 3 to provide stabilization. If this were not done, the copter would have flipped over.

Rob

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Roberto Navoni]

Hi Robert,
i agree with you , but we are testing the drone in limit situation ...
not in standard situation , and only in that situation appear a
possible bug. I was there and i see what happened i was quite
incredule about it .
I think that is important to undertand why this happen. Could be
somethings related to PID on altitude hold
best
Roberto

[Robert Lefebvre]

Roberto, what happened is this.

The copter has unbalanced motors output, and is underpowered. Mot#2 is pegged, and Mot#3 could not increase without flipping it.

[Emile Castelnuovo]

I agree with your analysis Robert.

Seems there could also be a problem with motor1 as it is always higher than the others.

And definitively underpowered :)

E.

[Marco Robustini]

I agree with Roberto and Emile, at this pwm value it's impossible that the quad will remain stable, when working close to full pwm can happen a flip.

Marco

[Marco Robustini]

Sorry, i meant Robert...

Marco

[Roberto Navoni]

Hi Marco,
i agree with you but i disagree respect of final result ... If you
check in the log the drone start to descend you see the value of motor
go down .... and the battery was quite good ... I understand that the
gas in loiter was at 3/4 but no reason because the drone start to
descend .. the drone flip on the land not in air because had an hard
landing .
Best
Roberto

[Robert Lefebvre]

Roberto, the value of 1 motor goes down.  Another motor is at maximum power.  This is how quadcopter flight works.  It has to balance thrust or else it flips over.  Your thrust is very unbalanced, and one motor was on maximum.

If all the motors had increased thrust, it would have flipped over and come down much faster than it did.

[Marco Robustini]

I analyzed the log, still requires investigation into the code.
CH2 output go at the max pwm (and remains linear) at 37.65, the other three motors seek to compensate to keep the level, but then there's a sudden height drop while there is a command to nearly full throttle, there's something wrong, code ignored the throttle control, the quad was not supposed to fall so quickly.
My two cents...

Marco

[Robert Lefebvre]

The sequence of events seems to me to be that first Mot #2 reached it's limit.  Then the Altitude sunk below the Desired Altitude.  Then Throttle Out went to maximum, then Throttle In went to maximum but at that point there's nothing left to give.

[Roberto Navoni]

Hi Robert,
the drone was in position hold and the stick was in the middle when
the pilot see drop of altitude put the stick up but the drone contine
to go down no reaction the strange is that only mot 2 go up but the
other go down at 1200 - 1300 .. could be a problem on stability patch
or other kind of bug related ...

[Robert Lefebvre]

Did it roll over on the way down?

[Roberto Navoni]

The drone broke only one propeller on the land so don't crash hard ...
roll when was on land

[Robert Lefebvre]

The fact that it did not roll over on the way down, is indication that stabilization was working exactly as it should be, and the reduced motor output on 1, 3, and 4 was required to keep it level.

This really can't go two ways.  Either the controller was giving exactly the correct throttle to keep it upright, or it wasn't.  In this case, clearly it was giving the correct throttle.  Unfortunately, motors 1, 3, and 4 had to be reduced due to lack of power on #2. 

This is the entire point of "Stability Patch".  Once an output is saturated, either at full throttle or min throttle, then the other 3 motors must be used to balance the copter.  It was decided that it is always best for the copter to descend straight down rather than flipping.  If motors 1, 3, and 4 had been allowed to rise to slow the descent, it would have rolled over, as #2 was already running at full power.

[Al B]

Robert,

Is that "Stability Patch" part of the 3.2.1 release or is it part of the Master branch and targeted for the 3.3 release?

[Luca Angelo Micheletti]

Thanks Robert,
We will check if we have a problem with motor or ESC on output 2 and we will do the same tests with a different setup ( motors and propellers or payload)

[Robert Lefebvre]

The"stability patch" has been in place for a long, long time.  2-3 years, maybe even more.  There are some tweaks to it in 3.3.

On 16 April 2015 at 15:41, Al B <cagi...@gmail.com> wrote:

Robert,

Is that "Stability Patch" part of the 3.2.1 release or is it part of the Master branch and targeted for the 3.3 release?

--