motors not stopping with zero throttle after flight

[Andrew Chapman]

Hi everyone, I had a bit of a scare with 2.9.1b today, after landing the motors wouldn't stop spinning, which of course set in my OH GOD ITS OUT OF CONTROL!!! instincts...

I tried trimming the throttle down further too, no effect.

You can see in this graph that THR_OUT increases relative to THR_IN when I first spool up (THR_MID is set to 720), and it seems to be an additive rather than multiplicative increase, so when I drop the throttle to zero after landing I still have THR_OUT at >130 and high MOTOR outputs:



I started to try disarming but it looked like it was going to yaw on the spot, so I reached between the spinning props to pull the power. Not the safest operation! In hindsight I probably should have held it against the yaw building up and tried to disarm.

So it seems something is a little amiss in the THR_IN/THR_OUT/THR_MIN/THR_MID logic - has anything changed there or is this likely to persist in 3.0?

AC.

[Jonathan Challinger]

Do you have a tlog of the flight? Do you know what mode you were in?

--
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/groups/opt_out.
 
 

[Andrew Chapman]

I was in AUTO for most of the flight, I *thought* I was in STAB for landing but I just double-checked the logs now and it looks like I was in LOITER, I mustn't have flicked the mode switch fully back.

Wow. I wouldn't think of deliberately landing in anything buy STAB, and I'm surprised I didn't notice the difference at the time - repositioning in LOITER in 2.9 can be a bit of a pig, and I took it down from quite a height and pin-point landed it on a small platform in a strong wind.

Good call Jonathan, thanks.

AC.

[Marco Robustini]

Hi Andrew, you can't disarm motors after automatic autolanding if the accelerometers goes bad due landing too heavy, after a few seconds that the quad is on the ground the accel "relax" and autodisarm is ok.
I had already reported this thing to Randy, but surely if you switch to Stabilize you're able to shut down the motors.
In case of LAND due failsafe and accel too shaken there's no chance to disarm them, you have to wait for the code to do it alone (and I never like this).
It would be very useful to be able to do it by force, for example by giving an "emergency command" for this kind of problems, such as "yaw and ale" to full left position for x seconds (like 5) for force disarm.

Marco

[Olivier ADLER]

If landing has been correctly detected you should have EV = 18 (DATA_LAND_COMPLETE) in the APM log.

Marco, a kill switch on CH7 or 8 has been requested in the issue tracker. This will be even better than a stick disarm because it will be instantaneous.

I did ask for kill + parachute as well on Ch7 and 8, so that we can have a simple way to manually control a parachute. This is interesting when flying in areas where safety is a primary concern.


Olivier.

[Marco Robustini]

Yes Olivier, a kill switch is a good idea but dangerous, if a user switches accidentally it's a sure crash.
Anyway i agree, with all my 3d heli i've the know "throttle hold" in a switch.
Personally I would prefer a combination of CH7 or CH8 kill switch + "yaw to left", reduce almost to zero the possibility of being wrong.
What do you think?

Marco

[Robert Lefebvre]

Are you planning on using the CH7/8 as a *disarm* switch, or a motor off switch?  When I proposed it, I was thinking it was motors-off, not disarm.  Making it into a disarm is a bit dangerous.  Do that in the air accidentally, and there's no hope, even if you switch back quickly.

I would leave it as a motors-off, but not disarm,  If you want to disarm, you can stop the motors, then do the disarm command.

[john...@gmail.com]

I still feel that zero throttle should kill motors regardless of flight mode. It's the only "kill switch" that you instinctively have time to do if there is a serious problem. I don't care how smart the autopilot is, I still want the final say if the propellers should spin or not.

- JAB

[Olivier ADLER]

On the other hand such a kill switch is dangerous, can cause ESC resync problem if used during fast descends.

I prefer to have a red button on my RC transmitter, actually it is only an idle / throttle cut switch, (does not work in all modes), but i will change it to the CH7 / CH8 kill function if we decide to put that in the code.

Something important is too choose a transmitter switch design and do not change it after. If you change your transmitter switches location from time to time, it is dangerous because when you really need a function in a hurry you can mix up things.


Olivier

[Olivier ADLER]

Marco, obviously we can have both types and it's not much complicated to do that in the code :


- CH7 - CH8 kill switch with PWM > 1600 = motor cut working in all modes

- CH7 - CH8 kill switch with PWM > 1700 = motor cut + disarm working in all modes

- CH7 - CH8 recovery switch with PWM > 1800 motor cut + disarm + parachute output working in all modes



Like this users can choose the right type according to their safety level needs simply changing there end point adjusment for CH7 or CH8. First solution seems enough in most cases.

Another advantage of doing it like this, is that you can have both function (kill motors or kill + parachute) on the same channel, keeping the other one for Simple mode or waypoints saving for example.


Olivier

[Robert Lefebvre]

Nice idea.  The only thing is that this is operating sort of backwards from the way I have it working now for Helis.  You must switch the channel high to get the motors to turn.  Low shuts them off. 

This is because it is not just a kill switch.  It's a run switch too.  You can arm, but the motors will not turn until the switch is high.  This reinforces in the pilots subconscious, exactly what the switch does.  You must raise it to run the motors, and lower it to kill them.

I also agree with you that I don't want the motors stopping in flight every time the stick is on the bottom.  I have had this happen a number of times accidentally, and it's never a good thing.  This is why I created my kill switch in my transmitter.

[Olivier ADLER]

Yes Robert it's working like this on my RC too.

This is just a matter of reversing your RC channel in the transmitter so that it is adapted for PWM = high for motor cut.

At the code level, to get it safe, we'll need to pool the channel 7 or 8 value instead of just reading a channel change event (must detect the right state regardless if we have a change or not).

Olivier

[Robert Lefebvre]

I didn't mean phyically up/down on the switch, I mean for the PWM.  I guess it doesn't really matter, but in my mind it's more logical where Low = Motor Off.  High = Motor On.

I could flip it in the Heli code to be the opposite, but is set up now where Low = Off, High = On.

[James Collings]

i also feel very strongly that throttle at zero should stop the motors no matter what flight mode.  its very simple and is exactly what any user would expect in stabilise, so extending this to all flight modes i think would be intuitive and logical.

(i am still here, not contributing much, but this is one of my personal gripes with arducopter)

ragards,

james

[Robert Lefebvre]

I'm extremely strongly against this.  Why do you want an invisible portion of your stick travel to result in insta-crash?  The problem is, you don't know when you reach the bottom, until you reach it, and then in many cases it's too late to save it.  I've had a number of crashes due to this.  Probably 5.  I've never had a situation where I thought "gee, I wish I could shut the motors off in Alt Hold".  I have had once or twice where I forgot I was in Alt Hold... but that is not the fault of the throttle stick operation, it's a fault of forgetting.

If you want 100% solid motor shut-down, put it on an independent switch.  This is what people have done since computerized radios were invented.

[Andy Newman]

I am also against having the motors shut down in all modes.  I use zero throttle for fast controlled descents in alt hold and loiter.  It would be too awkward juggling the throttle to get the fast descent in a controlled manner without stopping the motors.

Andy.

[Olivier ADLER]

I'm fully with Robert on this one. Most mid-range RC gear do allow to setup a throttle cut / motor idle switch. And we'll certainly have CH7 / CH8 motor cut for version 3.1. So every users should be able to find the best setup through those two solutions with a preference for the CH7/CH8 motor cut that will work with all flight modes.

If some users want to do it the unsafe way, cutting motors with throttle low then it's always possible to give them a parameter to do it. But obviously it's not a safe solution.

Technically i think that we must not allow that because we are dealing with ESCs and BackEMF sensing brushless motors. This mean that synching can be done reliably only when the motor has enough rotationnal speed. To start the motor, the ESC need a special slow start sequence without backEMF sensing, If a user put the throttle low in a panic, this is cutting the motor, and it is very possible that he could rise again the throttle still in panic, asking to the ESCs to resynch during a period where the motor is rotating very slow (without BackEMF sensing), but not fully stopped.

In this case, the ESC will send the motor start sequence, but the motor is not fully stopped. This can create a phase conflict and the associated motor freeze. This is worst if the motor is spinning in the reverse direction because of autorotation after a copter drop.

Some ESCs are more tolerant, other ones are very intolerant to this problems, and high performances firmwares like SimonK are well known for resynchronizing problems. But only ESC with coder inputs are fully blind to this problem. And we don't have that on copters.

I feel that many crashes have been induced by this specially with beginners that can panic very fast and do not have a great control of RC sticks specially for the throttle one. More Arducopter must deal with and help a lot of beginners.

For this reason i think that it's better to never allow stopping motors with throttle low. This gives an imaginary feeling of safety, not more and can create more trouble than save situations.


The true safety is a red (yes really it must be red using a red shrinking tube) button on your RC transmittter that will cut motors regardless the mode. I did it like this since my first APM 1.4 after thinking a lot about this problem and i will never change that.


More, because the motor cut switch (CH7 or RC throttle cut) will have priority when flying with two RC radio in teacher / student mode, it is really easy to cut motors if something goes very wrong without even taking back control with the RC student witch.


Speaking about safety, it's a good idea to put shrinking tubes of different colors on transmitter switches, to allow for direct visual recognition of switch functions.

For example i'm using those colors :

Red : motor cut

Yellow : RTL

White : Simple mode

Blue : Alt_Hold mode

Green : Auto mode

No Color : Pos_Old and Loiter modes (3 positions switch)



As a general rule it's better to use one switch for one function. This is easier to memorize. This is doable using flight conditions using a Gyro channel for CH5, something available on mid range RC systems. And still put RTL as the top priority flight condition.


Olivier

[Jonathan Challinger]

Thank you for saying what I was thinking, Robert.

I want to take this a step further: no more throttle cut in stabilize, or at least a parameter to turn it off. Min throttle forces the motors to min throttle. It may tumble but it won't stop the motors.

 This way it can be recovered instantly, otherwise it can't necessarily be recovered at all - hence the fall I had from 1000ft with PLENTY of time to recover, during which I was wildly sawing on the throttle to no avail. What if that fall had happened to land on a person or valuable property (car)?

It is objectively and provably so, so, so, so much more dangerous to have the motors cut out that I just don't understand why we do it at all. If a prop hits you at min throttle it might cut you. Oh no. However, if a copter lands on you while falling at 35 m/s and spinning at 2000dps, it's going to do worse. If a copter flips because a motor failed to restart and it hits you, there will be motors running higher than min throttle and it will do worse still. Even on top of that, it has further benefits - you know it's armed because the motors are spinning, and you know it's disarmed because the motors aren't spinning.

At least no stopping while armed should be the default parameter setting, and at most it should be hard-coded.

[Robert Lefebvre]

Can you imagine flying a gas powered plane, where, every time you accidentally dropped the throttle too far, the motor stopped?  No, that wouldn't work, that's why you typically have a low/high idle, and motor stop switch.  Safe, high idle in the air, low idle on the ground, and then throttle cut to shut it down.  Not saying we need 2 switches now, but just the 1.

And I'm not saying we do away with arming.  You still arm, but nothing happens until you raise that switch, and it won't arm unless the switch is low.  So you must arm with the switch low, then raise the switch, and now you have throttle, with no accidental mid-air shut-down.  If you switch down that throttle cut switch, it shuts the motors off, no matter what mode.  If you arm, but don't raise the throttle cut switch, it disarms after 10 seconds.  Same as it does now.

Very safe.  The pilot learns to know where that throttle cut switch is.  You can put red tube on it if you want.  On my DIY transmitter for my Octo, it's actually an LED lit switch.  Lights up when it's on.  The only problem I have now is that, I have a couple times landed in Alt_Hold, and switched the switch low, and the motors keep running until I put it in Stab mode.  This is ONLY a problem because of the way the code is now.  I have thought about making it force Stab Mode when the switch is dropped.  But I won't need to do this with Ch8 switching.

I've only had 1 accidental Ch 8 motor shutdown on my heli (again, this code is already functional on Heli), it was 100% pilot error, and there's no way to prevent it without making the system unsafe.  In a panic, I threw the wrong switch because I was using my Futaba.

[Jonathan Challinger]

Now it's personal. Let's get this arming situation fixed. I mean seriously, you can't even see the red led in sunlight.

[Robert Lefebvre]

What did you do exactly?

[DanielB]

Hi Robert.

"..every time you accidentally dropped the throttle too far, the motor stopped?"  IMHO, this is, however, exactly the point. 

If you've been flying a giant plane (have 2 with a bit over 14lbs AUW, with >100" ws for 3D) flyer (among other airframes, ..been at it since 1995), then that's exactly what you'd need, full down throttle to idle (about 1.5 k rpm) and a kill switch, ..override-priority on all modes.  In fact all our clubs here in (as far as I know) requires a kill switch (on board and a RC/TX activated) for obvious safety reasons.  So far last year, I had one tested with UDB4/MatrixPilot precisely because of its ability to override waypoints mode (auto mode, in APM terms) and any mode with TX input, so one can invoke TX-manual override, including an engine kill, anytime in case of any contingencies..

IMHO, the lack of this feature may be a showstopper for APM to be suitable for larger-much heavier airframes.

So I would recommend at least an all mode kill switch override feature as a must have option (perhaps as an advance parms option).

Just some thoughts.

Cheers.
Daniel

[Robert Lefebvre]

On the airplane side of things, yeah, absolutely it's a problem.  As far as I know. I don't use APM Plane, but I hear those guys talk about unintended ground starts.  They don't even have an arming proceedure.  They should also employ a Ch8 kill switch that will shut off the motor, no matter the mode.  And that could even be extended to allow the user to program in fixed servo positions, such as to cause it to go into a spin.  That would be optional, but would be the "get out of the air NOW" button.

For copters, cutting the power gets them out of the air pretty quick. ;)

[James Collings]

ok i understand your point of view regarding the insta-crash - though im sure you understand that im thinking about the situations where the crash is already assured and its all about damage limitation.

a dedicated kill switch achieves exactly the same without the alt-hold or loiter worries.  but along this route it seems to me it has to be absolute standard procedure, i mean, no exceptions to the process of starting and stopping the craft.

and so the question is this, could this replace CH7 funcionality or is it going to have to go on CH8?  in doing this are we setting a new minimum for the number of channels required to run arducopter?

regards

[Olivier ADLER]

Because it will be a new flight mode it will be possible to put it anywhere : Ch6, Ch7, Ch8 and even CH5 with a higher PWM value so that an auxiliary channel will not be wasted.


Olivier

[Robert Lefebvre]

It is absolutely SOP on my machines.  I can't arm with the switch high, and the motors won't spin up in Stab (my default mode) until the switch is high. So you MUST arm with it low, and then switch it high.  The only thing that doesn't work is that it won't shutdown in Alt_Hold or Auto.  I have to switch to Stab first.  But this is only because the code hasn't been implemented in Arducopter yet.

If you have a 6-ch radio, I'd think you could continue operating as you are now.  This will be an optional mode.  Or maybe the standard, but with the option to defeat it.

[john...@gmail.com]

Just remember to take into account that if you have a radio with less then 8 channels, the unconnected channels will still report as a valid signal with default value (center position) on the ppm side.

- JAB

[Maurice Barnes]

I am just seeing this thread in my research on motors shutting down in flight. I am running ArduCopter 3.1.2 and was transitioning from RTL to Stabilize (had so me GPS glitches) and my throttle was low.. When it switched modes the motors immediately went dead and the copter fell like a rock.. I tried to raise throttle but motors trying to start at that time couldn't save it as it flipped over in mid-air. Is this the expected behaviour? I really would like a low throttle idle motors as opposed to shut-down.

Regards,

Maurice

[Jason Short]

Properly calibrated ESC's should spin the motors at idle. If not, increase your idle speed in your parameters. 

Jason

For more options, visit https://groups.google.com/d/optout.

[Maurice Barnes]

Thanks for the reply jason.. The esc I have is the Dji opto 30. Which parameter would increase the speed?

Maurice

[Craig Elder]

This is the parameter

http://copter.ardupilot.com/wiki/arducopter-parameters/#Motors_always_spin_when_armed_MOT_SPIN_ARMED

[Maurice Barnes]

Thanks Craig..  I have tried that parameter all the way up to 100 without any change as well as the Throttle_MIN as well. None of these have any impact. Strangely enough if I flash with the PX4 firmware it works as it should.. Motors Spin on Arm but I cant get it to work with ArduCopter 3.1.2 and the latest build 3.2 Dev (03/11/2014). I have been trying to resolve this for weeks and since my crash this weekend I have decided to throw the issue out to the community. My RC3 MIN is set at 998 via calibration. Once I get over this hurdle I think all my problems will be solved.

Regards,

Maurice

[Maurice Barnes]

Correction That would be all the way up to the maximum value (I think its 300) not 100..

Maurice

[Maurice Barnes]

Just for completeness. I adjusted the throttle range on my transmitter to allow the motors to SPIN on ARM and now have my setup the way it should be. ( I was trying to not have to do this) I re-calibrated with a Throttle MIN at 1092 instead of 998 and all seems well.. I will do a test flight and confirm all is well. Thanks for the responses.

Regards,

Maurice

[j.go...@gmail.com]

I have to do the same to get spin on arm to work. But I just need a few units, like 3 or so. I think its because of the fluctuating throttle from my aurora 9.