Radio failed triggered failsafe, but it just fly away

I try to fly my quadcopter a little higher today, and it triggered failsafe.
But instead return to home, it just stay where it is without descent or return to its armed location.
After about 10m, the battery went out, it crashed.

I got the log but dont know what is the problem. It is REALLY appreciate that anybody could have a look at it .

Thanks in advance!

You had quite a poor GPS lock, only 6-7 satellites and an Hdop of 2.39-2.68.

Might have been at least part of the problem as it’s recommended to have >9 satellites and an Hdop of less than 2.00 preferably 1.50

[quote=“MarkM”]You had quite a poor GPS lock, only 6-7 satellites and an Hdop of 2.39-2.68.

Might have been at least part of the problem as it’s recommended to have >9 satellites and an Hdop of less than 2.00 preferably 1.50[/quote]

Thank you for the reply. This seems one of the problem but I don’t think this is the major one.
Because the the GPS position for launching site and crash site is quite precise on the map. And the logged route in the sky is not that bad.

I really need to why this happened and how to prevent it next time…

In Mission Planner, what are your settings for the failsafe parameters? If you want it to RTL when it hits the fence, you have to program that into the fence settings.

Just gone through your log again.

Line 410 is where the quad takes off and you start flying around in stabilise.

The problem is that up to that point you only had 5 satellites and an Hdop of 3.94, which means the quad didn’t know where home was.

When you hit radio failsafe it went to RTL and started flying around aimlessly due to a poor number of satellites, a high Hdop and not actually knowing where home was.

I would suggest you arm the quad in Loiter mode as this will then do a GPS pre-arm check which makes sure you have a good enough satellite lock. Once armed, then you can switch to stabilise and take off.

[quote=“MarkM”]Just gone through your log again.

Line 410 is where the quad takes off and you start flying around in stabilise.

The problem is that up to that point you only had 5 satellites and an Hdop of 3.94, which means the quad didn’t know where home was.

When you hit radio failsafe it went to RTL and started flying around aimlessly due to a poor number of satellites, a high Hdop and not actually knowing where home was.

I would suggest you arm the quad in Loiter mode as this will then do a GPS pre-arm check which makes sure you have a good enough satellite lock. Once armed, then you can switch to stabilise and take off.[/quote]

Thank you for the analyze. That sounds reasonable ,but I still have some question.

  1. Is there any limitation for choosing the home location? For example the quadcopter will not use any GPS data whose Hdop is above 3 or something like it?

  2. How can I tell whether GPS data is good enough for RTL before launch? I have a GPS indicator led from the APM, I always wait for that led to be solid before I take off.

  3. Is there any failsafe for RTL mode which no home is located? I think we should add one.

  4. After review the log again yesterday, I have a guess myself. After the radio failsafe, the quadcopter began to generate NTUN log. According to that log, the quadcopter was trying to get a position at (2371.761,-7944.613). I am not sure what does this mean, but I think that is a position at 83m away in some direction which is quite good on the map.

But the quadcopter never made it. So I guess the quadcopter tried to return but failed fighting with wind. This is only a guess so I still need some more data to support or deny it. Could you tell me where to check for example the output power of motor or flying angle while returning?

  1. The apm set’s the home position when it gets armed. So if you have a gps coordinate that is 50mtr or 50miles away due to low sat count or high Hdop it will still use that coordinate as the home point. Though you say it shows a good start and crash point on the map so it sounds like it knew where it was supposed to be going.

  2. If you try arming in loiter mode it uses the GPS pre arm check to make sure it’s good. If you can’t arm in loiter then don’t try any flight mode that uses GPS. If it arms in loiter, switch to stabilise and fly, or leave in loiter and fly.

  3. Not really, the apm will always get a home location when arming it’s just that that location could be different to where you actually are.

  4. Ntun logs are created when flying in an auto mode (loiter, auto etc something where the apm is in control).
    What they do show is that the DPos and Pos (DesiredPosition vs ActualPosition) match well on the Y axis but not on the X axis so the quad wasn’t doing entirely what the Apm was asking it to.

Under the Att selection there is Des(ired)Roll and Des(ired)Pitch vs there actual values and it seems the quad is constantly struggling with forward motion, it can’t get the forward pitch angle that the apm is asking for. Under Ctun it shows that the actual altitude is matching the desired altitude as well. You would need to enable full datalogging in order to get the motor outputs though.

The correct answer to this one is that you most likely have issues with CG being too far forward or too low, and your angle_max is set too low. Center your CG as much as possible, and also set RATE_PIT_IMAX and RATE_RLL_IMAX higher (2000-4500), then do an auto trim flight, then do an autotune with no payload and the copter at its minimum flying weight.

The navigation system never failed, but this copter just isn’t in control like this.

[quote=“jschall”]The correct answer to this one is that you most likely have issues with CG being too far forward or too low, and your angle_max is set too low. Center your CG as much as possible, and also set RATE_PIT_IMAX and RATE_RLL_IMAX higher (2000-4500), then do an auto trim flight, then do an autotune with no payload and the copter at its minimum flying weight.

The navigation system never failed, but this copter just isn’t in control like this.[/quote]
Thank you!
I think this would be the answer. btw , what does CG mean?

Center of gravity.

I attached my gopro camera to the quadcopter that day. I think this made the CG a little forward than usual.
Last time I attached camera before the crash, I think I can handle the quadcopter good enough, so I did not realize this could make that different.

So here is question

Should I stop switching between no-camera/camera mode while I practicing? For example I should practice without camera till I am satisfied with it and then attach the camera, do a tweak and never take the camera off after that? Or is there any param that I can change to make the quadcopter adapt with or without the camera?

I am really new to this, any suggestion will be appreciated.

Increase RATE_PIT_IMAX and RATE_RLL_IMAX to 2000-4500, do an autotune without your camera attached, change max lean angle to at least 30 degrees.

Thank you again, I will try it after I fix my quadcopter