First of all, let me introduce my quad configuration and the crash details.
We have a Tarot T960 hexacopter configuration based on a APM 2.6. We have tested many times and it flies really stable.
This morning, we decided to do the first flight of a modification based on this frame, but in this case, just with 4 arms, in quad configuration. The controller in this flight test was a pixhawk with APM 3.2 version.
We were testing loiter mode in order to measure the time autonomy with a new battery ( Multistar 10000maH). It was stable and after 3 minutes of flight, something happened and drone fell free and crashed.
I have downloaded the log, but after analysing it, I can not figure what happened and which was the problem.
Please I need your help in order to determine which problems we had in order to not commit the same mistakes.
* I recorded the crash, so I will update the topic with the link of the video. Probably it helps in the analysis.
*I have attached the log file.
It does sound like, it could be a hardware fault. Did you check the battery itself for any lose wire? Also did you do a voltage reading of the battery after the crash?
Furthermore, what ESCs are you using?
ARMING_CHECK = 0
Warning: Pre-Arming Check is not Active (0=Off & 1=Fully On)
ARMING_CHECK = 0
FS_GPS_ENABLE = 1
Warning: GPS Pre-Arm checks are disabled, while GPS Failsafe are Enabled.
Warning: GPS could be unstable while the GPS failsafe requires it to be stable.
Warning: Recommend having both Enabled! Refer to Issue 1487 for more information.
COMPASS_OFS_X = -33
COMPASS_OFS_Y = 172
COMPASS_OFS_Z = 21
Warning: One or more compass offset is not set correctly (-150 to +150 recommended).
Carlos, I'm thinking mechanical failure. The log does not indicate any kind of software error in the autopilot. On the contrary, as it was falling the auotpilot was attempting to correct. I am seeing the current drop to zero right as the freefall starts. That indicates a power failure to all 4 motors... but the Pixhawk still recieved power. Very interesting...
Is the new battery you're testing significantly heavier than what you normally fly? Can you post the specs or links to the new and old battery?
Maybe PWM connector to motors popped out of pixhawk? If that happens, the motors turn off. It happened to me, thankfully while I was running the copter strapped to the bench. After that I gued the connector in. If you are able, definitely strap the copter down and move the throttle all around. It's a safe (so long as you do it securely) way to find potential problems.
Thank you so much Stephen Zidek for your clarification. As you said, the region with zero current at the first minute is due to the landing of the quad, I did a first flight in order to be sure everything was OK ( with no luck).
I agree with you and I think is mechanical failure because the log doesn't show any sign of software problem. About batteries... well, I used similar batteries in a hexacopter with same ESCs and motors, but it was the first time in this configuration and I wanted to test the flight autonomy on this platform. The battery is a Multistar 10.000 mah.
I took the quad to my house and it is still like the day of falling, so tomorrow I will try to analyse connector or burn cables. I think it could be useful and after analysing the log, we are ready to find the real problem. I won't forget to take pictures of damages :S
Thank you so much for all your support, I love this community!
problems with Quad have been already detected in the first minute of flight.
Current went down to 0 value (all motors gone temporary disarmed),
verified by AccZ went to -40, so 4 x G
Throttle totally inresponsive
thrOut went down to 0 in 1st minute
you could have saved your Quad setting proper failsafes to ground your quad if current went to 0 in the 1st minute.
Not sure while Kalman filter failed to safe your quad.
On another Current down to 0 even ( motors disarmed) throttle failed to go up again to safe your quad.
I need your telemetry log to study if the problems are not due to low radio signal .
Check your battery and power distribution for burnout connectors.
check your battery, connectors and power distribution wiring.
Darius, you still have a lot to learn about logs before you hand out advice. According to the log, Carlos took the copter up, then landed it. The region you see with zero current draw, and with the bump in accZ, is due to Carlos landing his copter. An instantaneous spike of accZ down to -40 does not indicate freefall. However, Carlos did take the copter back up and park it in loiter mode. During loiter, after a few minutes, the altitude drops sharply and the accZ floats near 0. THAT is the sign of freefall.
There is no failsafe setting for if current draw goes to zero. I can't see a scenario in which such a failsafe would be useful.
You can check under the RAD tree on the log browser for radio strength. Carlos' signal strength was fine. There is also an error tag for if radio is lost. Please read the documentation.
An instantaneous spike of accZ down to -40 does not indicate freefall.
who claimed otherwise ?
AccZ = -40 = 4 x G, so 3 xG up in the sky.
How is going up vs. landing ?
Current = ZERO failsafe is clear in case of burnout, battery issue, as reported.
This is exactly the Mission Termination Command invoked by Mission Termination Unit.
At low altitude, drone can still be nicely grounded if ESC has motor braking set off.
What was seen on video is drone's free drift with all motors unpowered (disarmed).
How to save ESC PWM value to log file for every ESC to calculate RPM for every motor ?
I opt for All RPM technology QuadPilot to stop crashes and power down in midair.
Can Pixhawk support ESC with RPM sensor ?
Freefall resets AccZ to ZERO value.
had similar problem and was the ESC programming. they cut off some motors due to low battery. this esc feature needs to be disabled. you have it disabled on all esc?
Carlos, adding to what cala and Michael are finding I see that current goes down to 0 quickly at the same time. that may be because AP is cutting some motors to stabilize, but also it can be some issue with power ditstribution to escs.