So assuming this was in flight and rc signal was lost + no gps lock/home point recorded... can we assume that because it would be in air and sensing such it would just have hovered (until battery was low) rather than auto land and shut down?
Phantom 4 motors randomly turned off - bug or feature?
- Thread starter Dingoz
- Start date
So assuming this was in flight and rc signal was lost + no gps lock/home point recorded... can we assume that because it would be in air and sensing such it would just have hovered (until battery was low) rather than auto land and shut down?
With no GPS it would be in ATTI, and drifting but yes - if it never acquired a GPS lock then I would expect it to continue to drift until autoland for low battery. If it did acquire a GPS lock then it should establish a home point and hold there until autoland.
Meta4
Premium Pilot
- Joined
- Jan 8, 2014
- Messages
- 15,300
- Reaction score
- 9,242
You would have to be flying under cover or have GPS hardware issues to not get GPS when up in the air.
Even if you launched before recording a home point (you might launch from under cover), when the Phantom gets a good skyview it will get GPS reception and record a home point there.
Once it has GPS it will have position holding ability.
So then the autoland/motor off event in my test case on the desk was simply because it knew it had not even taken off yet and/or sensed nil movement so though it had landed -0 and shut down motors. Of course this is an easy test to do again (and just turn of RC while on desk in atti with no lock)... wait... why would it go into autoland at all if it knew it had not taken off or already landed - or is that just the named procedure anyway. Confused 
1. RC disconnect event (seems to be some bug but not really sure)
2. Autoland event (so if in atti with no gps lock and no home point recorded it should have just hovered and drifted until battery ran out and then auto landed - it should not have started auto landing with 50-60% battery as was the case OR because it had never taken off and it knew it could not hover as such is this aut land entr)?
3. Motor off (well that one is easy it was in auto land mode - sensed was landed already and shut off motors)>
UPDATE: So the home point might have been recorded (looks like 7-8 sats were locked at one stage or more before RC disconnect + auto landing + shutdown. But at the time of RC disconnect only 6 sats were locked up until auto landing triggered.
1. RC disconnect event (seems to be some bug but not really sure)
2. Autoland event (so if in atti with no gps lock and no home point recorded it should have just hovered and drifted until battery ran out and then auto landed - it should not have started auto landing with 50-60% battery as was the case OR because it had never taken off and it knew it could not hover as such is this aut land entr)?
3. Motor off (well that one is easy it was in auto land mode - sensed was landed already and shut off motors)>
UPDATE: So the home point might have been recorded (looks like 7-8 sats were locked at one stage or more before RC disconnect + auto landing + shutdown. But at the time of RC disconnect only 6 sats were locked up until auto landing triggered.
Last edited:
So then the autoland/motor off event in my test case on the desk was simply because it knew it had not even taken off yet and/or sensed nil movement so though it had landed -0 and shut down motors. Of course this is an easy test to do again (and just turn of RC while on desk in atti with no lock)... wait... why would it go into autoland at all if it knew it had not taken off or already landed - or is that just the named procedure anyway. Confused
1. RC disconnect event (seems to be some bug but not really sure)
2. Autoland event (so if in atti with no gps lock and no home point recorded it should have just hovered and drifted until battery ran out and then auto landed - it should not have started auto landing with 50-60% battery as was the case OR because it had never taken off and it knew it could not hover as such is this aut land entr)?
3. Motor off (well that one is easy it was in auto land mode - sensed was landed already and shut off motors)>
UPDATE: So the home point might have been recorded (looks like 7-8 sats were locked at one stage or more before RC disconnect + auto landing + shutdown. But at the time of RC disconnect only 6 sats were locked up until auto landing triggered.
A homepoint was recorded after the motor shutdown, at 645 s.
The homepoint was recorded at exactly the point when gpsHealth became greater than 3. The numSats was less than 8 in the early part of the .DAT and gpsHealth was also low during the early part. Then both improved without the AC being moved. Perhaps @Dingoz 's presence shielded the AC from GPS signals and then he moved out of the way.
I suspect that RC:connected represents the state of the connection in a network protocol layer sense. And, that the P4, for whatever reason, broke that connection and wouldn't have responded to stick commands although it would have received them. RC:sigStrength is derived from the number of frames lost in a given time frame. Since RC:sigStrength was showing 100% the entire flight that has to mean that the AC was receiving packets from the AC.
In regards to the GPS it was indoors so to date in similar position it generally takes a while to get 6+ sats and sometimes never does. It also fluctuates quite a bit... so generally struggling to pick up and maintain sats inside.
In summary of the information and theory etc, so far, can weight be put on one or the other cause:
Unknown: Why did RC disconnect happen ("User error" or "Fault")
Unknown: Why did drone enter auto-landing after RC disconnect ("Feature" or "Fault")
Unknown: If drone was in air vs idling on ground at the time, would auto-landing instruction still have been given the same?
As the operator at the time, I know I was only using the app to toggle on/off main switch for obstacle sensors (it turned off successfully first time but would not turn back on - but looked like it was on screen (green slider switch in position) - until you backed out of the screen and came back in and the setting would always be reverted to off again. Then it just shut down. To me it appears a bug in that phase of trying to toggle sensor on/off caused an RC disconnect (perhaps it was aware sensor command was given but was not taking/saving and considered that to be disconnected). Then the second event of auto landing occurred - this concerns me more in theory if it can not be explained (another bug? a drone randomly auto-landing into the ocean? a drone randomly dropping out of the sky?). Then the results of such a case "User Error", "Feature/Normal" or "Fault".
Unknown: Why did RC disconnect happen ("User error" or "Fault")
Unknown: Why did drone enter auto-landing after RC disconnect ("Feature" or "Fault")
Unknown: If drone was in air vs idling on ground at the time, would auto-landing instruction still have been given the same?
As the operator at the time, I know I was only using the app to toggle on/off main switch for obstacle sensors (it turned off successfully first time but would not turn back on - but looked like it was on screen (green slider switch in position) - until you backed out of the screen and came back in and the setting would always be reverted to off again. Then it just shut down. To me it appears a bug in that phase of trying to toggle sensor on/off caused an RC disconnect (perhaps it was aware sensor command was given but was not taking/saving and considered that to be disconnected). Then the second event of auto landing occurred - this concerns me more in theory if it can not be explained (another bug? a drone randomly auto-landing into the ocean? a drone randomly dropping out of the sky?). Then the results of such a case "User Error", "Feature/Normal" or "Fault".
Meta4
Premium Pilot
- Joined
- Jan 8, 2014
- Messages
- 15,300
- Reaction score
- 9,242
Why it entered autoland is the mystery.
Once it entered autoland, it quickly worked out that it had landed and shut down which is quite normal.
We know that Phantoms don't randomly enter autoland or shut down in flight.
Unless you can replicate the condition, I wouldn't be over-concerned about it.
So I tested a manual RC disconnect (powered off RC while stationary on ground with props spinning):
83.114 : 386881214 : 4039 [L-RC]rc_connect:1=>0
83.114 : 386881437 : 4039 [L-RC]rc_func_disable @ [85973 ms]
87.155 : 405067629 : 4241 [L-FMU/LED]action changed. rc completely lost
0)97.254 : 450512210 : 4746 [L-FLYMODE][Ctrl<6>] REQ_RC_LOST AUTO_LANDING ctrl_auto_landing
97.293 : 450688952 : 4748 [L-FMU/MOTOR] Stop. reason:ACT.landing
97.293 : 450689207 : 4748 [L-FMU/MOTOR]Total start times: 1, time: 39.26
97.293 : 450689477 : 4748 [L-PWM]set actuator mode:IDLE
97.293 : 450689943 : 4748 [L-FLYMODE]CTRL reset all by motor stopped
97.294 : 450694318 : 4748 [L-FMU/FSM]state changed. cur: motor off
97.295 : 450697032 : 4748 [L-FLYMODE][Ctrl<1>] REQ_RC_NORMAL ATTI_HOLD ctrl_gps_atti
97.311 : 450768167 : 4749 [L-IMU]set imu switch level = 1
97.314 : 450781678 : 4749 [L-FLYMODE][LANDING] reset landing status
97.873 : 453299971 : 4777 [L-RC]Start landing. rc vib!
And the results are the motors stopped almost instantly (leds flashed yellow for a few seconds then motors off). The log events look identical as to what happened when I did not actually switch off the RC in the original incident. So the auto land event is triggered by RC lost (while stationary on ground) and motors off is the logical next step once auto landing is complete.
The mystery that remains is what caused RC disconnect originally (I still suspect it was a bug somewhere in the chain and something to do with the sensor setting switch not working at the same time.)
On a side note I find it slightly amusing when pasting the event log text a sad face automatically is picked up from the text for RC lost which actually looks likes this : ( 0 ) Fitting
.Similar threads
