Hi All
The Problem:
My Phantom 3 Advance had a loss of control whilst flying today. The following is an annotation of my flight record;  It took off well climbing to a height of 29m without any issue, it also was listening to the "forwards, backwards, side-side" commands. At 28 seconds into the flight I descended to 9m and paused. At 33 seconds, the drone switches from GPS to P-Atti (without my command). At 34 seconds; without any control input, my drone flew at 11.6 m/s in the "backwards" motion. I immedietely tried to correct this by controlling "forwards" with no success. I was able to control climb& descend, and climbed to 33m whilst still moving violently out of control. At 39 seconds, the drone enters back into P-GPS, however still no control with the right stick is noted. At 43 seconds the drone again entered P-Ati mode, followed by two warning displays; "Propulsion output is limited to ensure life of battery" & Magnetic field interferece exit P-GPS mode". At 52 seconds, the drone entered OPTI. At 54 seconds I activated return to home, and the drone landed safely back at its home point.
After The Problem
After landing safely, I recallobrated the controller and ascended to a height of 3m, where the drone behaved normally to all of its commands. The drone flew for approx. 10 minutes and landed safely without any further issues.
Back Story
An earlier flight that day (5 hours before) had displayed the following warnings which appeared around 10 minutes into flight simultaneously ; "Ultrasonic Systems Error. Land immiediately", "Motor Obstructed, "Remote Controller Calibration Error". I landed immideitely following. I researched the faults, made sure the motors where spinning freely without any friction & re-calibrated the remote. I then had an entire battery's life worth of flight with no other warnings, and the drone behavaed normally.
After returning home I successfully updated the remotes firmware. I tried many times to update the drones firmware without success. Following reconnecting the controller to the drone, I then had the out of control episode mentioned above.
Any advice or comments would be appreciated.

Regards
Cam

That sounds like you messed up your compass by launching from a steel or reinforced concrete surface.

You were lucky to get away with it this time.
Many others have done that and not been so lucky.
Never launch from steel or reinforced concrete surfaces.

Thanks for the reply. I launched from a grass sports oval.

Cameron9007 Posted at 2018-7-10 07:44
Thanks for the reply. I launched from a grass sports oval.

Without flight data, we can only guess.
Go to http://www.phantomhelp.com/LogViewer/Upload/
Follow the instructions there to upload your flight record from your phone or tablet.
Come back and post a link to the report it gives you.

Labroides Posted at 2018-7-10 07:47
Without flight data, we can only guess.
Go to http://www.phantomhelp.com/LogViewer/Upload/
Follow the instructions there to upload your flight record from your phone or tablet.

Here is the log;
http://www.phantomhelp.com/LogViewer/NRSWJK0II2MMIAL93Z90/

This comment is not meant to explain the event above ... but just to clear a point.

Many sports grounds have buried piping / cabling / all sorts of gear under that grass. You cannot be sure of its non-interference characteristics. Its same with roads / driveways and small backyards.

The Model dropping in / out of GPS / ATTI modes usually indicates any or combination of : obstructed satellite view, magnetic interference, aberration in control signal interpreted as  a mode change, overheating Flight Controller, faulty Flight Controller data inputs, bad cable connections, ... the list goes on because all the parts of the units are interacting .... so its a matter of check everything is correct and carefully test fly.

Re-calibration or playing with FW is usually not the answer ... fact that many of us automatically do this ... is no proof of it being right thing to do.

My usual action on any strange events in flight ? To change location to a clear known good area and put in fresh charged battery and take a careful test flight to see if it re-occurs at this safe location. If nothing happens - then I try to find out what it is at old location. A simple hand compass can often show anomalies ... looking at buildings / trees etc. to see what sky view is - note that only vertical is NOT sufficient - you need angled view as well. RF Towers ... military radar relatively near ? You would be surprised at what distance high power Military radar / comms gear can wipe out various other gear ... near military games area ? Often the GPS for public use is disrupted while they play ... just examples.

Nigel

Thanks for you response Nigel! A very helpful answer, I appreciate your advice.

Hi Cameron, sorry for the late response. Just to verify, has the drone been involved in the crash or hard landing before? Could you read the current aircraft firmware? Not sure whether the previous issue is a coincidence, you can upload the original flight record and the flight log in the aircraft for further analysis. We do not support to analyze the data from the third party software, please kindly note and understand.

This looks like a case of calibrating whilst over some sort of buried metal, as has been explained above.  This is enough to mess up the aircraft's baseline view of what is normal magnetic conditions and what is not.  When you flew away from the take-off point and cleared the hidden metal, the magnetic conditions were very different.  This causes unexpected sensor data to be returned.  This is what all those error messages mean, even though it doesn't seem accurate or particularly meaningful at the time.  The system compares the speed/compass heading/yaw/position and it determined there was nonsensical data that it couldn't rely on, with sensor data being in conflict with other sensor data.  The system decides that it isn't reliable to use GPS positioning, so it ignores that data.  This causes it to drop into ATTI mode.

In ATTI mode the aircraft will no longer maintain an exact horizontal position and will drift in the sky under manual control (altitude is maintained by a barometer measuring air pressure).  In this mode it will be at the mercy of prevailing winds, which will be different at altitude compared to that experienced on the ground.  It seems you experienced a 11.6m/s wind blowing your aircraft away "backwards".  This wasn't uncontrollable.  A Phantom 3 Advanced will be able to fly manually at 15-16m/s on full stick (it is only 10m/s in Go Home mode).  So you had sufficient power to overcome the wind if flown manually, but if you had issued a Go Home at that point it wouldn't have been able to overcome it and would appear to be standing still.  It would have taken a while to combat the 10m/s wind and fly in the opposite direction at 5m/s so it would appear to be unsuccessful or not obvious.  You would need to persevere with the counter-move for a while to get a result.  Climbing usually results in you flying into even stronger wind, and as you exaggerated it went even more "violently out of control", and may seem like the right stick is having no effect.  It is not uncommon for the aircraft to switch back and forth between GPS and ATTI modes in these circumstances.  What you are actually experiencing is "flying it manually", with no GPS position-holding help.  That does not mean it is out of control.

The Propulsion Output Limited error message is in response to the motors spinning as hard as they can in order to fight the wind.  It does this so they don't overheat and unnecessarily strain the battery over long periods.  This sometimes returns a Motor Obstructed warning for good measure, i.e. the motors are working harder than expected chopping through strong wind and it appears like an obstruction to the system.  Again, it's a general-purpose error message that isn't necessarily accurate or helpful.   Fortunately you had a good combination of manageable wind gust speeds and RTH power in order to get back home and land.  FYI it has been pretty windy on the east coast for the past week, possibly in your location as well.  I'd think twice about flying in 10m/s winds myself.

The problem seems to be confined to a bad calibration in a questionable location.  The solution is to go to a "good" location with no interference and calibrate there.   Don't calibrate again when you go to a new location -- there is no need and it could cause problems, as you experienced.

Despite thinking it is logical to assume a park is a good location without any interference, let's look at the area where you were flying.  It shows what appears to be a network of underground irrigation pipes and sprinkler systems.   This is a satellite view from 2001: