Yet another flyway in the north

[AH-1G]

I just calculated on Google earth, he crashed into the hill!
Elevation where he took off is 1215 feet height of the hill is 1609 feet.
Drone was flying 1605 feet, so he may have hit some trees

 

[DoomMeister]

Could be a good place to start looking.

 

[AH-1G]

opps highest point of the hill is 1609 feet.
I think it's going to be close to the the two structures on top of the hill. Hopefully.

 

[AH-1G]

@ukaleq
Maybe its on top of the hill between the red and white stack and the other structure just south, or in that vicinity?
Hill looks like a ski slope.

 

[Steve Carr]

It seems to be caused by some inconsistency between the GPS signal (which degrades at high latitude as the sats get "closer to the horizon" - although ca. 66°N should not be critical)

The Q and H series use GPS plus GLONASS which negates any effect from higher latitudes. 66N would not be an issue.

 

[h-elsner]

For some reason the system switched by itself from Angle to Manual mode and did not record a homepoint. I do not understand yet what really happened but the problem starts already before the RTH was set. It looks simlar to compass-related fly-aways that we have had in the past with Chroma (lost GPS, followed by fly-away with increasing and unusual high speed). But H Plus is another flight controller and anothe generation of copter, hmmm...
I don't think it stops by itself. Maybe the hill has taken over the task to stop the flight.

I'm still on checking...

 

[h-elsner]

After GPS was lost the copter was blown with the wind. Not a classical fly-away.
Why I think so: If we look at pitch and roll, the copter follows exactly the stick movements. But during the whole flight it cound keep nearby and on place only with stick on near maximum.
All sticks have maximum +/-70-80%. So, I think it was in turtle mode or slider something nearby. This makes it hard to fight the wind.
Air speed increases dramatically if sticks remain in neutral position especially at the end of the recorded flight. I have cut the last minute to make it visible.

The upper chart is distance to first dataset in cutted part of the flight. Copters pitch follows pitch stick. Full pitch down holds the copter in the same distance.

Same for roll, roll up holds the copter. Neutral let it flee.

The main question to Yuneec here is, why GPS was lost although the GPS accuracy is good (~60cm vertical, ~30cm horizontal), GPS fix is most of the time '3D position' but also 'DGPS/SBAS aided 3D position' (all from sensor file). So, why switch to manual mode without flight mode switch touched?

br HE

 

[AH-1G]

Didn't this all start when the RTH was activated?
As I also recall someone else had a fly away when he activated the RTH?

 

[h-elsner]

No, RTH was later (somewhere in the middle of the escape already 100m away) and had no visible influence on the behaviour.

br HE

 

[AH-1G]

Then I misunderstood post #2

 

[DoomMeister]

Then I misunderstood post #2

@AH-1G, the mode automatically switched to Manual well before the RTH was initiated, and I could see the craft responding to the control inputs. Yet I missed seeing the drift that Helmut detected, before the RTH was initiated. It did seem that control was lost after RTH was initiated though.

Great job picking that up Helmut! That is why I always want to see what you read in the files. Did you notice the split Sensor files for the flight just previous to the flyaway? It was like there was a break in the telemetry from the PX4.

 

[AH-1G]

@AH-1G, the mode automatically switched to Manual well before the RTH was initiated, and I could see the craft responding to the control inputs. Yet I missed seeing the drift that Helmut detected, before the RTH was initiated. It did seem that control was lost after RTH was initiated though.

Great job picking that up Helmut! That is why I always want to see what you read in the files. Did you notice the split Sensor files for the flight just previous to the flyaway? It was like there was a break in the telemetry from the PX4.

so I'm looking at sensor file #32 is this what your talking about, showing only half of the graph?

 

[DoomMeister]

If you take a look at the kml file that gets created when you open the Sensor file you will see the PX4 quit recording GPS position at the point it starts heading east at an accelerated rate.

 

[AH-1G]

I'm still learning how to read all the data let alone navigating through the software.

 

[DoomMeister]

Hey, I’m still learning too. Helmut is about two light years ahead of me at deciphering the data and he digs even deeper into the Sensor files. There is more data in them than what gets extracted with his program.

 

[h-elsner]

I'm still learnig too. Every new case has new surprises.

Following assignment can be done:
Telemetry_00017 --> Sensor_00030.txt + "CRITICAL: 'ALL DATA LINKS LOST." + Sensor_00031.txt
Telemetry_00018 --> Sensor_00032.txt (part of)

Here is the time line:

Picture for pitch updated with RTH:

Why sensor file 32 is somehow cut is not clear for me. Maybe at this time the 5.8GHz connection was lost. I guess the sensor data are transmitted over serial connetion to camera and seed to video downlink.

My Typhoon H created empty sensor files in case no camera is mounted.

br HE

 

[DoomMeister]

My Typhoon H created empty sensor files in case no camera is mounted.

I’ve been flying my TH without the camera because the CGO3+ is out for repair (thankfully on its way back today). I’ll have a look at my flight logs. That is interesting.

 

[ukaleq]

Hi guys,

Thanks for your work.

In the case of the sensor files, what I experienced when flying with no camera last summer is that there were no files saved at all in this case.

The 5.8GHz link on that particular camera has always been very unstable and I indeed lost it quite often. I had told Yuneec about that.

I checked the wind direction at that time, it was between 250 and 260 degrees. Speed was around 9m/s with gusts around 14m/s. On the ground, I really didn't feel much wind though but the copter was higher.

My recollection is that as in previous flights, I had to fight some drift almost from the beginning. I remember having adjusted the turtle/hare speed control. But the question is: if I successfully managed to keep the aircraft within a reasonable range in the beginning, why did it leave me no chance at all when the fly away started? I have the impression it really accelerated suddenly and became unresponsive to stick input, which prompted me to trigger the RTH mode.

Is there a way to prove from the data that I did not switch off the GPS manually? Because that's most likely what Yuneec will claim. The only thing I did was to switch from angle to RTH. Given that the manual mode is buried in the software it can't have been switched accidentally.

 

[ukaleq]

Other questions:

• what is the correspondence between the flight modes and the numbers displayed? Looking at previous flights I keep seeing frequent changes...
• is it possible to tell the position of the turtle/hare control?

Anyway, I think the software is really botched. If the copter, for some reason, switches to manual mode, there should be a message that says it clearly (also with a sound, we are supposed to fly LOS and not to be always looking at the screen). It should also alert you on the speed control setting because indeed if you were at the lowest possible setting when GPS was active (so as to maneuver the copter very accurately) that becomes a potential danger when the copter abruptly switches to manual.

 

[ukaleq]

If you take a look at the kml file that gets created when you open the Sensor file you will see the PX4 quit recording GPS position at the point it starts heading east at an accelerated rate.

I am a little confused at this point. Is the GPS drawn from the sensor file or from the telemetry file? It's not clear to me what kind of information the sensor files contain and how it can be displayed in the software. When opening the sensor file, I see a table of what looks like hexadecimal numbers.

In general, are you surprised that the 2.4GHz link was cut so quickly?