Flyaway full speed left

I’ve been following this forum for a long time but this is my first post.

I’m flying with a perfectly stable and tuned quadcopter with 3.1.5.
The problem is that yesterday is went full speed left after entering loiter mode.
No damage to the model but I lost confidence in it and now I don’t want to fly in public places (problem because I use it for estate photografy).

Log shows a first rollin spike accompanied by a radio failsafe that clears in a few miliseconds (no problem I can live with a rtl radio failsafe).
My problem is that near the end it shows a second rollin spike and this time the rollin command is sustained causing a flyaway, but NO RADIO FAILSAFE. There is a gps error during this flyaway but is only when it crashed into a bush.

If it was a radio failure I’ve already repositioned the receiver and sat to best positions, but I just don’t want it to happen again

I was using an old 2200mah 3s battery that might not have the necessary punch and VCC fluctuates a bit (not sure if it happens with other batteries)

Can someone take a look to the log file and help me understand what when wrong.

Alien 560
apm 2.6 with LEA-6H
power module
multistar 3508-640kv
afro 30A esc

AR7000 receiver with sats.

Thanks a lot.
Best regards.

well, judging by your description and flight path, poorl placed magnetometer is most likely your problem - and I think your APM2 lost confidence in your config/build as well :slight_smile:

  • you did not log MAG, so it’s hard to say for sure - but it’s very,very likely.
    GPS glitch did not cause this.

You are not ready for commercial operations. - things can get worse.

I´m sure it´s not a magnetometer placement, since it is placed in a 20 cm pole and compassmot is very low (don´t remember the value but my inicial setup covered it and all the values are well within range) and I have never experienced a flyaway before in over 20 flights of the quad, like I said before all the flight modes are usually extremely stable and controlled, it was just this one time, if it was a constant I wouldn’t be worried since it would be a bad config, but the randomness of this kils my confidence.
I have about 15 years of RC experience (planes and helis) and my thought is that this was a radio glitch (pitchin and rollin spikes are way out of the max value I can input with the sticks) but I was hopping someone with more experience in log reading could confirm this.
I have been doing photography with a conventional heli without a FC with no problems and I intend to use the quad for the same job.

Just repeated compassmot:
“Interference at full throttle is 3% of mag field”.

Seems unlikely to be a compass problem.

This seems like a GPS glitch. The copter thought is was somewhere else, and corrected that by moving to that location. This is an inherent problem with a device that can move quickly in 3D space to a new location. Cars and planes cannot do this, and are less prone to such glitches.

The solution, i think, is to take off in loiter, with HDOP being less than 2m. You then don’t get a correction in the air when changing from stabilize to a GPS based mode

Hdop never goes above 1.9 and satcount is always above 9, the only time it goes above 1.6 is when I get an error (radio failsafe around minute 4 and the second time it triggers a gps error, but at that time I already had a bumpy landing, as seen when you graph gps glitch vs accX).

Graphing gps_lat/long vs rollin shows that the coordinated only change after rollin spike to -74. So gps error seems to me to be unlikely.

I’m starting to think I have a darker problem than a simple radio jitter. Maybe something related to Vcc, small 3S 2200mah 20c not capable of providing the necessary current to keep everything working??

HDOP is an indicator, not an objective value. GPS location is also an indicator, it can at time be really inaccurate! The issue is that you can fly under the control of GPS, but the initial position is likely to read ‘off-centre’ of that. When you go into GPS mode in the air, that delta will be corrected, and at speed (unlike a car, or plane which have psychical constraints for just changing direction). Once corrected the system will feel ‘locked-in’ and normal.

I’ve flown a lot with 2200mAh 3S Batteries. No issue whatsoever. Unless Vcc shows a drop below 4.5V in the log, its likely not to be the issue.

My worst glitch i had was changing to Loiter from stab, and the flight path moved more that 25m. it was really alarming. Very fast and resulted in a crash. Randy has added extra code to detect location changes at quite that speed, but it can still occur.

That makes sense!!
And the new GPS position correction would be commanded by a rollin input???

Sorry if the last post sounds a bit sarcastic, It was not my intention. :confused:

please observe how many even points is recorded, and the fine curved flightpath.
there is no big jumps in position.

  • is there a chance the GPS/magnetometer become loose and relocated ?

Gps module is firmly attached to a pole on the frame.