I think I need some help with the data log analysis. I try my best to find the root cause for 3 crashes. Is it possible that someone can double check the logs?
My conclusions:
Copter 1:
problem with Lightware LW20 Rangefinder
Seems that the EKF doesn’t see the unhealthy signal…
Copter 2:
FC switches all the time from compass 1 to compass 0 and back. Maybe the mounted gps module get’s loose over the flights.
vibration compensation on, maybe a problem with a motor/propeller
I have seen a GPS glitch message, but the GPS HDOP is in a good range during the full flight… Why the gps glitch appears?
Copter 3:
GPS module lost suddenly satellites, switch to land mode
All could use GPS_GNSS_MODE,65 or maybe 5 - test to see which works best in your region.
All 3 copters seem to have a low C battery, the voltage drops a lot during these issues - maybe due to props hitting the ground Do the ESCs have their low voltage cutoff disabled?
Copter 1
Rangefinder - copter appeared to fly into the ground even though you gave throttle input. Probably switch into Stabilise mode if that happens again - at least until the rangefinder issue is sorted. I’m not very familiar with rangefinders - your settings look normal to me.
Maybe the rangefinder needs a better power supply, and possibly some shielding on the wiring - they seem to be common issues.
There is some other tuning issues - like why is I term almost double P term - they should normally be equal except in very rare circumstances.
Copter 2
GPS glitch is a message indicating GPS position and IMU-calculated position dont agree, or GPS heading and compass heading dont agree. So it’s not necessarily an issue with the GPS, but usually the compass calibration and interference, or vibrations.
Vibrations below about 15 are OK, but lower is better. Over 15 to 20 is “this can go either way” and over 20 needs attention. 30 is bad and is actively giving you issues.
Motor output and vibrations were increasing, motor outputs went to near maximum and oscillations started. I suspect this was all a self-perpetuating loop → more noise affected stability causing increasing motor output oscillations, causing more noise and vibrations, causing more oscillations…
It seemed sluggish to climb, and motor outputs going to near maximum indicates this copter is a bit under-powered.
There’s a few minor tuning issues that didnt necessarily contribute, but tuning could improve after vibrations are fixed, and this will make the copter more resilient.
Copter 3
Connections? Vibrations are worth some time and effort.
The loss of GPS was quite some time back in the log.
Landing and attitude control was going very well until right at the end.
Pitch and roll departed from desired near landing - I cant tell if the copter was touching the ground or not. It didnt seem like GPS position was the problem. The position controller doesnt show anything of significance, like a desire to move.
EDIT:
So my text is fairly general - it would take a long time to fully analyse those logs.
I would say generally:
Minimise vibrations - time spent on this is well worth the effort.
Improve tuning (including compass cal) - better tuning allows developing issues to be obvious sooner.
Rangefinder - unsure, do some searches and see what others did.
Thank you very much for have a look at my logs. I totally agree with you, it takes some time to analyze the logs seriously. I spend yesterday 2 hours to them…
Regarding the batteries, those are only 25C rated. But it seems never a problem so far…
Generally, those copters have between 250 and over 500 flights and works normally very reliable. But I have mention that copter 2 and 3 have some crash’s during their lives. Most of them were with trees and similar obstacles but also tilt over right from the start due to that fact that not all motors starts (dshot issue). I believe that those impacts could damage the electronic components and of course also the hole mechanics… I forgot to mention, I’m using on all copters CUAV parts (Nora FC, Neo 3 Pro GPS module).
Some notes…
Copter 1
The Lightware LW20 rangefinder is directly powered by the flight controller. Due the huge amount of successful flights with LW20 rangefinders I assume that maby the lens from the rangefinder get damaged, covered with dirt, or some wiring issue… Regarding the tuning issues. What’s the name of the parameters with the suspicious I and P terms?
Copter 2
I believe that something gets lose on that copter and cause the rising vibrations, maybe the gps/compass module, that would also explain the compass issue. The vibration levels looks a lot better during the first flights of this copter (before I gave the copter to the pilots)… It seems that the pilot doesn’t realize the damage and keep flying…
Copter 3
Good hint, I will definitely check the cables and connections. If I have mounted a second gps module, is the flight controller able to switch to the second module and keep flying? According to my knowledge, Yes?
The specific tuning/PID terms I mentioned were
ATC_RAT_PIT_I
ATC_RAT_PIT_P
ATC_RAT_RLL_I
ATC_RAT_RLL_P
The P and I are usually the same value for RLL and the same for PIT.
The fact that they are different (I higher than P) indicates someone did some manual tuning, but didnt fully understand the Arducopter system.
There is probably a few other details you could look at, like Harmonic Notch Filter, then run Autotune.
A second GPS could keep working - provided whatever kills the first doesnt also kill the second one.
There is options for Blend or Use Best and so on. It can depend if they are identical units or not.
Yes, I manually tune my copters. I tried autotune a few years ago, but it produces not usable results. So I decided to to it manually. Regarding the high I term. I noticed that a higher I term help to keep the copter stable during descending and being affected by its own downwash. Higher P and lower I terms gave me more wobbles during filming. What’s the general rule in arducopter, P is double than I? Can this rule used for all type of drones (small quad, big filming copter with gimbal damping system).
I always use the harmonic notch filter. It’s a great tool to make the copter even more stable.
Maybe it will test the dual gps use in the future. Should be an improvement for the reliability.
Almost always P = I
Autotune can produce poor results if it is being affected by noise or some other physical issue, so if Autotune doesnt work properly there is usually a fixable reason why.
Instability during descent can be MOT_THST_EXPO amongst other things.
Do the ESCs have their low voltage cutoff disabled?