Hi,
I experienced a crash with a hexacopter that runs with ardupilot 4.0.3, whereupon I cannot find a correct explanation why this happens. I attached the crash logfile and a log file one flight before the crash.
Take-off weight was 8.4 kg. The copter is designed for a total weight of up to 12kg.
System history. This copter model was set up with ardupilot 3.6.x. The basic ATC_xxx flight parameters obtained in this way were adopted during the update to copter 4.x… In the past we have not observed any comparable attitude control behavior.
During the error analysis following the crash, we noticed that a parameter was not accepted and was at a different value than expected by us. This is ATC_THR_MIX_MAN. We flew the value to 0.5 in all copter 3.6.x. In the crashed system it is 0.1. We do not know if this is related to the crash, and besides we do not know why this parameter was wrong on the system. Maybe a parser error in Mission Planner maybe an human error. Normally we use the Missionplanner diff tool to migrate our basic parameter on new systems, and mostly we set check all checkboxes.
Crash happens in Mode Stabelize.
What happened during the three testflights and the last one with the crash:
We were about to fly useful values for PILOT_SPEED_DN in Loiter Mode. we tested the values 0 (default?), 75 and 140. we deliberately stayed below 150cm/s, i.e. 1/4 of the average induced speed.
During a test flight before the crash, I noticed the rapid thrusts of gas(stabeize mode) bring the copter out of attitude controll. The first time it happened at a low altitude. so that the copter staggered to the ground and attitude control started again as a skid of the landingskid touched the ground. in the following crash flight I provoked this again at a height of approx. 40m. I gave a series of fast thrusts and pitch and roll commands. And it happened again. The copter got out of control so quickly, that he falling on the ground. Countermeasures, reduce gas, accelerate or simply put it, my futile interception could no longer prevent the crash.
Your copter seems to have issues with the tuning. Based on the parameters you did not followed any guidelines for tuning heavy copters. ATC_ACCEL_P_MAX and ATC_ACCEL_R_MAX are way to high (actually default) and ATC_RAT_xxx_FILT are kept on 20 (against the recommended ~10).
I do recommend to follow the tuning guidelines https://ardupilot.org/copter/docs/tuning-process-instructions.html and do an autotune (with 0.05 -0.06 aggressivity) WITH THE LIGHTEST take off weight that you intend to use. Note; after tuning NEWER flight lighter than your tuning was done for.
In the log you sent I don’t see any crash, the last flight is a nicely executed landing in loiter mode, with a nice touchdown. Pitch and roll and throttle following exactly the pilot input with a nice less than 1G hump at the end. I hardly call that a crash.
thanks for the hint and the time you took for my question, and i agree totally with you.
i know the tuning guidelines and i tune an other big system in april.2020 where i following the suggested guidelines. and i start as interpolated valus for 17" props. mean, atc acc r and p max at ~70000 and also lower filter value at 10 according to the instructions INS_GYRO_FILTER / 2. and this copter are also get an roll/pitch autotune. the result is more than ok and it is flying well. in the moment and because it is also an expensive system i do not want to test this throttle movements to “maybe” reproduce the posted problem on this other big system. first i am interested expert opinions.
the new tuning values from this develop system, maybe solve the problem. in the moment i try to understand why the throttle movements can get the roll/pitch controller out of control so much and so fast. if your suggestions solve the problem we are happy, if not we need more options.
I fly 25+kg MTOW hexas and tuned them according to the guides. they fly like on rails. No more than .1-.2 degrees difference between desired and actual angles. So I can assure you that the guidelines are working just fine.
the landing on end of log23 is smooth. pilot_speed_dn is at 75cm :). i fix the google link. the link to the crash log file(24) is now under the second picture.
/g
wolke
Oh yeah, it looks like a crash
At first look its a feedback caused by wrong tune values. To put more to the table, at 10m alt, the crash check activated due too large deviations from desired angles, and stopped the motors.
@Eosbandi,
thanks to your first tip, we were able to fix the error. The problem was due to incorrectly set ATC parameters. The copter is now flying great. Following the guidelines for tuning heavy copters do the job :).
