my quad with Speedybee f405 v3 Stack and AP 4.5.7 crashed strangely in a simple flight. The contact between the nickel strip and the wire in the Li-ion battery pack was destroyed. I don’t know if this is the cause or the result. Please help me find out the cause of the crash.
From my understanding, the power to motors spiked, seemingly due to application of throttle, this increased current to an extreme, to 50 A, after which the log ends. This suggests that battery might have failed.
I never use welded nickel strip connected batteries, I prefer to solder. That is foolproof. Potentially it reduces the quality of the battery due to overheating during soldering process, but the battery can never fail like this disconnecting completely. The other option is to have two batteries.
Now, there are a few issues I am not so clear about: first, despite initial climb, the baro seems to indicate beginning of descent, depsite full power still being applied. But considering that RCout all are equal and high, current is high, and that the pitch and roll are pretty close to vertical position, the copter should have continued to climb, which is not reflected on the graph.
The other thing which perplexes me a little, is that throttle seems to be applied gradually, but the power spike is very pronounced, i.e. initially, the increased throttle does not have any effect, and then the copter shoots up. But that could be due to POshold mode which was waiting for the throttle to go past the mid position,
Thanks for your reply. I use PILOT_ACCEL_Z=250 and PILOT_SPEED_UP=250. So vertical acceleration and speed are restricted enough in AltHold and PosHold modes. This helps to avoid high current values. And the throttle is to go past the mid position as you noted.
It doesn’t look strange. Vbat plummeted, Current spiked, which would be expected from an internal battery failure it looks like, and logging stops from FC shut down. Where is the mystery?
The mystery is this: the log does not show the failure of the battery per se, the VBAT does not “plummet” due to failure but reduces somewhat due to a normal voltage sag which is due to high current, it goes from 15 to 12 V. That is normal for 50 A.
I can not understand well the baro alt data during the last 500 mS. It begins to climb, that is OK, and then it begins to descend sharply, despite the full power, i.e. 12 V x 50 A still displayed on the graph. The Vz from GPS seems to indicate a more correct story, i.e. the copter was descending, the power was applied and the descent began to slow down.
My alternative hypothesis, which is probably not very likely, is this, could the baro sensor had failed in the sense that it started to show descent, as far as I know, the ardupilot takes baro data as primary source for alt/Vz, so that this provoked a much higher current, this demand was reflected by the RCout, and this high current would have provoked the battery failure. The copter has several ups and downs in the log, but not near the 50 A mark.
I would bet that the battery failed with a very high probability, but I would not say that this is reflected in the graphs.