Homebuilt quad with APM 2.6 running 3.1 release. This quad is overpowered and hovers at about 38% throttle (14" props 690kv).
The rate PIDs were auto-tuned with the only change being STAB P reduced back to 4.5. It flew fine in Stabilize and had about 45 minutes of flying time prior to this incident.
Last night it briefly lost RC and hit the throttle failsafe RTL. I didn’t realize that was what happened quickly enough and it crashed before I could return to Stabilize.
In the attached log the interesting stuff begins when RTL kicks in on line 3706. DRoll and Roll (and pitch) begin to diverge immediately and the vehicle wobbles its way into the ground.
I’m guessing it’s a tuning issue (outside of the basic rates). All thoughts are very welcome.
[attachment=0]20140106.jpg[/attachment]
Motor logging was enabled which might help. No motors “peg” initially so it doesn’t seem to be a complete failure on one corner. Assorted motors start to peg during the “fall” - presumable to try and gain control.
At the time RTL kicked in, it was flying fast forward at about 30 degrees nose down.
“Home” was directly behind the quad so the first RTL action was to crank up motors 1 and 3 to try and pitch back the other way. The pitch forward has only just started to reduce when it cut those two motors and cranked up the left rear (perhaps indicating a “weakness” in that corner?)
Pitch and Roll seem to be “kind-of” following the desired pitch/roll but then overshooting (perhaps as the big props got cranked up)
on-board video is available but probably not very useful. Orientation looks like at least 90 degrees at times (perhaps inverted but hard to tell). Motors sound like they’re revving as requested and the props I can see in the video (2 front) all keep spinning at various speeds - until it hits the ground and the cow starts licking them
Brendan, I have a large homebuilt quad with 14" props and 470 KV motors and have experienced something similar. When executing a RTL, if it was faced in the opposing direction it would jerk hard around and seem to lose stability, and start fluttering down like a leaf. Popping it into Stabilize would level it just before reaching the ground. I posted some queries online but never got any answers. Since then I did an Autotune and haven’t seen that behavior any more. So I think my issue was one of tuning, but I think big, slow props and long arms (in my case) complicated things.
It looks like you have some sort of weird coupling between the roll and pitch. Your yaw in and yaw out match very well.
Thanks for the reply Phil. The altitude fence action was to “do nothing”, it was the RC failsafe that triggered the RTL.
Unfortunately the APM didn’t handle the sudden change to RTL and got grumpy (it had a cow man!!). This seems to have happened to other folks over the past month with a similar result (in this forum).
It would be good to get some suggestions about, say which PIDs to back off, etc other wise I’ll just stick to more mainstream motor prop combinations. It’s a shame because the 690kv / 14" combination gave me an easy 15 minutes out of a 4200 4S battery, it was quiet, smooth and had heaps of power for carrying gimbal, etc - a very cruisy quad until RTL struck it down .
I’ve had a look and it looks like a mechanical failure or a very bad tune because everything is doing ok for the first part of the log and then suddenly the actual roll and pitch do not match the desired roll and pitch.
I’d guess a failure of the back right motor because the roll goes negative (rolls left) and pitch goes positive (rolls back) at about the same rate.
The failure happens as RTL is entered but this could be just because of the sudden request to slow down. Before this point the copter is flying very gently. Maybe the ESCs or motors just couldn’t handle it.
It’s not the firmware suddenly asking for a strange lean angle (i.e. desired roll and desired pitch) or sending a weird output to the motors. Those parts look normal.
Ok, after some offline Discussions with Randy, it looks like this has been a mechanical failure. Looking at the logs, the copter is trying very hard to level out, but is fighting a loosing battle.
Most likely something has come loose as the copter did the sudden switch to RTL.
I guess mechanical failure seems like the only option - although I haven’t been able to find anything physical. One of these 690Kv 22 pole “pancake” motors did loose sync during the first auto-tune despite using the latest SimonK firmware. Perhaps that’s what happened again when RTL was hit. I think I’ll shelve those motors for good.
[quote=“Brendan”]Thanks for taking a look at this folks.
I guess mechanical failure seems like the only option - although I haven’t been able to find anything physical. One of these 690Kv 22 pole “pancake” motors did loose sync during the first auto-tune despite using the latest SimonK firmware. Perhaps that’s what happened again when RTL was hit. I think I’ll shelve those motors for good.[/quote]
Sir i personally feel that the problem lies with the SimonK Firmware for de-syncing issue.
This is my theory–>
Sir I was thinking of this problem and i came up with an explanation as to why the stalling happens.
My theory is that simonK flashes ESC’s have the brake function of the ESC set very low. As a result of this, when throttling down from a high position, the motor and props’ inertia tends to cause the motor to continue spinning ahead of the intended position. However, if the brake function of the ESC was set to high (on simon k un-flashed ESCs’) a strong electromagnetic field would be generated to quickly brake the motor and slow it down to the desired rpm and position.
I am just 19 years old so i dont really have much experience. Do tell me if I am right or wrong.
.