I've crashed 3 birds in a row, I need someone else's opinion on what caused this crash. I have all of the logs posted below minus the .log file as there is a 7mb limit.
I have people depending on me to fly these things, any help would greatly be appreciated!!
The weather was perfect for flying, winds aloft were about 12-17mph, kp=2 and no other weather anomalies were noted that day.
Version of firmware is noted in the logs.
The end result was a spiral to the ground, I just don't know what caused it to enter the spiral in the first place, it was a good flight up until...
Replies
UPDATE:
So, I've been working on this. I've found that there was a design flaw. I was using the Castle Creations 10 Amp BEC. This provides 8 amps continuous if you run it on a 4S system. This bird, however, runs on a 6S system. This means the BEC's continuous output drops to just 4 amps of output. There are 4 digital servos on this bird each could draw up to 2 amps. SO, in a hard turn, etc, it was browning out the servos and crashing. That was the theory.
It was also a possibility that one of the inboard servos had one stripped gear, single tooth kind of thing. It was replaced.
I replaced the BEC with castle creation's bigger BEC that gives 13 amps continuous at 6S.
I also replaced the airspeed sensor unit, the tubes and the pitot tube itself.
It flew for one hour and 20 minutes doing donuts in the sky, that was 4 crashes and one good flight.
On this flight I had the arsp calibration variable turned on. When it landed I turned the calibration off (as per instructions). The arsp ratio variable was 3.9 something, WHICH IS WAY OFF, but it flew.
So today, I went out with some confidence that it would fly again. I was wrong. Today it crashed with all of the same symptoms. I call it airspeed creep. I can't solve this. My company has folded from it.
The good news is that the newly uploaded logs CLEARLY shows this airspeed creep in action. Wind speed was 5-8mph. No other weather variables of note.
The first log (short one, number 50, the pitot tube was covered) shows the airspeed immediately jump to 20-28mph on start up. I saw this, rolled my eyes and rebooted the aircraft after I checked other functions successfully. This starts the second log (number 38). Around the 25% mark the airspeed shot up to 27mph. I did an airspeed calibration and it went back to normal. Feeling desperate (company was riding on a good test) I launched in auto mode.
Shortly after the take off command was complete, she stalled and crashed. WTF is this? It is beyond frustration, I just laughed at it. I'm beginning to think I have an autopilot that is damaged. I'd like to know what to put on the tombstone of my company's grave.
airspeed creep.zip
looks like something failed, a control surface or servo,
the last ~35seconds is a struggle to maintain desired attitude, fine before that.
I had had issues with the air speed artificially jumping up from the 3-8mph nomial at rest and then going to 27mph and staying for about 10 seconds and then jumping back down. That sensor had been repaired. It was replaced before this flight. I could not see any indication that airspeed was incorrect, could you?
I did notice that the servo (on ONE testing only, non repeatable) "stuttered" when held in stabilize mode at about 20deg up angle. It was quite severe, certainly enough to crash a drone. It would not repeat this behavior if in manual mode no matter what angle I held the plane OR what manual angle I commanded the control surface to be at. This flight passed all flight checks including a stabilize up-down-left-right check. No flutter was detected.
It would make a lot of sense that a gear was striped in one of the servos but if that is the case I can't find it now now (any ideas on detecting that?)
It would also make sense that the autopilot (pixhawk) was causing that flutter somehow as it will not repeat in manual mode.
I can not make it repeat that flutter currently.
What makes you think it was a failed servo?
I see nothing strange about IAS or GS when the trouble starts.
Only servo output that tries to get desired roll (and not getting close to it.)
It happened at a WP, where desired roll jumped from ~0 to ~30 degree, if something mechanical is going to give, that's a good time.
test servo within full operating range in both directions (servo tester) while loading it (use fingers).
depending on the wreck, it's not always servo, it may as well be a control rod, hinge (hard to know if it's broken) - then you have the option of power loss at harder load on both..
That's why I did not say "a servo" , but "control surface or a servo" - which covers everything between , be it electicity or hydraulics :)
Close examination of the wires to the control surface shows nothing. The links are solid in the control surfaces, I can not move the linkage out of its seating. Load testing all four surfaces (flying wing) with my finger on them reveals no flutter. I tried to get them to fail. They are what was recommended by the manufacturer as far as strength goes but that does not mean they are correct for the loading. They should be though.
Any other possibilities?
If the aircraft was in a stall or near stall condition due to low airspeed and then commanded a servo to move, that too would look like a failed servo yes?
I say that because of airspeed creep, its been reported before. If you look at RC3 (throttle commanded) it goes steadily down for a good bit. As airspeed artificially creeps up the throttle would go down proportionally until a stall occurred and that would not show up in a standard log. I don't know how to correlate the data to figure our if that is the case or not.
if you see logline ~275000 you are climbing fine at 30% throttle, 24 should not be a stall then, also, you have good groundspeed where the problem starts, unless it was very windy, and you turned into tailwind, I don't see stall as an option, also, after a while you had 60 deg nose down, that would make it level off if it could.
Instead - I see good airspeed increase when diving, telling me it was ok.
Can't say anything about build quality, or power redundancy.
I am not a big fan of wild guessing, so i am quitting now :)
I see your points, TYVM for your time! I really appreciate it!
I checked the tlogs, wind speed was 6mph at start, the aircraft turned into the wind which was about 10-13mph at the time of the first nose down incident. Your right, it is looking like a bad servo, I'll retest tonight.
Again, TYVM!