MK Crash: Hey guys, I was maidening my x8 with MK electronics this morning and flew 2 flights with it before there was a failure and the craft pitched towards the ground from about 10m altitude. Brief description of the flight as follows: - 1st flight, everything went well, flew for 3-4 minutes on attitude mode and landed to check the logs. - 2nd flight, flew for a minute on manual (aka Free or attitude) mode, the aircraft was pitching nose forward slightly so I added trim, negative nick, a few clicks and didn't see much of a change, and added a few more clicks and didn't think much of it. Continued to fly, flicked to altitude (Vario) mode to test and the aircraft was operating as expected. Altitude holding well, losing some height with forward flight, but regaining once slowing to a hover. This is the exciting part - I flicked back to manual mode and the aircraft pitched violently to about -60 degrees nick. I corrected briefly after a loss of altitude and the aircraft continued to pitch - 20-30 degrees nick and in a panic I pitched it further into the ground with it struck and flipped. Looks like some frame, motor and prop damage from my initial inspections. Luckily the rollcage protected the electronics and I think they are ok. I checked the transmitter after the flight and I had trimmed the nick to -12 clicks, which left me wondering, was it that when I initially trimmed, it wasn't affecting the aircraft, but when I flicked to altitude mode, then back to manual, did the trim initiate? Or was it a software bug that was exposed when I flicked back to manual mode. Please check the logs or offer any insight to help solve the problem. I was pretty rattled after the crash as the aircraft came straight at me from about 50m away and crashed right near my feet. I'll be scared to use Altitude hold again if it turns out to be a bug. I'm hoping that it has something to do with the Trim I added, however I don't understand why the trim wasn't affecting flight as I was adding it initially. Aircraft Details: Frame: Droidworx Skyjib HL X8 Motors: Tiger 3515-15 ESC: Herkuless III Batteries: Zippy 8000 6s (x2) Gimbal: Dummy Weight MKStack: FC 2.5 Navi 2.1+External compass, GPS with shield. (All brand new, with latest firmware FC2.06) GPX overview: Flight date: 13/10/2014 10:02:27 AM Flight time: 10:02:27 AM.4 - 10:04:35 AM.0 Duration : 127 secs, 00:02:07 Batt. time : 130 secs, 00:02:10 Start Location : -33.7804166 / +151.2746131 @6.6m Elevation(GPS) : -0.21 4.07 14.042 m (min/avg/max) GPS alt. (raw) : 6.303 9 20.555 m Altitude(Barom.): 0.05 5.69 17.05 m Vertical speed : -2.62 0.06 2.82 m/s Max speed : 20.4 km/h Max target dist.: 0 m Max distance/LOS: 26.2 m / 29.8 m Sats : 8 10 11 Voltage : min. 21.4, max. 23.1 V Current : 47 73 111.8 A Wattage : 1034 1632 2392.52 W Capacity: 2601 mAh Motor1: 1.4 7.6 20.2 A Temp: 25 31 35 ∞C Motor2: 1.7 7.8 11.2 A Temp: 26 31 35 ∞C Motor3: 1.4 8.9 17.3 A Temp: 27 31 36 ∞C Motor4: 3.4 9.3 14.6 A Temp: 25 31 35 ∞C Motor5: 1.7 10.0 16.6 A Temp: 26 31 35 ∞C Motor6: 2.9 9.9 15.8 A Temp: 25 31 35 ∞C Motor7: 1.7 8.9 15.3 A Temp: 26 32 35 ∞C Motor8: 1.7 8.5 13.9 A Temp: 24 31 35 ∞C Magnet Field: 68 81 91 % (!) Magnet Inclination: 38 45 70 deg No errors found Attached: GPX, MKTool Settings, Graph of Nick angle vs Stick input, Graph of Nick angle vs Roll, pictures of craft. My buddy filmed the last 2 seconds of flight as the aircraft impacted. I will link to that once I've got it from him. Video here: Let me know if I can provide any more details. Cheers, Cam
On lines 237-240 of the included GPX file, the copter does some violent pitching, and the sticks seem to be all over the place. I wonder if you had a radio glitch, or something like that. It also appears ALT HOLD was disengaged. If you step through the GPX file at those lines with the sticks window and the OSD window open, you'll see what I mean. I'm not sure you should use TX trim if you're also using AH. I've never had to trim my transmitter. Better to adjust your copter's CG and motor tilt.
Hey Steve, thanks for looking. Yeah at the moment I disengaged altitude hold the copter pitched to a crazy angle, and then the rest of the time was trying to correct for it. I never normally trim but I wanted to see a stable hover before re-caling the accelerometers or adjusting COG. Hopefully that was the mistake and not a radio glitch. Anyway you can think of to check for radio glitches? Attached is a graph of Pitch vs RC pitch I forgot to include above, and Pitch vs Roll. You can see where the pitch occurs, and then my reactions to correct it.
Any other insight into what might have caused this? pretty reluctant to engage altitude hold again if I've exposed a bug somehow
There's always a possibility that you have a defective Flight Control board. Do you have another one (or somebody's you can borrow) to test with? I'm very suspicious of the severe amount of TX trim you put in, and what might have been the result of that. But I'm afraid we don't have a whole lot to go on.
Is there any chance that you accidentally moved the control sticks and recalibrated the ACC while the copter was not level. It is so easy to do (left stick, top right hand corner and hold briefly). It has the effect of readjusting what the ACC thinks is level -- and it will feel like you have to put in trim to counteract it. It is just possible that you inadvertently did this between the two flights -- and it would produce exactly the effect you describe -- suddenly the copter is no longer flying level (because it was inadvertently told a new "level"). I must admit I'm struggling to work my hypothesis to fit the situation where it's not until you go fully manual before the real trouble starts.....but.... Andy.
Steve - I do have a friend nearby with a board I can test. I'll chat to him about it. Andy - I don't think that's possible, as I began the flight in manual before switching to altitude. If the ACC was not level, it would be likely I would see that behaviour immediately after takeoff, rather then 2 minutes into flight. It's a strange bug I can't get my head around, I plugged the board back into MKTools yesterday and ran scopes for nick/roll while moving the board around and they seem spot on. The logs look accurate as well. The only thing I can think of to test if I exposed a trim bug -- (where trim applied only became active once I switched from manual->altitude-> then back to manual) -- will be to tether the aircraft and repeat exactly what I did, and see if the aircraft pitches violently or not. That won't be for 2-3 weeks whilst I wait for parts and a chance to rebuild.
I dont think it's related with AH. In the GPX file I see that you engaged the AH in mid flight and there seems to be no problem. How do you power the FC? and How do you make the connection between Herkules and FC? (I2C) Do you use 5-pin molex cable for powering the FC from Herkules?
That's true ozkan, I don't think it was AH itself, but it happened at the instantaneous moment I disengaged AH, so something related to that moment of AH -> manual? FC is powered via the molex cable as you say. I2C with the 5-pin molex cable from the top herkules board to the FC. Thanks for looking at the logs
Is it possible that you disengaged the AH when the throttle stick was down? Is it possible that, you were descending with AH on (with this case, your throttle should be down the half way), and the moment you turned off AH your gas stick was still down but you couldn't compensate since you were too close to the ground and it caused crash? When I look at your logs, I can see your gas stick movement at line 238-239. Edit: I think your logs stopped recording at some point. According to the altitude, we can't see the sudden drop in the altitude in the logs. To understand if the ACC-Z sensor is functioning properly, connect it to MK Tools and check the ACC-Z value. When the FlightCtrl is horizontal leveled, the AccZ value must show a level of 670-699.
Throttle stick was around halfway, likely above 50% as I am used to compensating to maintain lift when disengaging AH so I don't rapidly lose altitude, so I don't think that would be a problem. I was around 11m at the time, the aircraft wasn't dropping like there was a lack of thrust, I heard the motors go into overdrive which I assume was the front 4 acting to pitch the aircraft to -60degrees nick. The log for Altitude is strange. Is it possible that the attitude of the aircraft and the speed of decent didn't allow for accurate z-logging? Both batteries ejected from the connections at the point of impact and the logging would have stopped, but that doesn't explain the lack of accurate altitude values. I will check the AccZ value and respond if it isn't within the values you suggest
@Ozkan: The logs will not have stopped being recorded -- what happens is that the log file data is temporarily stored in the Atmega CPU chip;s RAM and only written out to the microSD card every few seconds. When a crash occurs and the Navigation Control board suddenly loses power, the data in the CPU's RAM is lost -- and that's the last few seconds of flight, unfortunately. Holger at MK is aware of this issue. The technical challenge is that writing to the microSD card is a relatively slow process so it cannot be done very frequently. I've certainly made the point to him that it would be highly desirable to flush the RAM data out to the microSD card more frequently for precisely these kinds of situations. Andy
@Ozkan - Tested the Acc Z value and it's right where you suggest it should lie, about 678-80 at the level zone. Any more suggestions to double check? I can't rebuild for another week, but hopefully a tethering test will prove if the problem is repeatable or not. Thanks for the help so far everyone!
I heard back from Holger that, if there is an error code detected, the data is immediately flushed from CPU RAM out to the microSD card. He is also going to experiment and see if he can flush the data out more frequently so that less flight data is loss in case of a mishap. Andy.
This's a weird issue. Personally, I would look if this was a power & connection issue. One of the weakest point of the MK system is the connection between PDB and FC. The 5-pin molex cable has failed several times for me. After these incidents, in addition to Molex connection, I manually solder +/- lines between PDB and FC. I also solder C&D lines as an extra security measure. But with Herkules III, it's hard to solder c&D lines, you need to use the molex cable solely. I'd recommend to double checking the molex cable you are using, put a hot glue on both end to make sure they stay in their place intact. Make sure the ribbon cable which connects the two sides of the Hercules III is also safe: PDB can get hot and it can melt the ribbon cable if it's touching the heatsink. To prevent this issue, I would put a heat shrink around these ribbon cable as well in case it's touching the heatsinks.
Good tips Ozkan, I've always felt like the ribbon cable is a bit flimsy. So I'll hot glue that and the molex and go from there.