My Skywalker came to a sticky end this weekend. Strange things where happening while I had it on the ground waiting for the next flight. It was still on since the last flight and I was looking over the new plans, when I saw the flight data suddenly go insane! I was convinced that the airplane was doing climbs, rolls and dives while on the ground.
Have a look at the attached telemetry about log 36% through. It was indicating no motion for 10 minutes and then studdenly it started to roll and climb and dive on the APM Planner!!
Switching on stabilize confirmed this as the ailerons and elevator went insane trying to compensate for the phantom movement.
I cycled the power and all seemed well, put it up in the air, had a test on stabilise, and I switched it to autopilot. I turned my back and heard people around me gasp and I knew. The plane cratered, turning my Skywalker into a million pieces of foam.
Has any body got any idea what causes this phantom movements?? I'm running APM 1 Firmware 2.34.
I have never done any analysis on log files. I thought I would look at yours and give you my observations/narrative of what I think is happening. Someone else here may be able to tell you more correctly. This is just my quick attempt to understand what is happening, so that I learn to read it better.
APM is in manual.
throughout the log there is a yaw reported, as if the airplane is spinning slowly. There is no input from the accelerometer, and this is not reflected in the magnetometer, maybe it's coming from the GPS?…. there is no yaw input, and the turning is very slow, but constant, and a slowly increasing rate. I bet the plane was sitting on the ground, but for some reason the yaw value just keeps rising very slowly. About 1-2 degrees a second.
At 36.20%, the ground speed starts to produce results that are not consistent with the accelerometer. There is no movement of the plane, but GS rapidly increases to 25, while AS is zero, accelerometers are zero. It then gets erratic, changing to zero, 17, etc.
This suggests some GPS problem?
At 36.78%, the gyro shows movement (roll), and this seems to be supposed by accelerometer values.
By 37.37%, the gyro reports a full roll all the way over. There is no AS, no GS, no inputs or outputs to throttle. The plane just appears to be rolling over in place at about a rate of 6 rolls per minute. The roll is always to the right/starboard. This continues, and a slight pitch osellation begins, and gradually increases. At 38.97%, the pitch is -50 degrees, and the roll continues to the right at 6 rolls per minute, now with significant pitching, as if the plane is wobbling around the longitudinal access (while AS, GS remain at zero, altitude all remain static.) The pitching varies from -80 to about -5 degrees during this time (it does not go above the horizon.) At 41%, the plane pitches down past 90 degrees, and it may be relevant to note that the extreme pitch seems to occur at the same time as the roll is at it's most extreme. The heading changes continue in a clockwise direction from the very start of the log file, and continue to rotate, having accelerated to roughly 4.5-5 full rotations per minute.
The "roll" and "yaw" continues with pitch fixed between -50 -70 for a long time.
At about 34.27%, the GPS lock is reportedly lost, going to "no fix" for a brief moment. Sat count drops to 3. It returns after two seconds to a sat count of 4. Then 6, then five, then four, then the fix is lost again as it returns to 3. There are no significant changes to altitude, no throttle inputs or outputs, no AS or GS during this time. The Sat count bounces a few times between 3 (no fix) and 4-6. At about 55.50% the fix is lost for an extended period of time at at 56.10% the nose of the plane pitches up past the horizon for the first time since the "roll" started The altitude increases by a meter, and accelerometers confirm vertical movement. Maybe you are picking up the plane to look at it?
At 56.30% the GPS fix is restored, while the plane is 'lifted'. However, the right roll continues throughout what I suspect is your handling of the plane. I doubt you are rolling the plane, with those big wings… At 54.32 the GPS sat count drops again to 3, and then locks back on a second later. It is lost again by 54.93, restored, and lost again several more times for between 1 and ~8 seconds each time.
At 56.87%, you switch the APM into stabilize while still presumably holding the plane up about a meter off the ground. The plane continues to think it is rolling to the right, and pitching down -30 to -60+ degrees. GPS fix keeps coming and going. There has been no change in throttle position since the beginning of the log file, suggesting that it is displaying the "zero" position. At 58.69% you put the plane back on the ground. It still thinks it is rolling, which we can discount, since it continues to show zero AG and GS, and a plane with wings that size does not simply roll over while on the ground.
At 59.03% the ground speed increases to 1.1, increasing to 1.4 at 59.60%, showing several seconds of movement, with zero AS, before gradually slowing down to zero at 60.23%
At about 61.60%, you switch into manual. Throughout this time, GPS sat count bounces between 3 and 5, with the fix reported lost each time it dips to 3.
At about 62.70, the display freezes, suggesting that you reset the APM and the telemetry data is lost. This is confirmed at 62.81, when the mode status changes to initializing. At 62.90% the APM has rebooted. It may be nothing, but during the whole first part of the log, the voltage was reported at 16.13-16.16v, while the post initialization, the voltage is shown as various low levels from just over 5v down to 0.0 volts, then it starts to increase, and reaches 16.20v at around 64.63%. Perhaps you have changed the battery, and only afterwards connected the balancing port?
Someone seems to be handling the plane at 65.07%, lifting it up a little. It is switched into stabilize, and rotated by hand to check the control surfaces, roll, then pitch.
At around 67.40%, the throttle is increased for the first time during the log, followed afterwards by a major acceleration along the x axis, and Z axis, and a drop in Y, this may be a launch, as the APM seems to have gained 5 meters. However, AS and GS still show 0.0, and there is only one satellite in the sat count. There is no GPS fix. We are flying without an Airspeed sensor, and without a sat fix! This is not going to end well…..
The plane goes into a slight left bank and continues to rise, leveling off at 29 meters AGL at 68.69%, then rolling into a right bank and gaining in altitude slightly. It might have caught a thermal, it rises up quickly to 50 meters by 69.29%, pitches up 10 degrees, and begins to lose altitude. It still has no GPS fix, the sat count is still zero. It cruses around, banking slightly left and right at 58-60 meters AGL, still no GPS fix. The throttle is kicking, and the battery is pulled down to 15.60v from the draw.
At 70.48%, the APM is switched to AUTO. We are at 71 meters AGL. However, we still have no GPS fix. Did I mention that this is not going to end well?
We climb over 80 meters, then fall off again to 73 and back up to 86, climbing fast at 12 degrees. Climb to over 90 meters, before losing altitude again to the low 80's, then back up to 90 meters. The plane seems to be circling in a left bank while holding at around 90 meters AGL. It still has no GPS fix, and is set to fly to WP1, but it has no bearing, since it has no location. This is about 72.60%.
It throttles back slightly at 72.90%, 93 AGL, still in auto, still showing 0 AS, 0 GS, and a sat count of 1. At 73.13%, the APM is switched back into stabilize, level the plane off and reduce altitude gradually to 32 AGL while in a left bank, before leveling the wings again at 75.63%. There is still no GPS fix at this point, no indicated AS or GS. You bring the plane down to 20 AGL before rolling right into a bank, then into a left bank at 16 AGL, pulling out at 11AGL and gliding down about 6 AGL, then maneuvering a bit rolling left and right around 4 AGL. At 77.70%, there is still no GPS fix, sat count is 1. AS, GS is zero, and we begin climbing to 10 AGL, then rapidly to about 40AGL before banking left, circling, and dropping to 23 AGL. Still no GPS fix. 79.88%. We roll into a right bank and then level off at 20AGL. We drop to 17AGL, then switch to manual by 80.30%.
We fly around in manual for a bit, at 20-35 off the deck. Sometime just before 83.85% we switch stabilize back on (though you don't seem to need it, you fly nice.) We are at 19 AGL, no GPS, no indicated AS or GS, and the mission is set to head for WP1, which is at an unknown distance, if we were not in stabilize. We get to 2 AGL at 84.71. I'm going to guess we are flying along the contours of the land, as I doubt you got within 2 of the ground while in a 30 degree left bank with those big beautiful wings, so the place you are flying over is lower than where you launched. However, we don't have any location data, since the GPS sat count is still 1. 84.71%. At 84.82% we are at 0 ALG and at 85.18% we are at -1 but still moving, more or less level but at -10 degree pitch. By 85.23% we are at 0 ALG, and the senors are still reporting variations, but only small amounts. This suggest you landed the plane. At first, I thought the craft was crashed, but at 86.64, we see more motion, and the APM is switched to manual. We still have no GPS fix.
Somehow, you survived the whole flight w/o a GPS fix. I was sure that flight would end in a crash.
You seem to be walking the plane, in hand, somewhere. I can see the accelerometers and gyros are consistent with a human walking through a field while holding the plane at a slight right angle, probably in your right hand.
You set it down at around 88.70%. It is in manual, no GPS fix. The throttle is off, but the battery is down to 15.89v.
The plane sits on the ground for a while. The slow yaw rotation of 1 degree per second can be seen. Most likely that was occurring during the flight, but we cannot separate that false movement from the real movement, since we have no location, so I am just guessing. The plane is clearly not moving now, even if the compass is reporting a rotation of -1 degree a second.
The link quality is down to 64%, but I'm not sure why that is.
At 95.16%, you switch to stabilize, possibly preparing for another flight. At 95.56, the plane begins to move, and you handle or move the transmitter (ch4 in moves slightly) at 96.90%. The plane is in your hands, and you throttle up at 97.30%. Here we go, into the wild blue yonder! Oh, there is no GPS fix, AS 0, GS 0, but we are clearly accelerating forward and up.
It is a gradual climb to 45 ALG in stabilize and then up to 64 AGL, where, at 99.34% with 1 sat count, we switch into AUTO, set to go to WP1 (which is an unknown distance away, as we have no location data.) The plane takes up a 20 degree left bank at about 17% up pitch (what is your stall angle, anyways, on that plane?) It rolls level and loses altitude to 56 (we *were* stalling!) then at 99.60% the plane pitches down hard to 35% (far more than the default parameters normally allow) and the signal is lost, with the last packets showing the plane at 15AGL. It has, no doubt, cratered.
Analysis: Something odd happened during the ground portion and should be investigated. More than one issue seems to be happening then, and the gyros and GPS are affected. Possibly radio interference, or magnetic and radio interference.
However, after the reset, no GPS lock was made. A successful flight was conducted, but AUTO was not working as expected, so you flew in stab and manual. You landed, and then relaunched, but without a GPS lock.
After switching into AUTO, the plane banked left, then leveled off, then nose dived at more than -30 pitch angle from about 56AGL into the ground as fast as the throttle and gravity could take it. I suspect that the telemetry link was degraded, and there is better data on the APM than in this log. There will be a latency. We may find, on the APM's data, that it got a GPS lock around the time it cratered. However, I cannot tell from the signal shown in the APM because it seem to be a counter, and not an absolute value, and gets confused when I scrub forward and backwards in the log file.
I'm sorry for your loss. Did anyone get it on video?
Wow your blow-by-blow analysis is spot on! The reason for the GPS loss is probably because I put it down under a tree. I did not realise I didn't have GPS lock and could not figure out why the plane was wondering out of bounds, so I brought it back to land. I only realised the cause after looking at the logs when I got home.
My BIG concern is why the sensors went crazy while it was stationary. This happend on a previous flight and I had to fight the sabilisation (I thought it was an airframe issue) to bring it down to a crash land.
I'm very worried!!
After all that work, I thought I'd make the effort and edit and upload the video :) http://youtu.be/17GaSx7UbJY. On the video, when you hear the motor speed up - that's when I engaged auto - and ran to the laptop to check where it's going, NOT looking at the plane otherwise I would've been able to save it.
The camera's lens was ripped off so that's it's last video as well!
I don't think the crash was due to the loss of GPS lock - do you (I flew fine the first time sans GPS)? I htink the sensors went mad again.
What could be the cause???
You mean the XBee 900mhz transmission is influencing the APM sensors??
Wow that's quite troubling. It's never happened to me in 40+ flights, HOWEVER it started happening the moment I moved my APM from the main compartment in the fuselage to the top one underneath the wings where the ESC normally goes. Then again, it only seems to happen when the APM is on for a long time (30 minutes). I'll do some tests and confirm. I'm attaching the log to the original post.
Any other suggestions?
I've had a GSM tracker on top of the wings, just above the APM...
is it at all possible that any of the componets got wet from condensation of humidity?
We had something similar happen when flying thru fog.
Just a thought
Thanks Andreas, no. I had a landing in wet(ish) grass but the APM is high and dry under the wing.
its simply a barometer, if the atmospheric pressure was dropping as you ran the test this is interpreted as an increase in altitude by AP. ive seen the same thing in arducopter leaving APM running on the desk for hours.
i dont know how much pressure difference equates to 64m of altitude, but it doesnt seem wildly out of range.