2nd flyaway and crash in 8 months - BEWARE and install SCR!!

[CloudsNeverDie]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

There's something strange happening with the compass at 1:30 in the video. If you look at the little red arrow in the radar at the bottom left, you can see it starts out facing North. Then over about 4 seconds the arrow turns clockwise until it's pointing South. It doesn't look like the Phantom itself actually turned because the gimbal keeps pointing North and you don't see any legs in the video as the compass turns. So this looks like it could have been a compass failure to me... or it could be a gyro failure if the Phantom uses both compass and gyro to figure out its heading (I don't know). Just another possibility to consider.

Was this a Phantom 2 V3, with the new compass design?

 

[LarasLTU]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

There's something strange happening with the compass at 1:30 in the video. If you look at the little red arrow in the radar at the bottom left, you can see it starts out facing North. Then over about 4 seconds the arrow turns clockwise until it's pointing South. It doesn't look like the Phantom itself actually turned because the gimbal keeps pointing North and you don't see any legs in the video as the compass turns. So this looks like it could have been a compass failure to me... or it could be a gyro failure if the Phantom uses both compass and gyro to figure out its heading (I don't know). Just another possibility to consider.

Was this a Phantom 2 V3, with the new compass design?

That was very good observation indeed, I think we all missed it.

I'm not sure about the version, but it was purchased in July, 2014 so definitely not V3.

BTW, I'm still waiting for feedback from vendor, I assume he contacted DJI and possibly is waiting for their response (at least it was the same workflow when I lost V+ for the first time)...

 

[yorlik]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

wow. great catch on the compass! makes one watch it even closer... continuing the idea, your compass seems to stay around 1pm position up to 1:08, then , with the bridge staying in the same spot, the compass turns CCW to 10am position! does the gimbal also stop sideways motion from showing up? If so, this could just be it working its magic. If not, then it began reporting bad data CCW before going off 180 degrees CW....

 

[CloudsNeverDie]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Yeah, good spot with that CCW rotation earlier on in the flight!

All kind of reminds me a little bit of this... it's a different board but the principle is the same and they're all using MEMS gyros I'm sure.

https://www.youtube.com/watch?v=AowEPJUFE-M

 

[LarasLTU]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Very good catch guys - that's the beauty of Screen Recorder as different people can identify very important aspects based on the facts and not only assumptions. I'm sure there should be enough information for DJI to make the investigation, question is if they will be willing to....

I tried to put below some sequence, my actions and recorded video helps to get the parameters from DJI app.

The bridge is practically South - North line and I wanted to have panorama with sun directly from behind.
1:24 - Distance: 10.9m Altitude 103m Speed 0.7m/s, North indicator is at 12:00 position. I'm not touching the right stick, distance is quite constant and GPS is maintaining it. I tilt up the gimbal to get better view.
1:24 - 1:26 Distance: 11.5m Altitude 103.3m Speed 0.3m/s. North indicator is at 12:00 position. I'm turning to the left to have 2 rivers joining on the left on my screen. This tiny move to the left is very similar to the video shared by CloudsNeverDie, when Compass get's wild. It's also interesting, that at 1:28 I get 8 satellites, while it was 7 all the time before.
1:28 - 1:33 Distance 21.1m Altitude 102.5m Speed 2.5m/s, North indicator is at 6:00. That's a moment I realized that it started drifting away very fast to the real North and it wasn't responding to the right stick. I then switched S1 to ATTI mode but Phantom didn't react to the right stick - it was still drifting away.
1:38 Distance 49.5m Altitude 108.3m Speed: 13.8m/s, Norh indicator is at 6:00. I'm not sure why Altitude has increased - it could be that it reacted to my left stick as I was trying not to loose height. Or maybe actual altitude wasn't 108.3m as from video it looks that it already started going down from that moment. I was desperately trying to push left stick up (to get height) and right stick down (to come back) without any success and soon (1:45 I'm shouting if anybody can see it) I wasn't able to see it anymore as it was moving very fast and I was trying to watch the phone for flight parameters.
1:47 Distance 104.8m Altitude 10.1m Speed: 25.7m/s, North indicator is at 4:00. I switched off the controller with the last hope that RTH will help. I wasn't aware that it was only 10m above the ground. By switching the controller off I lost the possibility to see the exact area of the crash - this is something to consider in the future. I was lucky that it landed within 100 meters and Find my Phantom indicated quite accurate position where it crashed - it was right below when connection was lost, but at that time I wasn't sure where it was and was still hoping for the sound of returning Phantom for additional minute.
1:50 Wifi Disconnected
2:10 Wifi Extender battery indicator grayed out

Having those facts it looks to me that Compass was definitely one of the factors impacting this behavior. It clearly shows wrong North. GPS parameters look very valid, it shows the 7-8 satellites all the time, Distance/Altitude/Speed as well.

Frank: do you still believe that GPS was the root cause in this case and that your mod would help to prevent it? I'm really interested in your observations and appreciate your help. I want to thank all the members sharing the ideas and providing the arguments - this is what this forum is intended for, isn't it?

 

[yorlik]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

I am not a pilot but think from your video the circle red v also shows rate of climb and horizontal orientation... never saw descrip of this ib manuals. But watch slight colors in background of the circle.. maybe more data to analyze on actual motion.....

 

[yorlik]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Looking again at the round display and it seems obvious it DOES show IMU data with horizon line.... looks like maybe your IMU was out of control watching this too: I stopped video @ 1:38 - horizon line in circle shows only slight maybe 10 degree tilt to the RIGHT, yet your camera shows full 90 degree tilt TO THE LEFT.... Unless some magnetic anomoly, UFO, or something was near your drone, it seems your recording should be sufficient to warranty this!

 

[flyNfrank]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Frank: do you still believe that GPS was the root cause in this case and that your mod would help to prevent it? I'm really interested in your observations and appreciate your help. I want to thank all the members sharing the ideas and providing the arguments - this is what this forum is intended for, isn't it?

1st let me clarify that it is not a problem with the GPS. It's a plug and connector connection issue that has a effect on the GPS readings. Now based on that, I would put up the deed to my house this is what happened in your case. I will go grab and copy what I just posted in a thread of mine on this issue and paste it here for you below.....

The Plug & Connector for the GPS are the only combo together under the top cover. When the top cover gets removed, that combo is effected. Either the plug gets pulled, or the top cover is flipped over and can dangle from the combo. Both methods apply some kind of overall stress. If you were to look back a few pages from this one, a member has found that corrosion to the copper on both sides of the connection. Using Contact Cleaner has turned up some great results. I believe the members name was "copythat" or something similar. He discovered it after trying everything else.

As I have stated before I have actual data taken from both of the two flight loggers I always use when flying. There is a brief interruption to the gps connection which causes it to become confused on the quads location. The result is, the quad "appears" to switch into Manual Mode and because our setup requires the controller lever to be above the middle position in order to maintain altitude. With the levers position not up would equal not enough applied power, and the quad drops at a angle until either reaching the ground, or if the operator switches the S1 switches out and back into the GPS Mode. I was at 1000ft altitude the 1st time this happened which gave me enought time to figure out I could save it doing the mentioned S1 switch process.

I had the flytrex live and at 1st I thought the problem was related to that devices wire harness. I opened the cover and unplugged the flytrex wire harness and plugged the GPS plug in it's place. However, same results. This is when I used the end of a plastic tie strap to wedge in between the plug and connector to eliminate movement as much as possible. I flew over 25 flights and it never was to happen again.

Now, if you look one page back RichWest put together complete info for a 100% fix to this issue. I plan to go over ALL connections not secure with hot-glue using contact cleaner, and then I'm going to replace the gps plug and leave the connector as is.

 

[Meta4]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Hard to make sense of that explanation.
If the GPS data was wrong and the Phantom believed it was in a distant location why would it dive?
It wouldn't. - it would fly straight and level because the Phantom doesn't get altitude info from GPS.

 

[BlackTracer]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Hard to make sense of that explanation.
If the GPS data was wrong and the Phantom believed it was in a distant location why would it dive?
It wouldn't. - it would fly straight and level because the Phantom doesn't get altitude info from GPS.

If it were trying to fly from point A to point B and point B was on the other side of the earth, would it not try to get there through the earth not around it? Just sayin. Maybe it would take the roundabout way.....

 

[Meta4]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Look at the the image and you'll see that this Phantom has tipped to the right so far that is almost upside down.
On the far right of the screen you can see the motor mount and one of the legs to show you the orientation.
Look at the angle of the horizon on the left.

A phantom doesn't fly anywhere like this - even to the other side of the earth.
It has gyro sensors to keep it level.
The simplest explanation for the cause is that either the flight controller was crazy or a problem with an esc or motor that makes it impossible for the flight controller to maintain level flight.
It's got nothing to do with compass or GPS.

 

[CloudsNeverDie]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

I'm increasingly leaning towards gyro failure. ESC or motor failure doesn't explain the red arrow turning without the craft turning at 1:30ish.

 

[BlackTracer]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Look at the the image and you'll see that this Phantom has tipped to the right so far that is almost upside down.
On the far right of the screen you can see the motor mount and one of the legs to show you the orientation.
Look at the angle of the horizon on the left.

A phantom doesn't fly anywhere like this - even to the other side of the earth.
It has gyro sensors to keep it level.
The simplest explanation for the cause is that either the flight controller was crazy or a problem with an esc or motor that makes it impossible for the flight controller to maintain level flight.
It's got nothing to do with compass or GPS.

This is why you make the big bucks meta4. You are right that wherever it was flying it would not fly with that attitude.

Wouldn't you love to hear what dji software and hardware engineers comments would be with regard to this frame? I sure would.

 

[Meta4]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

This is why you make the big bucks meta4

To justify my charge rate, I'd better throw in a few more to give a more complete view..

 

[yorlik]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Watching recorded video yet again, it is obvious there is NO GPS data glitch all the way thru to loss of signal; speed and distance is rock solid throughout. I cannot see how how anyone would maintain a GPS or GPS connector issue theory with the data solid against it.

Even if it somehow caused a microsecond GPS glitch that did not show up on the recording, I think - but not sure without going out and flying to verify - my distance and speed go away as soon as I leave GPS mode; so that is another clue it was not a GPS or GPS connector issue. I guarantee I would NOT bet the farm on this scenario.

I notice the earth curvature from camera is concave then convex (or vice versa) in last 2 pix; so in one the drone is upside down in the other it is right side up. All in that 1:40 sec so it is rolling pretty fast.

Key possible clue I see from that is the IMU display shows it right side up and NOT upside down or at the heavy angle the other pix shows.

I have spent 40 years as design engineer in servo loops and their usage. This IMU display (lite blue sky above darker blue horizon & acceleration (height of this line above or below center) shows it absolutely does NOT match what is happening. This is the feedback. The feedback information is wrong. So this tells me that the control is NOT at fault - if the controller or computer was at fault the IMU would still show correctly reality as it spins over as shown.

It also seems fairly clear it is NOT any ESC or motor failure since again, the IMU data is totally false in multiple places throughout the recording, and if it was a motor or ESC failure, the IMU data would still be reporting correctly; it is not. So the compass going south and the IMU not following what is really happening proves to me it is an IMU/compass problem, nothing more.

 

[max]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

I notice the earth curvature from camera is concave then convex (or vice versa) in last 2 pix; so in one the drone is upside down in the other it is right side up. All in that 1:40 sec so it is rolling pretty fast.

I agree with everything you've said (looks like IMU failure failure to me), except for the part quoted above. You're just seeing the barrel distortion of the lens having different effects on the horizon depending on where it is in the frame. If the camera was actually rotating you'd see the horizon tilt angle change. Here the camera is just panning to its right. Keep in mind it's on a gimbal so who knows what the quad itself is doing in relation to the camera (we do have some idea due to seeing the arm/landing gear in the frame)

 

[yorlik]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Tnx for correction.

 

[Meta4]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

Yes .. the fisheye curves any straight line (that doesn't pass through the centre), away from the centre.
So the Phantom is tipped way off level flight but we can't see that it is rolling.
The gimbal will attempt to steady the image too - up to a point.
I'm not sure what the artificial horizon in the radar display is doing - that's confusing but I have a feeling it's a symptom rather than causative.
But what we have for sure is a Phantom that has lost thrust on one (front left?) motor and is unable to maintain level flight and falls from the sky.

 

[mendezl]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

good analysis... IMU failure sounds like the most likely cause of the crash.

DJI needs to helps you out gettting a new one, they should since it is under warranty.

 

[LarasLTU]

Re: 2nd flyaway and crash in 8 months - BEWARE and install S

good analysis... IMU failure sounds like the most likely cause of the crash.

DJI needs to helps you out getting a new one, they should since it is under warranty.

Update form my side:
I was asked by seller to fill in RMA form and send phantom to DJI in Germany for investigation. I'm not sure I would like it replaced and I would prefer to get full refund and wait for some time, possibly purchasing V3 latter (as we know they solved/improved much more areas than they are officially advertising) or maybe looking for any alternative. I was checking warranty terms and it's not stated what is the period for the whole Phantom (e.g. it crashes), but only separate parts, which looks strange, isn't it? http://www.dji.com/service/warranty-period

BTW, when I purchased first 2 Phantoms which crashed I was looking for the cheapest option in the market (Europe). Can you share your experience with Amazon as a seller, because my understanding is that they treat customers much more better in general and maybe have even better support/response from DJI as Amazon is a big player. I think I saw somewhere that Amazon even accepts the loss to their credit and refunds easier. My question: is it worth to purchase from Amazon where the price is 100-200 USD higher, but this would possibly give much better support/feedback in case of troubles/warranty?