I'm always quite careful before I takeoff to avoid situations like this. I have a full pre-flight checklist that I follow every time. Tonight I had "safe to fly", and the only thing I noticed is that it took some extra time to go from 4 GPS to 9 GPS as it was sitting idle on the landing pad waiting for my commands. Typically I have a higher count but there were a couple garages to the south that I figured blocked a couple satellites. I waited a couple minutes and it stayed stable at 9 satellites so I started the motors, gave it a moment to update the home point and then took off. From the moment it was in the air it started to tip forward and backward as if i was moving the sticks to go forward and backward or as if it was detecting obstacles. I know it had low light and obstacle detection wouldn't work, but I've never had low light cause it to react to nothing.

I tried to bring it back to me but it ignored all stick inputs and was headed for an open field as you’ll see on the map. I pressed the pause button and it didn’t stop flying forward, then I tried to gain altitude but it only went up a couple feet before clipping the fence surrounding the field and falling down. The motors didn’t stop as DJI always says they will, so I had to do CSC to cut them. I was disappointed that it clipped the fence and crashed at first but in hindsight I now realize that if it didn’t, the drone probably would’ve flown away and I wouldn’t have been able to recover it since it was ignoring all input right after the IMU exception. Thankful I have it to send in for a swap.

Could one of you fine pilots tell me if this was a malfunction or if I just didn't have enough satellites? I have already chosen to skip damage assessment just to get back in the air, Care Refresh Express case is done and I've got a replacement on the way. Just curious for future reference.

http://www.phantomhelp.com/logviewer/HFAYD8IKCQ4OCX3BXNR3/

Thanks all!

Hi MacLover - I had exactly the same issue on the weekend.

This had happened a few times to me in the past, but I thought that it was perhaps just me or a random anomaly caused by where I was flying.  On the weekend however at a height of about 50m, my drone just started to fly away from me.  For about 30 seconds I had no control at all and it was only due to its height that it didn't actually collide with anything.  RTH and Pause on the remote control did not work - RTH producing an error on the remote control screen.

My warning LOG shows the following entries.  I am wondering whether you had the same or whether anyone else might be able to shed some light on the subject.

*IMU Redundancy Switch
*Compass Error.  Exit P-GPS mode
*In Flight, working IMU encounters heading exception, please switch to anti mode if craft behaves abnormally.
*Weak GPS signal.  Positional accuracy may be compromised.  Please fly with caution.

I have also had it in the past where whilst using active track, and whilst not in reach of the remote control, the drone has just literally stopped following me and just taken off into the distance.  Luckily I reached the remote in time and after a few seconds of trying to regain control, managed to get it to return.

Please, anyone else thoughts, experiences or advice would be greatly appreciated.  

Hi, I'm so sorry to hear about the accident. However, glad to know that you have got a replacement on the way, hope you can receive it soon. Take care and fly safely.

Appears to be yet another variation of this case to me... Although here it was the IMU instead of compass that was (mis)diagnosed faultily. The reasons may be hidden rebars in pavement (or some other type of compass interference), uncalibrated compass and/or IMU, etc. These drones should not go that mad because of compass errors, but it seems they do. I wonder how they suppose me to "switch to atti mode if craft behaves abnormally". I'd love to, but there is no way of doing so, unfortunately.

Very classical case: Weak GPS link and as it looks a good breeze of wind. Although the flightrecord indicates a GPS link at take off time it was too weak. 11 satellites are the absolute minimum for flying safely and if you can't get them where you are, change location. The movements at take off time were related to VPS not working because of darkness. Later you might have had the impression that it was ignoring your input but this was caused by the ATTI mode where only 100% stick moves will have an impact.
A pre flight check list is a very important thing, but it can only be useful if strictly adhered to. It can happen that abandoning a mission is the only option.

At 8 seconds throttle up aircraft rises to 6.2ft
At 11.9 seconds throttle down aircraft drops altitude to 5.2 ft
At 13.8 seconds throttle up aircraft rises to 10.2 ft
At 15.6 seconds throttle up aircraft rises to 12.1 ft

Above you can see that aircraft was responding to your stick movements, but weak gps and aircraft going to Atti mode you needed much stronger stick movements, but even then you would have been restricted to max height of 16 ft above take off.

At 11.2 to 11.9 you can see very small movements on elevator stick but it is having no effect on aircraft which was now in Atti mode and moving in another direction taking aircraft away.

I agree with asaw that it could have picked up interference on the ground but not enough to report compass problem but enough to cause conflict with IMU .

As you can see it was correctly responding to stick movements by the rise and drop in altitude, it was in Atti mode and would have had little or no horizontal hold so was drifting faster than stick movements you used to counteract this involuntary movement caused by going to Atti mode.

I believe this aircraft would have been controllable in an different envoirment, but looking at the area you took off from this made it very difficult.

We already know that problems with compass when placed near or on top of metal or rebar or ground with underground cables will effect compass and cause aircraft to have knock on problems with IMU causing aircraft to forget about gps leaving it to fly in Atti mode.

If you don’t mind could you explain where you took off from or had you calibrated recently.

asaw Posted at 2018-5-7 02:25
Appears to be yet another variation of this case to me... Although here it was the IMU instead of compass that was (mis)diagnosed faultily. The reasons may be hidden rebars in pavement (or some other type of compass interference), uncalibrated compass and/or IMU, etc. These drones should not go that mad because of compass errors, but it seems they do. I wonder how they suppose me to "switch to atti mode if craft behaves abnormally". I'd love to, but there is no way of doing so, unfortunately.

Although it says switch to Atti mode as it does in both spark and Mavic Pro its a bit of an anomaly and should be replaced by aircraft is switching to Atti mode, it’s been like that since Mavic was first introduced and requests have been made to change it. This case did go to Atti mode regained gps and again went to Atti mode, and this is the purpose of having auto Atti in these aircraft so if your in Atti mode and gps returns you won’t have to continue to keep switching to check if gps has returned as we know was the case with Phantom 1 & 2.

You can also see the anomaly in phantomhelp which doesn’t report magnetic or compass problem, while his own log does, we also seen some of these anomalies in the case you pointed out but we never seen the OP’s flight log until dji made a decision, I firmly believe that phantomhelp is still not correctly aligned with Mavic Air yet.

Wachtberger Posted at 2018-5-7 02:43
Very classical case: Weak GPS link and as it looks a good breeze of wind. Although the flightrecord indicates a GPS link at take off time it was too weak. 11 satellites are the absolute minimum for flying safely and if you can't get them where you are, change location. The movements at take off time were related to VPS not working because of darkness. Later you might have had the impression that it was ignoring your input but this was caused by the ATTI mode where only 100% stick moves will have an impact.
A pre flight check list is a very important thing, but it can only be useful if strictly adhered to. It can happen that abandoning a mission is the only option.

If it was just the wind and atti mode then there should bee clear connection between stick input and drone action. Now I don't know if the FW had been updated without IMU anc compass calibration, but there is something more going on here.

G_Sig Posted at 2018-5-7 03:07
If it was just the wind and atti mode then there should bee clear connection between stick input and drone action. Now I don't know if the FW had been updated without IMU anc compass calibration, but there is something more going on here.
[view_image]
[view_image]

You are taking these readings from phantomhelp Help although in slightly different format, OP’s flight log reported compass and magnetic problems also,  but not seen in your report.

hallmark007 Posted at 2018-5-7 02:59
Although it says switch to Atti mode as it does in both spark and Mavic Pro its a bit of an anomaly and should be replaced by aircraft is switching to Atti mode, it’s been like that since Mavic was first introduced and requests have been made to change it. This case did go to Atti mode regained gps and again went to Atti mode, and this is the purpose of having auto Atti in these aircraft so if your in Atti mode and gps returns you won’t have to continue to keep switching to check if gps has returned as we know was the case with Phantom 1 & 2.

You can also see the anomaly in phantomhelp which doesn’t report magnetic or compass problem, while his own log does, we also seen some of these anomalies in the case you pointed out but we never seen the OP’s flight log until dji made a decision, I firmly believe that phantomhelp is still not correctly aligned with Mavic Air yet.

The message "In Flight, working IMU encounters heading exception,please switch to atti mode if craft behave abnormally" is at 0m 14.7s. The craft already behaves abnormally at this time as pitch stays negative no matter what pilot does with the elevator stick (moves it up and down). And the aircraft first goes into ATTI mode only at 0m 18.9s. So the way how it does it automatically is insufficient.

asaw Posted at 2018-5-7 03:15
The message "In Flight, working IMU encounters heading exception,please switch to atti mode if craft behave abnormally" is at 0m 14.7s. The craft already behaves abnormally at this time as pitch stays negative no matter what pilot does with the elevator stick (moves it up and down). And the aircraft first goes into ATTI mode only at 0m 18.9s. So the way how it does it automatically is insufficient.

At 14.7 no elevator stick movement , throttle yes and aircraft responds you can see it climb lower and climb again this is correct.
op said his warning said magnetic warning  compass error exit gps, at 18 seconds Aircraft was upside down.
Are we to think phantomhelp is right and OP’s log is wrong, but that is his only log at the time and reporting through his telemetry.

You are seeing it late and insufficient because you are looking at phantomhelp log.

hallmark007 Posted at 2018-5-7 03:35
At 14.7 no elevator stick movement , throttle yes and aircraft responds you can see it climb lower and climb again this is correct.
op said his warning said magnetic warning switch compass error exit gps, at 18 seconds Aircraft was upside down.
Are we to think phantomhelp is right and OP’s log is wrong, but that is his only log at the time and reporting through his telemetry.

> At 14.7 no elevator stick movement

Because they had been before 14.7! Even the flight controller figured something was going wrong by that time!

asaw Posted at 2018-5-7 03:41
> At 14.7 no elevator stick movement

Because they had been before 14.7! Even the flight controller figured something was going wrong by that time!

At 14.7 there was throttle stick movement and aircraft responded by going up the down this is correct, this also goes along with the OP’s statement that he tried to climb to clear the fence, and op says his reports were exit gps. You can also see earlier movements with elevator that shows aircraft continuing on its direction but these stick movements were extremely small for aircraft to show any movement in Atti mode as it was moving faster than the stick movements could or would make any difference.
Atti mode as you well know demands much more positive stick movements to control aircraft, and it is always advised to immediately land when you get warning in this mode and all other warnings demand the same of the controller.

What you seem to be only looking at here is phantomhelp log and ignoring what OP said his telemetry reported, and if you only use what phantomhelp says then yes going to Atti mode was insufficient, but if you take what was reported by OP then what your saying might not be correct, and if you look at the attitude of aircraft in the first 15 seconds it had very weak gps and had no horizontal hold.

Again I would like to ask you to take a look at throttle movements to see if you think that aircraft did not respond to those movements, from what I see right up until the crash it did.

hallmark007 Posted at 2018-5-7 04:01
At 14.7 there was throttle stick movement and aircraft responded by going up the down this is correct, this also goes along with the OP’s statement that he tried to climb to clear the fence, and op says his reports were exit gps. You can also see earlier movements with elevator that shows aircraft continuing on its direction but these stick movements were extremely small for aircraft to show any movement in Atti mode as it was moving faster than the stick movements could or would make any difference.
Atti mode as you well know demands much more positive stick movements to control aircraft, and it is always advised to immediately land when you get warning in this mode and all other warnings demand the same of the controller.

hallmark007, stop making up stories. Attitude mode doesn't require any pilot input at all for the aircraft to remain stable. It also doesn't require any input for the aircraft to hold its position if there is no wind (and own downwash doesn't have an influence on it either). Pilot's input is only needed to maintain position if there is wind. In this case the aircraft's ground speed was negligible up until 15s. Besides, it is only your speculation that the AC had gone into ATTI mode before 18s.

asaw Posted at 2018-5-7 04:15
hallmark007, stop making up stories. Attitude mode doesn't require any pilot input at all for the aircraft to remain stable. It also doesn't require any input for the aircraft to hold its position if there is no wind (and own downwash doesn't have an influence on it either). Pilot's input is only needed to maintain position if there is wind. In this case the aircraft's ground speed was negligible up until 15s. Besides, it is only your speculation that the AC had gone into ATTI mode before 18s.

In attitude mode aircraft will not move only if there is no wind , as controller increased height and as we all know that the higher we go the more wind there is also controller is creating propulsion by simply raising his aircraft and when he does this you can clearly see aircraft moving away from homepoint, so for you to assume that there was no wind is just stretching it a little bit.

The OP has already said he got reports that compass error exit gps you are completely ignoring this simply because it doesn’t suit your point, so if controller gets compass error exit gps switch to Atti mode this has no meaning to you.

And again I will point out to you that quadcoptors are much more difficult to control without technology than helis .

The only difference between here and the other case you pointed to is OP has given us the correct warnings he received, you choose to ignore these because it doesn’t suit what you think should happen, but what I’ve already pointed out to you on several occasions is what happens and you continue to ignore this although it’s staring you in the face, WARNING COMPASS ERROR EXIT GPS.

The OP also goes on to explain how when flying at 50 m similar happened but he didn’t crash because he was above obstacles and if this flight was in a better envoirment then we may have seen a better result.

And again you refuse to look a stick movements and how aircraft responded because it doesn’t suit you, you refuse to see that OP got warnings, and you expect us to accept that you know that there was absolutely no wind at all completely calm, but no explanation for this because it suits you.
It’s easy to make up stories when you leave all that out of the equation.

hallmark007 Posted at 2018-5-7 04:40
In attitude mode aircraft will not move only if there is no wind , as controller increased height and as we all know that the higher we go the more wind there is also controller is creating propulsion by simply raising his aircraft and when he does this you can clearly see aircraft moving away from homepoint, so for you to assume that there was no wind is just stretching it a little bit.

The OP has already said he got reports that compass error exit gps you are completely ignoring this simply because it doesn’t suit your point, so if controller gets compass error exit gps switch to Atti mode this has no meaning to you.

> The OP has already said he got reports that compass error exit gps you are completely ignoring this simply because it doesn’t suit your point

No, I am completely ignoring this because it was another case with another person and the OP didn't write anything about his compass.

asaw Posted at 2018-5-7 04:55
> The OP has already said he got reports that compass error exit gps you are completely ignoring this simply because it doesn’t suit your point

No, I am completely ignoring this because it was another case with another person and the OP didn't write anything about his compass.

He did write about his compass otherwise how would we know.

Again I have tried to explain how these Aircraft work and what happens in similar situations.
You are coming at this on the back of your heli experience and telling us how it should work based on that, this is not a heli and does not work in the same way in Atti mode or in many other ways, you have already seen two very similar situations with very similar reactions from aircraft on this case everything was reported to controller in other case we never seen what was reported but my guess is it was similar simply because dji found the same cause right from the beginning.

There is no point in me trying to further explain what happened, because your on a different route than me, I’m talking about what happens and your talking about what you think should happen.

I urge you to go buy a quad which only has Atti mode with Camera preferably and post the video you get from the smooth flying and control you can master it with. Look forward to seeing, but as for this case as I said it’s similar to other case with same results, both could have been avoided very easily.

hallmark007 Posted at 2018-5-7 05:15
He did write about his compass otherwise how would we know.

Again I have tried to explain how these Aircraft work and what happens in similar situations.

I am approaching this based on my background in aviation, not only the RC helicopters. And based on that background, a quadcopter is a type of helicopter. So you can even swap flight controllers between them (given the quadcopter has a suitable ESC logic board).

And I am analyzing the flight log instead of making up stories how the aircraft went into ATTI mode when the log didn't show it, like you do here.

Ok name one dji Aircraft you can change RC without modifying?

The op got warning COMPASS ERROR EXIT GPS SWITCH TO ATTI MODE, he got these warnings from his dji data, yes dji data, yet you continually say he didn’t, that’s what they call ignoring the facts. It’s really strange that you continue to dispute this.

hallmark007 Posted at 2018-5-7 05:53
Ok name one dji Aircraft you can change RC without modifying?

The op got warning COMPASS ERROR EXIT GPS SWITCH TO ATTI MODE, he got these warnings from his dji data, yes dji data, yet you continually say he didn’t, that’s what they call ignoring the facts. It’s really strange that you continue to dispute this.

I didn't mention RC's (with these you can usually use the same unit to control your submarine, car, glider, and helicopter, not the case with DJI of course). I was talking about the flight controller - that's how DJI and other manufacturers call the autopilot that flies their drones.

What post did you take the warning from?

I appreciate all the replies. I agree this was likely due to a lack of sufficient GPS, but I agree with G_Sig, it didn’t seem like it was solely in Atti mode. Even at the very beginning when it started flying it was rocking back and forth aggressively and stick inputs didn’t change it. It was tipping forward and backwards a couple times a second with no serious wind. There was a little which I understand would’ve required sport and 100% stick input to counter but I knew it was more than just Atti mode when it started rocking and making extremely abnormal noise.

I had not updated to the latest firmware that came out in the last 48 hours, was still running the last stable version and had been flying for weeks in windy conditions with no issues.

maclover009 Posted at 2018-5-7 06:14
I appreciate all the replies. I agree this was likely due to a lack of sufficient GPS, but I agree with G_Sig, it didn’t seem like it was solely in Atti mode. Even at the very beginning when it started flying it was rocking back and forth aggressively and stick inputs didn’t change it. It was tipping forward and backwards a couple times a second with no serious wind. There was a little which I understand would’ve required sport and 100% stick input to counter but I knew it was more than just Atti mode when it started rocking and making extremely abnormal noise.

You can replay your flight log in DJI GO 4 on your mobile device to make sure when exactly it switched to ATTI (or maybe OPTI) mode. I must agree with hallmark, Phantomhelp doesn't always detect this correctly, but DJI GO 4 does.

Just checked. Looks like it didn’t go to atti until 18 seconds when it crashed. The ENTIRE time it was jerking back and forth and non responsive it was in GPS, with the IMU heading exception error. Nothing about compass. That was a different post. It only entered Atti mode once it clipped the fence and flipped. So what do we think caused this? Again I’ve flown in Atti mode before and this wasn’t it. I wish I could explain better what I meant by “jerking” but it was literally as if it was in sport mode and I was moving the right stick 100% up and 100% down every second.

What’s really weird though is it shows the aircraft facing me the entire time as it was moving away, but I had the green status light flashing me, not the reds. So it definitely thought it was facing the opposite direction. I do see where I pulled the right stick down 100% in DJI flight viewer to try and get it to come away from the fence but maybe that’s what made it shoot into the fence since the compass was flipped? I checked for errors before takeoff and had none. I always check.

You can see in these two photos what I’m referencing. At 8 seconds I try to move out and up into the street, at 11 seconds is my “oh crap” moment where I right stick 100% back to try to pull it away from the fence. At 8 seconds the forward stick made it move into the street, and at 11 seconds it sped up versus reversing as you can clearly see I called for.

Also, contrary to those that say I didn’t do 100% stick, here’s before it crashed where I had a couple good seconds of 100% left stick up and it was still in GPS. It alerted to the heading exception at this time and then crashed 2 seconds later switching to Atti mode once it landed upside down on the ground.

I know some of you may question whether or not it was actually facing me and just amature hour pilot error but you can see when you drag the timeline back and forth in flight viewer that even with right stick forward it moves away from it. If it was indeed facing me, right stick forward would’ve put it into a building versus the street, and when I did right stick back it would’ve done the reverse of that. Both inputs take the craft away from me into the fence.

maclover009 Posted at 2018-5-7 06:34
Just checked. Looks like it didn’t go to atti until 18 seconds when it crashed. The ENTIRE time it was jerking back and forth and non responsive it was in GPS, with the IMU heading exception error. Nothing about compass. That was a different post. It only entered Atti mode once it clipped the fence and flipped. So what do we think caused this? Again I’ve flown in Atti mode before and this wasn’t it. I wish I could explain better what I meant by “jerking” but it was literally as if it was in sport mode and I was moving the right stick 100% up and 100% down every second.

Based on the yaw, pitch, and roll readings between 15 and 20 seconds the IMU was malfunctioning. Whether it could have been caused by compass or not only DJI can tell. Common sense suggests it shouldn't have, but weird things sometimes happen with these drones.

Can you confirm at what time did it initially hit the fence? After 20s I guess?

It hit the fence right at the end of 17 seconds going into the 18 seconds. Started to spin around and do the “ive crashed” dance and then ended up in the street.

maclover009 Posted at 2018-5-7 06:59
It hit the fence right at the end of 17 seconds going into the 18 seconds. Started to spin around and do the “ive crashed” dance and then ended up in the street.

Then these violent oscillations in readings were caused by the drone's hitting the fence and the IMU had been fine...  Maybe flight controller misdiagnosed IMU as faulty based on wrong compass readings and then went nuts.

Either way, the drone thought it was facing me and applied both forward and backwards input as forward motion towards the fence. You can see the compass facing the wrong direction from takeoff, which is another thing I’ll be sure to check going forward before starting motors.

What is other B in the log ???

I think it’s actually a number 8, unknown error code 8. Wish I knew as well.

maclover009 Posted at 2018-5-7 06:44
Also, contrary to those that say I didn’t do 100% stick, here’s before it crashed where I had a couple good seconds of 100% left stick up and it was still in GPS. It alerted to the heading exception at this time and then crashed 2 seconds later switching to Atti mode once it landed upside down on the ground.

While it’s not ideal it’s the reality of what happens when you have problems with compass and then in turn IMU it’s a process your aircraft goes through when it’s getting conflicting data, it is very common that this occurs because of magnetic interference usually picked up on the ground, and this is why it’s important to have a preflight routine, the actions of a lot who fly these aircraft is as soon as they get good gps, they take off like nothing can go wrong, thank you for putting up both your log and information, as it will help others who read or study what can happen.
I also believe in an open field you may have eventually got more control, but hey.
This below might help simple exercise which all commercial flyers use all the time.

Tip To Help Avoid compass interference and crash.


1/ Never calibrate Compass unless prompted to.

2/ start AC leave until you receive enough gps lock ,

3/ if you get Compass interference turn off AC and move to another location,
without interference.

4/ In bottom left hand corner on your map you will see small red triangle, check to make sure that this triangle is pointing in the same direction (heading) as your AC, this will show good compass on the ground.

Raise AC to height of 8ft
Hover for 20 seconds
Fly forward 2ft
Backward 2ft
Left 2ft
Right 2ft
Up 2ft
Down 2ft
Yaw left
Yaw right
Each time returning to hover position
If you have a good horizontal each time, you will then know you have good GPS good IMU and good Compass, and your ready to fly.

This exercise takes less than 2 minutes and is worthwhile doing before every flight,

hallmark007 Posted at 2018-5-7 03:11
You are taking these readings from phantomhelp Help although in slightly different format, OP’s flight log reported compass and magnetic problems also,  but not seen in your report.

I don't see any of the usual compass, magnetic, or yaw errors in the logs.  Where are you seeing them?

KlooGee Posted at 2018-5-7 09:31
I don't see any of the usual compass, magnetic, or yaw errors in the logs.  Where are you seeing them?

The OP said he received through his app compass error, exit gps, switch to Atti , this is the second case I have now seen where phantomhelp does not report warnings received with MavAir, and also I’ve seen recently similar case with Mavic Pro.

I do think the phantomhelp is not fully up to speed yet for Mavic Air.



IMU Redundancy Switch
*Compass Error.  Exit P-GPS mode
*In Flight, working IMU encounters heading exception, please switch to anti mode if craft behaves abnormally.
*Weak GPS signal.  Positional accuracy may be compromised.  Please fly with caution.

maclover009 Posted at 2018-5-7 06:44
Also, contrary to those that say I didn’t do 100% stick, here’s before it crashed where I had a couple good seconds of 100% left stick up and it was still in GPS. It alerted to the heading exception at this time and then crashed 2 seconds later switching to Atti mode once it landed upside down on the ground.

You also had 4 out of 5 bars of GPS "signal level". And this is what it should mean:

LEVEL_0        The GPS has almost no signal, which is very bad.
LEVEL_1        The GPS signal is very weak.
LEVEL_2        The GPS signal is weak. At this level, the aircraft's go home functionality will still work.
LEVEL_3        The GPS signal is good. At this level, the aircraft can hover in the air.
LEVEL_4        The GPS signal is very good. At this level, the aircraft can record the home point.
LEVEL_5        The GPS signal is very strong.
NONE        There is no GPS signal.

hallmark007 Posted at 2018-5-7 09:37
The OP said he received through his app compass error, exit gps, switch to Atti , this is the second case I have now seen where phantomhelp does not report warnings received with MavAir, and also I’ve seen recently similar case with Mavic Pro.

I do think the phantomhelp is not fully up to speed yet for Mavic Air.

Is this normal reading?

G_Sig Posted at 2018-5-7 10:11
Is this normal reading?
[view_image]

Not understanding lots of course lock there, I’m certain MavAir doesn’t have Course Lock YET.