Welcome Inspire Pilots!
Join our free DJI Inspire community today!
Sign up

I2 Go 4 App apparently destroyed an iPad Air after FW upgrade

Joined
Mar 4, 2014
Messages
176
Reaction score
62
This is cross-posting from forum.dji.com because of the severity of the problem (Editor: please forgive me!)

Last night, I was using the I2, props removed, for about two hours, just familiarizing myself with the Go App 4. Everything was working flawlessly. I took numerous screen captures just to remind me where various icons were.

Today when I fired everything up, the GO 4 app indicated that I had "incompatible firmware" and that I had to upgrade it. I operated the "slider" switch and the firmware upgrade appear to complete correctly, albeit with a puzzling message about disconnecting and reconnecting the data cable (as though I had been using DJI Assistant rather than the Go app).

Therefore, I power cycled the I2, the RC, and the iPad Air.

Note: I do not know which version(s) of firmware were installed where last night (I hadn't got that far), nor do I know what the compulsory firmware update this afternoon installed. There was no indication of version numbers during the update.

The Go 4 app then indicated that the compass needed recalibrating (which didn't seem unreasonable given the firmware upgrade), so I went outside and completed the compass calibration. Just as I was coming back indoors (with the aircraft, RC, and iPad all still powered up, I got the "Mobile device CPU fully loaded. Related performance will be affected." I had it on screen just for long enough to do a screen capture. At which point the display on the iPad turned into a series of vertical colored lines and went black.

Now, no matter what I do, it appears that the iPad Air has been damaged and will not even start up. so I have no option to but to take it to the Apple store to have it repaired (of course, it's out of Applecare by now).

I'm very disturbed to see that DJI's firmware is "over-driving" the iPad so badly that it will destroy itself -- especially because I was using the setup flawlessly last night for two hours. It would suggest that the new firmware -- whatever version it is, has not been adequately tested and is capable of actual destructive behavior. I've searched the forum and can see reference to "Mobile device CPU fully loaded," but I have not been able to find out whether anyone else's iPads have been irreparably damaged. Anyone else had this problem?

I do have a second iPad Air, but I hardly dare connect it for fear it will destroy a second iPad.

Thanks in advance
Andy
 
Last edited:
As posted on the other forum it is not possible for the Go app or any app to damage your IPad, it's simply not possible on many levels.

You either have a damaged screen or lose screen cable or it's hw failure, there is a lot of protection o such as thermal throttling to prevent damage with heat.

That's not to say though using it in a case or leaving it in direct sunlight would not do it.

Have you tried pressing and holding both the power and home buttons together for 10 seconds ?
 
  • Like
Reactions: cvanhoose
Thanks for posting here as well, Mad_angler1. For the convenience of other forum members, I'll post my same reply from the other forum:

I would like to agree with you that it is not possible. I am merely recounting what happened.

It sounds like you have some evidence to suggest that it is not possible for an app. to "overdrive" hardware and thus cause damage. Can you share this information please? I am a forensic software analyst (I analyze computer-based evidence and testify about it as an expert in Court), so such evidence would be very helpful to know.

The screen was not damaged and, as far as I can tell, there was nothing that would cause a cable to come loose. It was not left in direct sunlight -- in fact immediately after the failure, the back of the iPad had no "hot spots" and did not feel particularly warm.

I have tried single button restarting, double button restarting, connecting up to a Mac (to see if iTunes would start up), and also connecting to a charging cable (to see if that would wake it up). The iPad was 100% charged and had been in used for about 20 minutes before it shut down. I have an appointment with the local Apple Store Genius bar tomorrow afternoon so I should know more once they have inspected it.

Thanks
Andy
 
As posted on the other forum it is not possible for the Go app or any app to damage your IPad, it's simply not possible on many levels.

You either have a damaged screen or lose screen cable or it's hw failure, there is a lot of protection o such as thermal throttling to prevent damage with heat.

That's not to say though using it in a case or leaving it in direct sunlight would not do it.

Have you tried pressing and holding both the power and home buttons together for 10 seconds ?

100 percent correct information.

OP may just have had a coincidence of issues at once.
 
When I saw the "incompatible firmware" it was a firmware update for the battery in the I2 and I was using an Air 2 at the time. Later on, I got another one, but it was for the focus wheel I have plugged into the RC and saw the blinking blue LED on the focus wheel so I'm guessing that was the "incompatible firmware" message for it.

DJI isn't really good at explaining what they are trying to tell you at times, imho, with "incompatible firmware." Somehow, it might be coincidence if your iPad is truly bricked. Also, I think the older iPad AIr is no longer working with the newest GO 4, but someone else may know better (DJI's approved list is getting too old too!). My Air 2 is having some connection issues now in the latest GO version where it never did prior. More connection errors too where I have to unplug it and replug the USB cable too to get a green Connected message in GO 4 or I get the red Disconnected message.

Good luck sorting it out.
 
100 percent correct information.

OP may just have had a coincidence of issues at once.

Interesting point, cvanhoose. Do you have any additional information, please? From an evidentiary point of view, separating causation from coincidence on a limited number of samples is very hard. You seem pretty certain that it is just a coincidence (and I really do not have enough information to disagree with you beyond the sequence of events that I saw).

It would be very helpful to understand why you feel that it is "100 percent correct" that it cannot have been the Go 4 app overdriving the hardware to the point of failure? Can you help with some additional details please?

Thanks in advance
Andy
 
Last edited:
Somehow, it might be coincidence if your iPad is truly bricked. Also, I think the older iPad AIr is no longer working with the newest GO 4, but someone else may know better

Thanks for posting Casey53 -- I agree, at this point all I know is the sequence:

1. Worked fine yesterday evening for two hours before firmware upgrade.
2. Mandatory firmware upgrade.
3. After 20 minutes, "Mobile CPU fully loaded."
4. iPad display disintegrates into a series of colored vertical lines.
5. iPad did not feel unusually warm to the touch anywhere on rear surface.
6. iPad died.

I'm hoping someone can provide additional evidence for similar (or identical experiences) so that I might be able to separate coincidence from causality.

Thanks in advance
Andy
 
I have an appointment with a Genius at the local Apple store this afternoon (Pacific Time). Beyond that, nothing new to report.

Before it died, the Go 4 app displayed a peculiar message (separate from the Mobile CPU message. It stated something like "incorrect position of batteries, please re-align." This is from memory (of a colleague and me) -- we took a screen capture of it, but it is entombed in the dead iPad. We both saw the message but it scrolled off the screen when the Mobile CPU message occurred.

So that's a second order question: What does such a message mean? Why would it appear after 20 minutes of use?

Cheers
Andy
 
Last edited:
Thanks for posting, cvanhoose. You may have misunderstood me -- I was not attempting to start a debate.
I'm just seeking information on possible causes, the better to understand the nature of the failure.

From your knowledge of the iPad, and presuming I am correctly understanding your posting, there is no means from within a software driver to misconfigure or overload the GPU. On the model A1475 (iPad Air 1, late 2013/2014), it looks like it's the A8X + GXA6850 GPU, system on a chip ("SOC")).

Based on your knowledge, if you max out the A8X (or the GPU for that matter), from what you say, the iPad should signal the app that it's going into thermal shutdown/self-protect -- that appears to have worked (hence the Mobile CPU message from the Go 4 app).

It's what happens after that that is unclear. I appears as though the GXA6850 GPU may have died -- the screen turning into a series of colored lines (running across the short axis of the screen) might be evidence of that. At the time, the RC and iPad were resting on a table inside our building. My colleague had just taken a screen capture of the Mobile CPU message. That capture occurred, and then we saw the colored lines before the screen went black.

From what you say, therefore, it appears that the fault might be in the thermal protection system that failed to shut down the SoC fast enough to protect itself?

Thanks
Andy
 
Last edited:
Will be interesting to see what the Apple gurus say.

That "Mobile CPU fully loaded" message out of GO 4 might be telling if DJI is demanding too much of the GPU in the old iPad Air. That message is one DJI was supposedly addressing, but given that GO is getting to the point you're gonna need something along the line of a Cray Supercomputer to run who knows? I think GO 4 is around a gig in memory now and it may be too much for these handheld things, short of going to a new iPad Pro with the A10X.

Good luck at the Apple Store, but likely you may need a new one too.
 
The software can not contain a script or trigger to overload the cpu or gpu. The iOS is sandboxed and it is nearly impossible to get an app approved unless you follow the guidelines and protocols. You must use the Apple core libraries when developing.

Unfortunately you may never know what happened unless you get Apple to re-micro solider the memory chip to a new iPad and grab the logs, thats if you have the log collection enabled.

I recommend the newest iPad and iPad Pro for its GPU and CPU. I also have used my iPhone7+ and the performance was par for go4 w/Mavic. Crystal Sky is the way to go, dedicated encoder.
Good Luck.
 
Thanks, Chris. I'll be asking the Genius whether it's feasible to get Apple to do a memory-chip-ectomy to a new SOC board.

I'm planning to get an iPad Pro 10.5 while I'm down there (and with Applecare!)

What kind of battery life are you seeing with Crystal Sky (either model)? Not quite clear from the documentation I've ready on the Crystal Sky. The manual indicates it can take either 2 x CS785 or one CS550.

I had thought of using a Small HD DP7 Hi Brightness -- I used them on the ALTAs I fly. It would need to connect up to the HDMI, port of course, not the USB port. The downside is they suck power -- around 24 watts.

Off to see a Genius!
Andy
 
The visit to the Genius Bar at the Apple store resulted in the following conclusion:

  1. More likely than not, the Go app's computational load on the CPU/GPU load caused one or both of the chips on the SoC to overheat leading to a thermal self-protect (hence the "Mobile CPU fully loaded message").
  2. The need to do the thermal slowdown/shutdown without the presence of a hot spot on the back suggests a very rapid rise in temperature.
  3. The CPU/GPU load may have been further exacerbated by taking a screen shot of the Mobile CPU message -- but that's not at all clear.
  4. It is a mystery why, immediately after this, the iPad died completely, suggesting (but not confirming) that the thermal shutdown mechanism did not act rapidly enough to prevent damage to the CPU/GPU.
  5. I cannot recover the screen captures made leading up to the death of the CPU/GPU but I can get a replacement iPad Air 1 for $250 (Applecare had expired -- as often seems to be the case! :) ).

Alternative explanation: The SoC was on the brink of failure (who knew? :) ) and being taken to the limit running the compute intensive Go 4 app had the side-effect of pushing it over the edge.

Was the Go 4 app causal? Unknown. It was the only app. that was running and was computationally intensive -- and the iPad was in Airplane Mode so not much else should have been loading the CPU -- other than, of course, iOS itself.

Oh...a new gotcha: The only new, fast, iPad Pro that seems to be approximately the right size for the RC is the iPad Pro 10.5 -- but it's 5mm too wide to fit in the tablet holder, so I'll need to get an adapter for the RC. There seem to be several around -- mostly 3D printed ABS as far as I can tell.

Andy.
 
The visit to the Genius Bar at the Apple store resulted in the following conclusion:

  1. More likely than not, the Go app's computational load on the CPU/GPU load caused one or both of the chips on the SoC to overheat leading to a thermal self-protect (hence the "Mobile CPU fully loaded message").
  2. The need to do the thermal slowdown/shutdown without the presence of a hot spot on the back suggests a very rapid rise in temperature.
  3. The CPU/GPU load may have been further exacerbated by taking a screen shot of the Mobile CPU message -- but that's not at all clear.
  4. It is a mystery why, immediately after this, the iPad died completely, suggesting (but not confirming) that the thermal shutdown mechanism did not act rapidly enough to prevent damage to the CPU/GPU.
  5. I cannot recover the screen captures made leading up to the death of the CPU/GPU but I can get a replacement iPad Air 1 for $250 (Applecare had expired -- as often seems to be the case! :) ).

Alternative explanation: The SoC was on the brink of failure (who knew? :) ) and being taken to the limit running the compute intensive Go 4 app had the side-effect of pushing it over the edge.

Was the Go 4 app causal? Unknown. It was the only app. that was running and was computationally intensive -- and the iPad was in Airplane Mode so not much else should have been loading the CPU -- other than, of course, iOS itself.

Oh...a new gotcha: The only new, fast, iPad Pro that seems to be approximately the right size for the RC is the iPad Pro 10.5 -- but it's 5mm too wide to fit in the tablet holder, so I'll need to get an adapter for the RC. There seem to be several around -- mostly 3D printed ABS as far as I can tell.

Andy.

Were they able to boot the device and see the log? A log will have to be gathered from the device or the iPad will need to be dissected for a conclusion.
FYi most genius from those bars do not know the mechanics of the devices. They have 3-4 steps of diagnosis to go down from a list that is supplied to them after you state your issues.
 
The visit to the Genius Bar at the Apple store resulted in the following conclusion:

  1. More likely than not, the Go app's computational load on the CPU/GPU load caused one or both of the chips on the SoC to overheat leading to a thermal self-protect (hence the "Mobile CPU fully loaded message").
  2. The need to do the thermal slowdown/shutdown without the presence of a hot spot on the back suggests a very rapid rise in temperature.
  3. The CPU/GPU load may have been further exacerbated by taking a screen shot of the Mobile CPU message -- but that's not at all clear.
  4. It is a mystery why, immediately after this, the iPad died completely, suggesting (but not confirming) that the thermal shutdown mechanism did not act rapidly enough to prevent damage to the CPU/GPU.
  5. I cannot recover the screen captures made leading up to the death of the CPU/GPU but I can get a replacement iPad Air 1 for $250 (Applecare had expired -- as often seems to be the case! :) ).

Alternative explanation: The SoC was on the brink of failure (who knew? :) ) and being taken to the limit running the compute intensive Go 4 app had the side-effect of pushing it over the edge.

Was the Go 4 app causal? Unknown. It was the only app. that was running and was computationally intensive -- and the iPad was in Airplane Mode so not much else should have been loading the CPU -- other than, of course, iOS itself.

Oh...a new gotcha: The only new, fast, iPad Pro that seems to be approximately the right size for the RC is the iPad Pro 10.5 -- but it's 5mm too wide to fit in the tablet holder, so I'll need to get an adapter for the RC. There seem to be several around -- mostly 3D printed ABS as far as I can tell.

Andy.

Hello Andy - this is Michael McVay replying...long time Freefly Forum member (although I think I have fallen off of the most notable members list recently). We are a DJI distributor and help with a lot of issues people have with their DJI products. We supply a number of our long time Freefly friends with their DJI stuff.

While I certainly can not say for certain what happened in your case, iPads have been routinely getting the CPU Fully Loaded message and then having some video breakup issues that appear as green lines, green pixels, blocky or frozen video, etc. A few things you can do to take a little bit of a load off your iPad is to turn off the video caching and to set the display resolution to 720 instead of 1080. Both seem to give the GPU a little bit of a break. The newer firmwares were supposedly going to address some of these issues but as you have found out, they do still show up - but not in all circumstances.

With your background, you might be able to get some additional information from the now failed iPad - but without that information it will likely only become a guess as to what happened and if the failure was coincidental or the result of too much too quickly for it.

I can offer a few ideas on moving forward with your inspire 2 and either choosing a display or using a dual display with your SmallHD monitor. We use dual displays on nearly all of our DJI set ups - modeling them very much like the Cinestar and Alta set ups we made. Secraft makes a DJI specific tray and a double mount crossbar that allows you to mount an iPad of your choice (all the way up to the big ones when using an XL size Mobile Grip (more on that when you click the thread linked below) as well as an HDMI monitor (daylight viewable preferred). On one of our more recent demo set ups we are using the small CrystalSky monitor on the left and an Atomos 1500nit monitor on the right.

I'd be happy to help you talk through this or get you set up with anything you would like to try...and while the circumstance to see you over here is unfortunate, I'm glad to get the chance to return the favor a little as you have made countless contributions to multirotor pilots for many years.

Just let me know what I can do next - or we can take this one to a PM or email as well.

See this link:
Some really cool and unique ways to set up your remote controller
 
  • Like
Reactions: Rick_L
Were they able to boot the device and see the log? A log will have to be gathered from the device or the iPad will need to be dissected for a conclusion.
FYi most genius from those bars do not know the mechanics of the devices. They have 3-4 steps of diagnosis to go down from a list that is supplied to them after you state your issues.

Understood re: the decision tree that the Geniuses have to use, but they're the first line of triage.

But, to answer your question re: booting the device. Nope. The iPad is now an ex-iPad, "it's kicked the bucket, shuffled off the mortal coil, run down the curtain and joined the bleedin' choir invisible!" (Thanks to John Cleese for that description.)

So no hope of getting the syslogs or the screen captures of the failures, I'm afraid.

I'm just going to commission two new iPad Pro 10.5's using PolarPro adapter thingies so that they will fit into the RCs.
Andy
 
Hello Andy - this is Michael McVay replying...long time Freefly Forum member (although I think I have fallen off of the most notable members list recently).

Hi Michael: Long time no speak! Hope you're doing well.

We are a DJI distributor and help with a lot of issues people have with their DJI products. We supply a number of our long time Freefly friends with their DJI stuff.
That's good to know!

While I certainly can not say for certain what happened in your case, iPads have been routinely getting the CPU Fully Loaded message and then having some video breakup issues that appear as green lines, green pixels, blocky or frozen video, etc. A few things you can do to take a little bit of a load off your iPad is to turn off the video caching and to set the display resolution to 720 instead of 1080. Both seem to give the GPU a little bit of a break. The newer firmwares were supposedly going to address some of these issues but as you have found out, they do still show up - but not in all circumstances.
Yeah, my research has revealed the situation as you described. I was using Go 4 just fine for two hours the day before this incident and I didn't want to change any defaults so I don't even know whether video caching was enabled by default.

I realize the code has to lots of pixel mashing and that's computationally pretty expensive -- as you say, I was hoping that the newer DJI firmware was going to solve that problem -- but it's not clear whether the issue originates in the firmware or in the Go 4 app itself -- I suspect that's where the main computational load lies -- the graphics processing is probably compute-bound.

With your background, you might be able to get some additional information from the now failed iPad - but without that information it will likely only become a guess as to what happened and if the failure was coincidental or the result of too much too quickly for it.
I'd like to think we'd be able to recover something but the iPad is inert (I'll avoid the Monty Python parrot sketch again).

I can offer a few ideas on moving forward with your inspire 2 and either choosing a display or using a dual display with your SmallHD monitor. We use dual displays on nearly all of our DJI set ups - modeling them very much like the Cinestar and Alta set ups we made. Secraft makes a DJI specific tray and a double mount crossbar that allows you to mount an iPad of your choice (all the way up to the big ones when using an XL size Mobile Grip (more on that when you click the thread linked below) as well as an HDMI monitor (daylight viewable preferred). On one of our more recent demo set ups we are using the small CrystalSky monitor on the left and an Atomos 1500nit monitor on the right.
I noticed that Secraft had a tray -- might go down that route. For a quick fix, I'll just use the PolarPro device extenders -- looks like they'll accommodate the iPad Pro 10.5 in a protective case -- which would be a good thing. The PadHat angled hoods are pretty good too. I tend to use the HDMI out on the RC feeding into an Atomos Ninja HDMI recorder and then into a pair of Zeiss Cinemizers. Then there is less need to rely on the iPad and the visibility of the Cinemizers is exceptional in full sun if you're wearing a brimmed hat.

I'd be happy to help you talk through this or get you set up with anything you would like to try...and while the circumstance to see you over here is unfortunate, I'm glad to get the chance to return the favor a little as you have made countless contributions to multirotor pilots for many years.
That's very kind of you to say that Michael. I'm glad to hear it!
I'm still flying the ALTA 6 and "the big daddy" ALTA 8. It's been interesting to see how both DJI and FF have evolved. The pilot community has also become so much more skilled and more knowledgeable in such a short time. A lot of pilots have put a lot of time into learning their craft. It's been impressive to watch.

Just let me know what I can do next - or we can take this one to a PM or email as well.

See this link:
Some really cool and unique ways to set up your remote controller
I'll contact you via PM or email. Thanks again for the kind words.
Cheers
Andy
 
Last edited:
I remember that parrot! :p

Hey "Ed" -- I wondered where you'd been? I thought that line might resonate somewhat. Looks like the forum here has been kept in good nick under your watchful eye!

Seems like my I1 batteries self-destructing saga has been followed by this unfortunate event. But thanks for letting me tell my tale of woe. Wish we knew more but the iPad was pining for the fjords before we knew it....

Cheers
Andy
 

Members online

Forum statistics

Threads
22,277
Messages
210,655
Members
34,326
Latest member
BobbyeriGop