Hi guys,
I am sure you have been agreeing with me that there is some going on with the Spark falling case, no matter it is hardware or software, but it is happening. And good thing that I start getting attention from DJI. See that thread here: http://forum.dji.com/forum.php?m ... 90&page=1#pid873699

So, I made my time last night and had a look around all the posts for "Spark falling off the sky". Apart from the latch issue where the battery not firmly connected (you can avoid that by checking the tightening before flight), I noticed that some of the falling Sparks were using separated battery (Which not come with the combo). For example:
http://forum.dji.com/thread-104883-1-2.html
Therefore, I start having some theory about the hardware compatible. I myself did experienced the same situation where my extra-ordered battery behaved really strange: Spark got ~18 GPS locked but it cannot record homepoint and kept staying in atti mode, sometimes jumped to GPS mode just mostly stayed in Atti, and signal lost very often. I then sensed the abnormality, so I immediately landed it, plugged into Assistant 2 and refreshed the firmware with the new battery on then everything is flawless until now. you can see that thread of mine here: http://forum.dji.com/forum.php?m ... 33&page=1#pid871329

So...guys, all are assumptions, I know some of you got angry because your Sparks fall for no reason, I do sorry for that and hope that DJI will have a satisfying solution. For those who not yet had the issue, please note that the threat is there, "not yet happened" does not mean "will never happen". My sugestting counter-measure for now is that:
--> If you bought extra batteries that come separated with your combo, please plug it in, connect to DJI Assistant 2 on PC and refresh the firmware with that battery.

Still waiting for some official answer from DJI but play it safe guys. I know with some $500 is not much, but in country like mine (Vietnam), $500 is like a fortune.
I make a POLL here to confirm my theory, so please vote that when your Spark fall, are you using the extra-ordered battery or the one come with your combo pack?

P.S./ I suddenly reallized it is quite funny that DJI staffs are supposed to to this kind of task (Collecting data and propose solution), rather than let a user do this. Anyway, it is for the good of the community so I don't mind.
Fly it safe guys.

I guess they cant speculate anything without concrete answer or enough evidence like logs, records, etc.  Is a bit strange having to "refresh firmware" when adding new battery though.. but good observation. or it could be they have expected x amount of fault tolerance, if it is still under the figure is still not 'critical'.. Imagine if they sell 1mil units but less than 0.05% report issues, it may not justify a recall or trigger a "response team"..

About your case, perhaps you can share your flight record and logs with DJI why you have 1x satelites connection but still in Atti mode..

Hm that's really interesting and would def. explain why some people got problems with the battery and some don't - probably a "battery batch" problem.
But as far as I know the DJI Go app should inform you on "battery firmware" updates. But if its just a "refresh" it's easy to do and hopefully gets rid of the problem.

@eylneb
If the battery disconnects or whatever then logging and video will also stop. The Karma had the advantage that the GoPro was still filming and therefor was proof of the sudden death ;)I hope the lack of information from DJI is because they just can't reproduce the problem - unless they got all the hardware including batteries send in.

You should put up an option for those who have had no problems with Combo or other batteries.

hallmark007 Posted at 2017-7-19 04:16
You should put up an option for those who have had no problems with Combo or other batteries.

Done
Thanks for suggestion

I have had zero issues with the batteries I received with the fly more or the fly less spark, I have had issues with erratic behavior while using gesture mode..

Mine too fell from the sky on the 3rd flight of a brand new separate battery. From about 6 meters high. I even caught it on a phone camera (poor quality but you see it fall)

From Spark FAQ :

    When does a battery enter Hibernation Mode?

    When a battery is less than 10% charged, it will enter Hibernation Mode after approximately six hours of nonuse.
    When a battery is completely drained, it will enter Hibernation Mode within 10 minutes of turning off.

could it be that for some reasons batteries get in hibernation mode during flight ?
Is it possible to ground/flight test the time counters that the battery believes it is with respect to those hibernation conditions ?
If not, shouldn't this be added to firmware quickly ?

I've hopefully not experienced spark fall (yet ?), I have 2 stock batteries and 2 extra batteries. I'm of course monitoring this subject very closely.
I believe that it should be addressed even very quiclky by DJI even if this affect only a very small percentage of users.
I saw DJI acknolegded cases where they admit that they was no pilot fault, so :

- either they consider and explicitely tells us that we have to live with this risk (which seems too high to me considering that it's a brand new product)
- either there is something ununderstood and they should make an official call to
* temporary stop of flights (as it can happen in general aviation after unuderstood accident)
* temporary restricted flight conditions, like ultra safe flights were you are certain to safely recover the hardware and hurt nobody.

The last proposed solution would at least providea possible way to accumulate global fleet flight hours with studiable logs and hardware...

What do you think ?

I purchased the Spark (which of course comes with a battery) locally at a big box. I flew that battery just fine for a number of flights. On a first flight on a new battery (purchased separately from DJI on Amazon) I had the Spark drop from the sky into a pond. I'm just adding my story in case it helps.

Jim

I haven't purchased my spark yet.. and I was thinking of buying the extra battery with it. Now I'm second guessing...

Matioupi Posted at 2017-7-19 09:12
From Spark FAQ :

    When does a battery enter Hibernation Mode?

Hi, thank you for your idea. That actually can happen, however, in my opinion, not likely the case here. My reason:
1-If that is the case, the programmer should have noticed that by now from many cases reported and have the fix.
2-If the battery enter hibernate (sleep mode), the only way to wake it up is to plug into the charger (as DJI explained), however, most of the cases users can still power on after the fall so I don't think the batteries were in sleep mode when  fell.
What do you think about my opinion?

So far no real action from DJI telling me that they do not know what is happening with their product. If yes, they should have told all the users to stop flying and release the fix by now. It really makes sense that way because there is no point for them to delay the fix if they now what the bug is. And the fix will not take too long if they know the bugs. In mathematics/ Technical, there are only 2 options: right or wrong, so if wrong and they knows what is wrong, it can be fixed easily. There is no grey zone that "the program is wrong but the product still working".
Not sure if you get my point but yeah, I have the reasons to believe DJI engineers are having no clue what is going on.
My Spark is currently on the shelf for until I hear something from DJI.

fans5d9349b7 Posted at 2017-7-19 08:47
My Spark shutting down and crashing. 70% battery.

Hi Mr. fans5d9349b7,
May I ask do you still have your Spark or you have already sent it to DJI for warranty?
We are trying to understand the issue if it is the battery matter or not, so, is there any chance that you can help to do a simple experiment. Just have Spark hover over something really really soft (a mattress for example), and at a very low altitude:
1-Using a new battery pack to see if it fall like in the video you attached?
2-Using the old battery (the one you used when if fall) to see if it fall again? If it fall again, we can 100% sure it is the battery issue that we are dealing with.
Again, this is a favor for all of us, so if you are not happy to do so, it is fine.
Thank you,

Here is 1 thread where the TS used 4 batteries..  & one having 7 batteries with no issue..

https://forum.dji.com/thread-103915-1-1.html
** but this doesnt represents a 100% working as it should... Just stating that people that have bought more batteries had successful runs with it

eylneb Posted at 2017-7-19 17:45
Here is 1 thread where the TS used 4 batteries..  & one having 7 batteries with no issue..

https://forum.dji.com/thread-103915-1-1.html

If you notice I was the one who just asked him so that the thread came to top.
I was asking for his experience with the batteries and overheat matter.
For now, I am trying to get more input, if,...the guys with falling Spark can help to do the experiment I suggested above.
I am worrying if it is not the battery issues, then surely it can only the quality control. Why? Because:
1-If it is software, all Spark use 1 similar code, therefore, if 1 fall, all must fall. That is why I have never believe the problem is Software until now.
2-It happen randomly, so I guess the Spark quality is not similar, therefore, you have some bad-quality products, some not. I am worrying this case might be true because DJI keeps silent with no software fix until now.

Phuong Do Posted at 2017-7-19 18:02
If you notice I was the one who just asked him so that the thread came to top.
I was asking for his experience with the batteries and overheat matter.
For now, I am trying to get more input, if,...the guys with falling Spark can help to do the experiment I suggested above.

Yes I know you posted there... sorry if you feel offended, i dont mean to. My goal was to also to also inform readers that it also worked as it should for some..

eylneb Posted at 2017-7-19 18:18
Yes I know you posted there... sorry if you feel offended, i dont mean to.  My goal was to also to also inform readers that it also worked as it should for some..

Hi man. no offense here. Sorry, don't worry, maybe I sound a bit serious up there, haha. I am just like you the normal user, and actually I am so new to drone.I have been working on some programming for arduino and quadcopter but nothing serious.

My purpose for those threads is to figure out the problem, therefore, my assumption could be wrong. Even I am wrong, I am still happy because it help to narrow down the issue, so, don't worry to say that I am wrong.
Feel free to give comment
Thank you

I'm new to drone to... been wanting to get one for the longest time but Phantom series was big & expensive, then came MVP which is small but still pricey..

Spark is just nice and while is still pricey but is within budget....

Nontheless we need more "logical" people like you to help the community to narrow down issues than to just shout n shout....

personally i've yet to have any major issues faced by some here.... when things happen i try to reflect what was wrong and what could have caused it... most time is because i get carried away or misjudge braking distance (in sport mode).. already crashed my drone 3 times due to that

haha

Phuong Do Posted at 2017-7-19 17:10
Hi Mr. fans5d9349b7,
May I ask do you still have your Spark or you have already sent it to DJI for warranty?
We are trying to understand the issue if it is the battery matter or not, so, is there any chance that you can help to do a simple experiment. Just have Spark hover over something really really soft (a mattress for example), and at a very low altitude:

I still have my Spark and I'll be carrying out such experiments over the next weekend.

Phuong Do Posted at 2017-7-19 16:46
Hi, thank you for your idea. That actually can happen, however, in my opinion, not likely the case here. My reason:
1-If that is the case, the programmer should have noticed that by now from many cases reported and have the fix.
2-If the battery enter hibernate (sleep mode), the only way to wake it up is to plug into the charger (as DJI explained), however, most of the cases users can still power on after the fall so I don't think the batteries were in sleep mode when  fell.

you're right, if people were able to repower on the field with the same battery as the one on the fall, this should probably not be a battery hibernation condition...

As for software vs hardware root cause, I have no opinion... Software can get really complicated right now and really unexpected things can happens.

As for the mecanical hypothesis :

- maybe some batch have defective contacts... Having close pictures of the battery and AC side electrical contact would be nice. Also having some tests gently shaking the AC + Battery that falled would maybe let reproduce the power loss if it's rooted to loose contact.

- could it happens that the button on the battery get "triggered" with the same sequence that the one that lead to power off ? (short push + 2 sec push) ? That seems rather improbable, but if the button is a bit sticky why not ? (could also be some debouncing issue at software level for reading the battery button state ?)

- other ideas ? ( Just for comparaison how strange things can go, I'm thinking about the strange "flash reset default" that happened on Raspberry Pi 2 dev board : https://www.raspberrypi.org/blog ... ree-physics-lesson/ ). Of course this kind of corner effect is very hard to imagine at conception time.

Matioupi Posted at 2017-7-19 22:58
you're right, if people were able to repower on the field with the same battery as the one on the fall, this should probably not be a battery hibernation condition...

As for software vs hardware root cause, I have no opinion... Software can get really complicated right now and really unexpected things can happens.

Hi,you made 2 good concern,i dont have the firm answers for that,but I have some opinions as well:
1.For the contact issue,that surely can happen,so that is back to the hardward issue.However,if the loose contact issue,shouldn't the battery still on after fall?Because Spark just loose the contact with battery only,battery itself is not cutoff.
2.For the sticky button on the programming side,i think programming for push button is not hard and it can be reuse,you can simple just google for button programming debounce,lots of samples.so I believe that is something to hard for DJI programmer.Or even in the worst case they did really fail on that debounce,it should happen with all battery since they all use the same code.

Actually,you can see that I originally thought there is some problem with the battery.However,today,to my surprise,the guy who got his Spark fall,told me he did try the same battery and the same aircraft after the fall.no cutoff...I guess that's why DJI keep being silent,they don't really know what is happening.
I am waiting for some more input,and hey,thanks for the comment again,please feel free to discuss my opinion above

Phuong Do Posted at 2017-7-19 23:49
Hi,you made 2 good concern,i dont have the firm answers for that,but I have some opinions as well:
1.For the contact issue,that surely can happen,so that is back to the hardward issue.However,if the loose contact issue,shouldn't the battery still on after fall?Because Spark just loose the contact with battery only,battery itself is not cutoff.
2.For the sticky button on the programming side,i think programming for push button is not hard and it can be reuse,you can simple just google for button programming debounce,lots of samples.so I believe that is something to hard for DJI programmer.Or even in the worst case they did really fail on that debounce,it should happen with all battery since they all use the same code.

Mayb you are right but I'm not so sure... It's quite easy to test, we should do the test to remove a battery while aircraft on without unpowering and check ow long it takes to turn off...
I don'thave my Spark with me right now.

xtro Posted at 2017-7-20 01:25
I wondered if it was not a problem of the sensor of temperature, if he is badly calibrated it can cut the power if it thinks that the temperature is too high

nice hypothesis, quite easy to test is to purposely fly/turn on the spark in very hot air (e.g. turn on and engines on inside a transparent box on sunshine) and sees if this brings CPU safety shutdown or somethink like that.

If this is the case, DJI could implement a "if temperature too high AND in flight"

Matioupi Posted at 2017-7-20 01:25
nice hypothesis, quite easy to test is to purposely fly/turn on the spark in very hot air (e.g. turn on and engines on inside a transparent box on sunshine) and sees if this brings CPU safety shutdown or somethink like that.

If this is the case, DJI could implement a "if temperature too high AND in flight"

i think that should be quite easy to tell with the proper log... as i think it is one of the items it capture.

eylneb Posted at 2017-7-20 01:41
i think that should be quite easy to tell with the proper log... as i think it is one of the items it capture.

expept if (for example) the frequency of temperatrue check and the duration of "filtering" this data + time to take the action is higher (takes less time) than the drop of bettery temperature to flight log.

I believe that trying to reproduce the behaviour is better than analyzing the logs, because it could then allow DJI to "debug" with their internal tools.

After three weeks flying (+/- 15 flights) without problems with the supplied battery I yesterday did my first flight with a new battery. Given the issues with new batteries I didn't dare fly high. But I didn't have any issue with the new battery as well.
So may be a bad Spark or Battery batch. Maybe s/n should be given to identify a bad batch.
Mine are:
Extra battery: 0C0AE5XB320RSU
Original battery: can not see it in the logs, will check this on the battery itself and update the post.

Matioupi Posted at 2017-7-20 00:02
Mayb you are right but I'm not so sure... It's quite easy to test, we should do the test to remove a battery while aircraft on without unpowering and check ow long it takes to turn off...
I don'thave my Spark with me right now.

Hi,
I have tested by turn on the battery (not connecting to Aircraft), and once the battery is turned on, it will stay power on until I double push again to turn off --> So this prove my point that if it is the contact matter, the battery should still be on after the crash But guys reported that battery was off when they picked it up.

xtro Posted at 2017-7-20 02:28
@ phuong do , you saw my answer ?

Hi Mr. xtro,
Yes, I was about to answer that. I was thinking about heat issue as well.

I did the small experiment today, thanks to Mr. Matioupi suggestion. I turn on the battery only, without connect to the aircraft. The battery itself can maintain the power-on itself without having to connect to the aircraft. Until you double tap it one more time to turn it off. So this help me to narrow down to 2 points:
1- Just like Mr. xtro's opinion, maybe aircraft was coded so that it cut off the power of the battery when overheat/ Or when there is some combination of elements (I cannot tell for sure unless there is code for the aircraft to review).
2- If Aircraft was never programmed to be able to cut-off the power of the battery by itself. Then the issue is surely not Spark, it is down to the issue of battery we are dealing with:
        2a. Again, maybe battery got programmed to shut itself off under some combination of elements.
        2b. Battery circuit/ hardware issue that some have problem, some do not.
I hope all above make sense, if not, please feel free to chip in your thinking.
I think we are getting close to the root of issue.

Phuong Do Posted at 2017-7-20 02:48
Hi xtro,
Yes, I was about to answer that. I was thinking about heat issue as well.

Hopefully one of the YouTubers will test this, heating up spark in controlled environment would answer if this is in fact an overheating problem

ernieken Posted at 2017-7-20 02:05
After three weeks flying (+/- 15 flights) without problems with the supplied battery I yesterday did my first flight with a new battery. Given the issues with new batteries I didn't dare fly high. But I didn't have any issue with the new battery as well.
So may be a bad Spark or Battery batch. Maybe s/n should be given to identify a bad batch.
Mine are:

i have had 47 flights at a distance of about 30000 feet and i had no issues with mine! i guess either i'm one of the lucky ones or maybe it will happen soon, either way i'm finished flying it over water until the problem is resolved!

Thor1 Posted at 2017-7-20 03:00
i have had 47 flights at a distance of about 30000 feet and i had no issues with mine! i guess either i'm one of the lucky ones or maybe it will happen soon, either way i'm finished flying it over water until the problem is resolved!

Hi Mr. Thor1,
As I said, we don't know yet if it is the heat issue or not. If it is the heat issue --> Only DJI can fix it now (My point 1 above). Also, if it is really the heat issue, I can have a picture like this:-DJI will never accept they have issue with "some" of the units, because if they said that, all the users will want to send it back to DJI for check, that will cost them a lot of money, correct? On the contrary, if they just let it go, and if your unit is faulty, you, individual only will send back for repair, the cost is not much. Doing the maths, you can see which way DJI will choose. Please note, this is only my opinion, please do not think I am bad mouth DJI, I truly love the company.
If it is point 2, where hardware issue with the battery: with your 47 flights at distance 30000 ft like that, I believe you don't have a faulty hardware. So don't let this discourage you

So, either way above, just keep flying.

Phuong Do Posted at 2017-7-20 03:12
Hi Mr. Thor1,
As I said, we don't know yet if it is the heat issue or not. If it is the heat issue --> Only DJI can fix it now (My point 1 above). Also, if it is really the heat issue, I can have a picture like this:-DJI will never accept they have issue with "some" of the units, because if they said that, all the users will want to send it back to DJI for check, that will cost them a lot of money, correct? On the contrary, if they just let it goes, and if your unit is faulty, you, individual only will send back for repair, the cost is not much. Doing the maths, you can see which way DJI will choose. Please note, this is only my opinion, please do not think I am bad mouth DJI, I truly love the company.
If it is point 2, where hardware issue with the battery: with your 47 flights at distance 30000 ft like that, I believe you don't have a faulty hardware. So don't let this discourage you

That is a good point. i understand why they wouldn't admit that. all i would like to know is when they expect to have resolved the problem, maybe even update us on if they are making any progress. they don't necessarily have to tell us what is wrong but how close they are to fixing it. the worst thing is not knowing.

Phuong Do Posted at 2017-7-20 02:16
Hi,
I have tested by turn on the battery (not connecting to Aircraft), and once the battery is turned on, it will stay power on until I double push again to turn off --> So this prove my point that if it is the contact matter, the battery should still be on after the crash But guys reported that battery was off when they picked it up.

this is not exactly the same test as turning on the battery connected to the aircraft and removing it while turned on...

with 24 cores in the Spark, a lot of thing will get very hard to debug because sequencing of events is not predictible.
I hope those 24 core stuff are not for critical flight fucntions which should be on only one core to get predictible behaviour.

Anyway, It could be that some other thread(s) lock up in some conditions, stick up one or more CPU core to 100% CPU usage resulting in high temperature which in turn make an OS/bios shutdown.
I have an old laptop where this used to happen all the time until I replace the thermal compound between CPU and radiator. It was heating too much and the bios or OS was shutting everything down right away.

This would switch the root cause to Spark (not batteries) and explain why batteries are still on.

DJI could issue a kind of "bench firmware" supposed to provide artificially high CPU usage (for ground tests). I guess that all the gesture recognition / intelligent flight mode requires much more CPU usage than normal mode. It would be nice to know what flight mode was on during incidents.
And to test the thermal issue hypothesis have in under heavy CPU load with some intelligent mode on.

(Could also be that the intelligent mode has been activated before in the same flight and some thread is not stopped properly when returning to normal mode, with high CPU load resulting)
When I see that shutting down the drone without stopping camera result in a corrupted video file, I really believe that there may exist serious and basic software flaws of the kind.

Matioupi Posted at 2017-7-20 03:32
this is not exactly the same test as turning on the battery connected to the aircraft and removing it while turned on...

Well, I would love to try it that way, but manual did say never remove battery from aircraft when it is turn on. See attachment.
Honestly, I am from Vietnam, and unlike most of you from a developed country with the strong currency. $700 is a fortune to me, So I will not riskdestroying my battery or aircraft unless someone is willing to sponsor me
If you can do us a favor and test, we are all very thankful for that, but I will not go against the manual

Phuong Do Posted at 2017-7-20 03:42
Well, I would love to try it that way, but manual did say never remove battery from aircraft when it is turn on. See attachment.
Honestly, I am from Vietnam, and unlike most of you from a developed country with the strong currency. $700 is a fortune to me, So I will not riskdestroying my battery or aircraft unless someone is willing to sponsor me
If you can do us a favor and test, we are all very thankful for that, but I will not go against the manual

thanks for pointing that out... Also I may live in a richer country i'm not willing to kill my Spark on purpose and ruin the warranty if DJI decides that I purposely did something wrong.

I saw a video on youtube where a guy is litterally torturing his Spark... maybe he could do those kind of tests :

Matioupi Posted at 2017-7-20 04:19
this is (one of) the kind of CPU the Spark features :https://uploads.movidius.com/1463156689-2016-04-29_VPU_ProductBrief.pdf
but i've not been able to find details with thermal specs for it/them

Thanks for the document

this is (one of) the kind of CPU the Spark features :https://uploads.movidius.com/146 ... PU_ProductBrief.pdf
but i've not been able to find details with thermal specs for it/them

There are approximately 21 people on this forum who have had crash or power cut off , with their spark, at least 7 of those have nothing to do with power cut off, 9 have had power cut off at very low altitude and 5 at higher Altitude.
1 has been given warranty , 1 has been refused warranty.
So all in all 14 have power cut off, of that 14 , 3 have produced logs for others here on the forum to look at, and yes they do show similar end to log when power cut.
So if we take law of averages some 20% of these power cut off's will be proven to be some other than power cut off. So we are left with 80% of 14 who it may be shown to have actual power cut off. So roughly 11 people with power cut off,
There are over 2000 people using this spark forum so we assume they own spark.
What we have is less than 1/2 % with problem of power cut , out of 2000.
This is less than 2 per week out of the 2000.
We haven't had any conclusive report from dji as to any case except 1 where drone was lost to water.

So if it's battery that might be causing this so far out of 2000, 14 batteries could be bad.

I know it's no comfort to those who have had problems, but with this type of technology is 99.5% Good or Bad. That's my question.

I also realise there are many other problems, but I'm talking about almost fatal problems here.

1-2% is too low To recall all spark.
How many sparks have been sold?