Wonder if anyone can tell me why Sparky went into landing mode ..can see by pics it suddenly thought it was less than a foot of the ground and activated landing , but why ..it was not a problem i almost imeadiatly cancelled landing and carried on flying ...

Interesting. Any chance you can share logs from that flight? http://www.phantomhelp.com/LogViewer/Upload/

I saw it too ...

But I do not remember if I was landing or not.

Wobbles , could you please sync the flight records, then provide me your DJI account email and the date this occurred? I'd like to look into it.

I experienced it too, but I should dig in all flight log to find out which one it was...

glad is was not just me then LOL , sent logs to DJI to see if they can figure it out, will update as i know

Monochrome Posted at 2017-7-27 15:43
Interesting. Any chance you can share logs from that flight? http://www.phantomhelp.com/LogViewer/Upload/

yea will do later when home ..

Deleted more times DJI go 4 on android and iOS use two devices.


deleted everything physically.

Lian82 Posted at 2017-7-28 03:47
Deleted more times DJI go 4 on android and iOS use two devices.

sorry dont understand what you are saying?

wobbles Posted at 2017-7-28 04:01
sorry dont understand what you are saying?

The moderator asked for flight data ..

I replied: "i deleted all apps".
easy no?

Lian82 Posted at 2017-7-28 06:01
The moderator asked for flight data ..

I replied: "i deleted all apps".

oh i see ,thought she just asked me ,

Lian82 Posted at 2017-7-28 06:01
The moderator asked for flight data ..

I replied: "i deleted all apps".

Not easy at all wise guy , In the first place,its not your topic and the moderator was talking to wobbles not you so A little understanding maybe ?

Moderators wont to see in to this because this may lead to those falling from the sky sparks

Hi, how do you guys record the screen with all the control display on it?

I think I found the flight were I had a similar alert :

https://www.dropbox.com/s/apzd16 ... 4-55-22%5D.zip?dl=0

http://app.airdata.com/main?share=muhbPP

at 5:59, it shows a confirmed landing... at 40 m above ground.

djiuser_LzVM4RJ Posted at 2017-7-28 07:59
Hi, how do you guys record the screen with all the control display on it?

I use Iphone 6s plus with IOS 11 (bETA) and its in the controll center

Szwedu Posted at 2017-7-28 07:53
Not easy at all wise guy , In the first place,its not your topic and the moderator was talking to wobbles not you so A little understanding maybe ?

Moderators wont to see in to this because this may lead to those falling from the sky sparks

could not have put it better ..

Just a theory, but..

Did you clean VPS sensors before take off? Spark does not appear to use the same ultrasound technology on VPS as P3 or P4, maybe the sensor are dirty and give incorrect data , making the drone think its very close to the ground and land when are 70m high or even on thinking the drone are close to a hand and  do a palm land.

Anyone knows whats the VPS techology used on spark? And why the sensor are not like P3 or P4 (ultrasound)? Looks like infrared.

Sorry for my english, its not my  first language.

VPS on Spark is infrared as front . If i dig up DJI statement on this i will show.

Continuing my theory..



Maybe thats why DJI on Spark User Manual said more than 1 time to clean or keep clean all sensors before flight, maybe, its just a theory.

Dji Spark Manual advises twice about keep sensors clean or this maybe cause problems. Maybe land at 70m high its one of the problems (just a theory), other problem may be the difficult to use hand gestures, people may forget about clean the drone before take off or dont even read the manual ( i not saying its the case of the owner of this post).

Dji implemented a bunch of new functions and technologies (new vps? new colision system, hand gestures, etc) to a "affordable price drone" that targets a new drone users that maybe never flow before a drone, when even the old drone users never have contact with this new technologies and functions at the moment, its really not a good idea.

All new technologies have their own peculiarities, in the case of a drone, this may lead to accidents and problems until its 100% learned.

Sorry for my english, its not my first language.

I see youtube videos from spark, and many of them after a crash dont even do a check on drone before take off again, common guys, EVERY drone crash is a serious thing to deal with, motors and blades need to be inspected to find missaligment or dirty inside motors, need to search fractures on frame, clean the contacts and inspect the batteries, and MANY others checks before take off again, even if spark is a small drone.

Drones are not made to crash! not even 1 time, its not a normal thing. Spark , besides his size, ITS NOT A TOY, and i am just observing people dealing with spark as it is a toy.

I cleaned the sensors after removing the plastic films at servicing. Then I've been checking and they kept clean.
I land on a clean surface and never touched them.
Moreover, I had this behaviour only once. A dirty sensor should trigger this all the time / more often.
Maybe it's the root cause in some other cases, but for the one I've been facing, I would not think so.

Szwedu Posted at 2017-7-28 10:20
VPS on Spark is infrared as front . If i dig up DJI statement on this i will show.

Maybe infrared if used as VPS has some tipe of peculiarities, like be susceptible to dirt on glass that covers it, fog or something that makes the infrared sensor think the ground its closer than they are. If infrared works calculating the distance to the ground as a laser tape.   Makes sense for me if there are dirt between the sensor and the ground, sensor will think dirty is the ground.

ITS JUST A THEORY.

guys you might have a point , i had done 300 miles on bike and was filming at 4 diff locations using 1 battery for each location so Sparky had been in and out of box all day...will clean them and see. in all honesty i hope it is that simple , lesson learned by me if it is .. cheers

In addition to the upload of flight log and airdata, I just did phantompilot upload :

http://www.phantomhelp.com/logviewer/RHESNAMDMS8HMZ747PVW/#

interresting is that is show a completely wrong VPS altitude. However, I can guarantee that sensor where clean.

Edit :

VPS should show 0 at such height above ground...
However, from 5:38, it show non 0 VPS altitude, even if it SHOULD be 0 because its far out of the sensor real range.
(and Spark is not above unusual surface : just grass)
Then VPS gets back to 0 and I guess this is what triggers the landing....

Matioupi Posted at 2017-7-28 21:55
In addition to the upload of flight log and airdata, I just did phantompilot upload :

http://www.phantomhelp.com/logviewer/RHESNAMDMS8HMZ747PVW/#

Am interested to see I what dji say on this

Matioupi Posted at 2017-7-28 21:55
In addition to the upload of flight log and airdata, I just did phantompilot upload :

http://www.phantomhelp.com/logviewer/RHESNAMDMS8HMZ747PVW/#

I see your flight log.

Thats what i am talking about! VPS reads are wrong in various moments of you flight, not just at 5.38. Maybe VPS have problems or it is very sensitive at the point it wil catch particles or fog in the air (i dont believe).

I opened a topic  with a theory that this wrong VPS read may lead to a drone shutdown midair, thinking that drone are landed on the ground or a hand when are  midflight .

I need to see  flightlog of some spark that fell out of the sky.

I read some time ago that thick fog triggered landing for someone. So maybe this is a good lead.

Hathdert Posted at 2017-7-29 01:14
I see your flight log.

Thats what i am talking about! VPS reads are wrong in various moments of you flight, not just at 5.38. Maybe VPS have problems or it is very sensitive at the point it wil catch particles or fog in the air (i dont believe).

We agree that those false landing mode switch should not happen and that they are issues with VPS (could be a bug walking on the Spark, I mean a real bug like fly or bee ?)
Anyway, there should be other software conditions to prevent this like IMU get constant values (Spark is still, or barometer altitude threshold)

Spark that fall in the Sky are shut down when their pilot get them back (even when the battery is in place)
After landing and engine shutdown, Spark is still on.
One possibility that would still link the two kind of events (Sparkk fall off mid sky and false landings) is if the battery shuts down when a violent landing / shock is faced. Any volunteer to drop a battery/it's Spark on the ground and sees what happens ?

i agree with your posts , it all makes sence , i really do think this issue is linked to sparks dropping out the sky,
As mentioned above if the spark reads 0 alt it will power off .
so here is a idea , why dont they build into the fimware/software that the spark has to read 0 alt for certain amount of time and it checks that all all other paramaters are checking ..IE its still and no movement .
and or you phsically have to press another button as well to finaly turn off.
ONE VERY IMPORTANT ADDIDTION : Is when this happened i was coming down from a higher altitude ..IE left stick fully down,,so putting two and two togehther
SPARK READS: left stick fully down ..lets check alt it reads 0 ..so turns power off...luckily for us on this post we avoided the shut off BY LUCK on my part....by immeadiatly realisng stick and canceling landing prompt, i recon on the ones where they fell it just went straight to zero and they had no landing prompt just a shut off..
Not sure if this makes sence to the brains on here , but me it makes a perfect sence and could stop us owners having all the grief when you loose or crash your sparky due to a fixable error..
cheers

I was also going down with stick full down

So conditions for shut off could be (to let Spark believe it has landed) :

VPS get some noisy measurement different of 0 so Spark believes it is close to the ground (maybe with additionnal condition of deacreasing values) which is the case in my flight (time 5m 55.8s)
Stick full down
VPS returns to 0 for a given time with stick still down

Without extra checks (IMU still and IMU height) Spark may beleive it has landed... DJI should tell us what the exact conditions for engine stop are.

Still this theory does not explain why fallen Spark were found OFF while after a regular landing, it's still ON.

wobbles Posted at 2017-7-29 01:51
i agree with your posts , it all makes sence , i really do think this issue is linked to sparks dropping out the sky,
As mentioned above if the spark reads 0 alt it will power off .
so here is a idea , why dont they build into the fimware/software that the spark has to read 0 alt for certain amount of time and it checks that all all other paramaters are checking ..IE its still and no movement .

The problem is that Spark are a "unusual" drone .

It can be used with controller, without controller, to launch from the ground, from the hand, to controll with the hand, with the controller, with the cell phone, etc.. So its not easy to make some security things.

Me for example, i dont have the controller, its will be not easy to me pull the trigguer down on my phone to shut off motors when land like you said.

One thing that emplain that type of difficulty:  THE SPARK WHEN USED WITHOUT CONTROLLER , CONTROLLED WITH A MOBILE PHONE - DOESNT HAVE A SHUTDOWN EMERGENCY BUTTON OR SEQUENCE - and its very very important . Every one that has the controller, just need to do a CSC and drone will fall of the sky on a emergency.

Before i have the Spark, i,ve had the Parrot bebop drone, i dont have his controller too, but parrot put on APP a emergency button to shut off bebop midflight ( its very dangerous), and for its very very importante.

So thats a example thats not easy for Spark to do some security things as are with other more simple drones like Phantom.

I'm just thinking that those hypothesis could be field checked and "reprocuded" by hanging the spak to some pole so that as some point it would pull down without VPS changing. It would not be completelty still (if hanged to just one string) and IMU to VPS height can be made different if you move takeoff and false landing location (e.g. on a hill following terrain slope)

Land the Spark without using the auto-land function consists in:

Force the spark to the ground keeping the left stickers down until drone drone shut off motors, the VPS read will go from some altitude  (1m por example) to absolute 0 altitude - the drone will basicaly knows thats not possible to go down and shut off, the IMU will know that the drone cant descend more too. (If the drone will land on a hand, probably it will ignore the IMU and GPS data, as the hand will note be so stable as the ground and not on real 0 altitude, and the drone relies the entire land just on VPS).

So, think with me..

Spark use a infrared type VPS, it probably works like a laser tape, its a beam, not a wave like a ultrasonic sensor (spark its the first to use this, all others use ultrasonic), IR beam its much more concentrated on a small area, if theres something between the tape and the wall, the thing that are between both will be the distance read. So, if theres something  between the drone and the ground, fog, bees, dirty, particles, the distance to the ground will be read from that thing and not the real ground.

If the VPS on Spark has wrong VPS data at 70m altitude, when the data need to be 0 , it tells other distance for that moment ( and it has! the flightlog dont lie!), with the wrong variation on altitude given by VPS infrared ,  the drone  may think that  are landing and forced trough the ground when its not! as i said before how it works, just imagine a wrong variation in altitude midflight at 70m altitude given by VPS from "16m" (wrong) to "0".

The drone will Turn off thinking it was landed or starting the auto land process.

Its just a theory.

Remenber, this type os VPS technology are not just new to us users, its new to DJI too!  Spark are the first DJI drone to use infrared type VPS.

And remember, even if DJI test the drone 2.000.000 hours before start selling on market. It doesnt matter, as the tests are probably on a controlled environment that dont have particles in the air or fog, or other type of infinite real world variations that can lead to a small or a big problems when flight on real world.

Just to verify, had you removed the thin film of the Vision Positioning System? Please check this.
Besides, was it a flight environment with fog or cloud? If it was, please try to fly in another place to test.  

Before the Galaxy Note 7  start his sells on the market, probably the samsung makes previously thousands of tests of different types, probably give the phone to testers use it for a good period of time and do many more security things, and at the final, note 7 start to explode after sells started.

And remember, its just a phone! Real world variations on a drone are much much more complex than others gadgets or products, its a flying machine!  Drones for sure need much more attention and time to release as close as 100% secure. I doubt DJI has this type of concern, given the problems since phantom 1 fly aways, phantom 3 shell cracks and brick updates, etc.

The p4p has infrared on sides so that would be the firdt to use it. I do agree the tech is new to dji tho

$gambino$ Posted at 2017-7-29 03:36
The p4p has infrared on sides so that would be the firdt to use it. I do agree the tech is new to dji tho

My mistake, so the spark is the second to use this type of VPS tech.

The spark probably its the first to use on main bottom sensor.

DJI Thor Posted at 2017-7-29 02:52
Just to verify, had you removed the thin film of the Vision Positioning System? Please check this.
Besides, was it a flight environment with fog or cloud? If it was, please try to fly in another place to test.

As stated on previous post of this thread plastic films were removed and sensors are clean.

No fog or dust the day of the flight either.