S-e-ven Posted at 2018-7-12 11:10
got that changed?

https://forum.dji.com/thread-111652-1-1.html

S-e-ven,

Oops! You may be right.
Did I get my wires crossed?
Let me double check this evening when I will  go with my Spark to play.

Funny thing is - the instructions I gave are copy and paste form DJI Support instructions regarding fail safe  set up for Spark. They sent them to me via e-mail just couple of days ago in response to my fail safe in ATTI query.

If what you say turns out to be true (I strongly suspect - it is true :-)) this will be lesson number 5 from this incident: Never trust DJI Support again - they know less than we do :-).

Actually, the last statement is probably entirely true. Sadly.

Mirek

Well... Checked it and S-e-ven is right. This option does not exist for Spark.

Moral of the story - treat information provided to you by DJI Support with utmost suspicion. Trust but alwsys verify :-).

Mirek

Wow, this was a great read and completely devoid of the typical WAAHHHHH MY DRONE IS GONE BUY ME A NEW ONE DJI OR ELSE!!! one usually sees in these posts. Glad I decided to read it!

Mirek, thank you for the extremely in-depth analysis that allowed the OP to retrieve his drone and get to the bottom of what happened!

OP, thank you for going through this process in a calm and reasonable manner, understanding and accepting the reason this happened was pilot error from calibrating the compass over a steel structure, and just generally being a good sport about all this! Very happy that you got your Spark back, and I hope you have much more trouble-free fun with it in the future! Also, now you know that if the Spark prompts you to calibrate the compass, most likely it is due to local compass interference and you should just move to a better spot rather than doing the calibration.

My big take-away from this incident is what Mirek was also trying to pin down; Spark's behavior in ATTI when it gets disconnected from the RC.

Now I know that in case of ATTI mode I should:

1. Try to fly the Spark back to me either VLOS or with the compass display in the app. (I already knew this and had to do it once.)
2. If I'm unable to fly it back for whatever reason but don't want it to drift/fly farther away than it already has, or just want it to land in place: Switch off the controller to disconnect! Spark should land after 20 sec, but will probably still drift with the wind during landing, since it will still be in ATTI.

This is a last-resort option of course, but good to know that it's there.

Mirek6 Posted at 2018-7-12 10:18
Hddeuce,

Perhaps it is common misconception. I see experienced drone  vlogers and forum members claiming 400 ft as legal limit in USA. I am from Canada so I am even more restricted (90 m).

Hdduece is correct. There is nothing new about this. In the US. 400ft is an FAA recommended advisory guideline for section 336 recreational RC flying.

Section 107 commercial flying is more striclty limited to 400 feet. For 336, quoting soemone from the FAA:

"Section 336, the provision specifically addressing model aircraft in the 2012 FAA Modernization and Reform Act, does not contain a definitive altitude limitation for model aircraft operations. Rather, it requires operation of model aircraft “in accordance with a community-based set of safety guidelines….” Community-based organizations, such as the Academy of Model Aeronautics, may establish altitude limitations in their safety guidelines that exceed the FAA’s 400ft AGL altitude recommendation.

Although such safety guidelines may provide for flight above 400 feet AGL, Section 336 also protects the safety of manned aircraft operations by requiring that model aircraft not interfere with and give way to manned aircraft. The state also explicitly affirms that the FAA may pursue enforcement action against model aircraft operators who endanger the safety of the NAS.

Sincerely,
Earl Lawrence
Director
Unmanned Aircraft Systems Integration Office"

On Camera View, tap on the three dots on the upper right hand of the screen.
Afterward, tap on the drone icon which is the Main Controller Settings
Find Advanced Settings.
Under RC Signal Lost, you can choose between three choices: Hover, Return to Home or Land

Mirek - I have never heard of this option. In DJI Go 4, I do not see "Advanced Settings" under "Main Controller Settings." What version of DJI GO 4 are you using?

marlowe Posted at 2018-7-12 16:10
On Camera View, tap on the three dots on the upper right hand of the screen.
Afterward, tap on the drone icon which is the Main Controller Settings
Find Advanced Settings.

marlowe,

Yes - I corrected myself in post 81 and 82 above. This option is not available for Spark - sorry.

Mirek

Firehawk989 Posted at 2018-7-12 15:29
Wow, this was a great read and completely devoid of the typical WAAHHHHH MY DRONE IS GONE BUY ME A NEW ONE DJI OR ELSE!!! one usually sees in these posts. Glad I decided to read it!

Mirek, thank you for the extremely in-depth analysis that allowed the OP to retrieve his drone and get to the bottom of what happened!

Firehawk989,

You write: "2. If I'm unable to fly it back for whatever reason but don't want it to drift/fly farther away than it already has, or just want it to land in place: Switch off the controller to disconnect! Spark should land after 20 sec, but will probably still drift with the wind during landing, since it will still be in ATTI."

No - don't switch off RC. Use your compass view and try to get your drone back manually. If you panic or really have no idea what to do, hit auto-landing button (not RTH) and for good measure put throttle all the way down and watch Spark going down (on metrics at the bottom of DJI GO screen). Switching RC off you give up.  Never give up unless your signal breaks.

Mirek

marlowe Posted at 2018-7-12 15:51
Hdduece is correct. There is nothing new about this. In the US. 400ft is an FAA recommended advisory guideline for section 336 recreational RC flying.

Section 107 commercial flying is more striclty limited to 400 feet. For 336, quoting soemone from the FAA:

Marlowe,

OK - you learn something new every day. This is why I am such a big fan of this forum :-). Thank you.

Mirek

Guys,

So here is a dilemma.

For the longest time I was looking for a definite proof of Spark fail-safe behaviour (when signal is lost) in ATTI.
I exchanged e-mails with DJI Support, I analyzed logs on this forum and I experimented with some friends.

Until today, I had very strong suspicion (confirmed by my previous analysis and some videos I have seen) that when signal is broken and Spark is in ATTI mode, it would start to land.
The only real question remained: does auto-land sequence start 3 seconds after signal breaks (as is with failsafe RTH sequence when connected to RC) or after 20 seconds (as is with failsafe RTH sequence when connected to mobile). Or, perhaps, somewhere in between? (but why?).

Today, for the first time, we saw a definite proof in this case. I quickly calculated ~20 seconds wait time after signal broke before Spark started to autoland in ATTI. I calculated 20 seconds because in 25 seconds between signal breaking and signal restoration Spark managed to drop 29 metres down. 29 metres at 3m/s takes 9.5 seconds. Not quite but close enough to explain what happened during 25 second gap.

And than second lightning struck in the same day! My friend 62+ posted video from his recovered drone – the case in South Africa which I analyzed month ago. See: https://forum.dji.com/thread-152623-1-1.html My analysis in post 20 and his video in post 41.

In that previous case Spark flew away in ATTI, signal was disrupted by a local LTE cell tower and 62+ lost contact with the drone. When ATTI conditions improved, drone switched back to GPS and started RTH, but it started from low position above the ground, which led me to believe that ATTI and signal loss caused auto-landing. There was, however, no data to calculate when auto-landing started. Until 62+ posted his video today and notified me. From his recovered video (he compares video from SD card and video from his mobile – which is disrupted with signal loss) it is very clear that auto-landing started very soon – several seconds (3? 5? 7?) – after signal loss.  3 seconds would br consistent with failsafe for RTH when RC connection lost.

Having just analyzed NTQLT case and proclaiming 20 second delay – this new finding completely blew me away.

I came back to NTQLT’s log and started too look at it from the perspective of auto-landing after 3 seconds – not 20. And it is hard to reconcile. 25 seconds with no signal and 29 meters drop does not add up cleanly to 20 second delay nor does it add up cleanly to 3 second delay. More like something in between these two numbers.
But than I remembered that auto-landing never starts full speed down. Spark has inertia and it takes several seconds for it to reach full speed of descent.

So we have 25 seconds:

• First 3 seconds nothing happens (countdown to auto-landing after signal loss)
• Next few seconds (let’s say 5) is spent on calculating RTH trajectory – if Spark was in P-GPS it would be turning towards HP now – not descending yet.
• Calculations fail – Spark is in ATTI – auto-descent starts.
• Next 5-8 seconds Spark starts to accelerate towards full descent speed (full speed averages between 2.6m/s and 3 m/s). Let’s assume it managed to drop 7 metres during this time.
• Next 22 metres it descends with average speed of 2.6 m/s – it takes about 9 seconds.
  

The above gives us 25 seconds.
Yes – some educated speculation here. But given new video proof by 62+ I must say that Spark cannot take 20 seconds to start auto-landing. 3 seconds plus some delays as above is feasible.

The bottom line is we know for the fact from NTQLT case that 25 seconds after connection broke, Spark was already 29 metres below and auto-landing at 2.6 m/s to 3 m/s.
This should be good enough for future cases and calculations where to look for lost drone.

My apologies for confusion. As you just witnessed reverse engineering of somebody’s design is not exact science. Lots of pitfalls and blind alleys. But eventually the true picture does form.

Any comments or other suggestions are welcome.

Mirek

Firehawk989 Posted at 2018-7-12 15:29
Wow, this was a great read and completely devoid of the typical WAAHHHHH MY DRONE IS GONE BUY ME A NEW ONE DJI OR ELSE!!! one usually sees in these posts. Glad I decided to read it!

Mirek, thank you for the extremely in-depth analysis that allowed the OP to retrieve his drone and get to the bottom of what happened!

"Try to fly the Spark back to me either VLOS or with the compass display in the app."

I am just not sure, that the Compass (radar view) display will give you the right feedback.
In Atti, with Compass failures and Position/GPS ignoring.
I just don't know, but in Atti, the bird is kind of blind to directions and locations, so what info would the app get, for flying it back only from the display?

or is (if it is not, DJI, would that be possible) the App as long connected to the bird getting GPS info for the map/radar view, even when the bird has to ignore , just say, 18 satellites?
What happens in Atti with the distance readings?
and where is it coming from?

And I do not talk about the "switch to atti" Drones, Phantoms.
There I asume that the app is fully aware of the direction and position, just the bird behaves, as if it wouldn't know.
How is that in a real Atti case? Spark just has "real Atti" cases!
What information does the app has, where does it gets its readings?

Mirek6 Posted at 2018-7-12 19:51
Guys,

So here is a dilemma.

I just think, it is time to open a own topic for Atti, with all the findings and the excamples and videos linked to it.
It is just more possible, that people read about it, BEFORE it happens to them, if it is a own
topic, maybe eben "digest" (DJI, are you with us ;-).
The original posts getting lost over time, mostly.
One topic can be bumped every know and then, by adding new findings to it!

Great news,, where about was it,, ?

Mirek6 Posted at 2018-7-12 12:13
S-e-ven,

Oops! You may be right.

They may well know less than you. They wanted me to jump through hoops just to even take a look at my logs. You solved it before could even provide the info they wanted.

Firehawk989 Posted at 2018-7-12 15:29
Wow, this was a great read and completely devoid of the typical WAAHHHHH MY DRONE IS GONE BUY ME A NEW ONE DJI OR ELSE!!! one usually sees in these posts. Glad I decided to read it!

Mirek, thank you for the extremely in-depth analysis that allowed the OP to retrieve his drone and get to the bottom of what happened!

Exactly! It's great to know the steps that can be taken in ATTI situation. Mirek did a ton of work to figure this all out. I'm glad he stayed with this, I didn't want to just find the Spark. I wanted to find out what happened so there would not be a repeat. I certainly did not expect him to put in all of his time to continue to pin down the cause. It was so much appreciated.

Firehawk989 Posted at 2018-7-12 15:29
Wow, this was a great read and completely devoid of the typical WAAHHHHH MY DRONE IS GONE BUY ME A NEW ONE DJI OR ELSE!!! one usually sees in these posts. Glad I decided to read it!

Mirek, thank you for the extremely in-depth analysis that allowed the OP to retrieve his drone and get to the bottom of what happened!

Exactly! It's great to know the steps that can be taken in ATTI situation. Mirek did a ton of work to figure this all out. I'm glad he stayed with this, I didn't want to just find the Spark. I wanted to find out what happened so there would not be a repeat. I certainly did not expect him to put in all of his time to continue to pin down the cause. It was so much appreciated.

Mirek6 Posted at 2018-7-12 18:55
Firehawk989,

You write: "2. If I'm unable to fly it back for whatever reason but don't want it to drift/fly farther away than it already has, or just want it to land in place: Switch off the controller to disconnect! Spark should land after 20 sec, but will probably still drift with the wind during landing, since it will still be in ATTI."

And don't lose sight of the Spark. I made at least two mistakes, beside the erroneous calibration. I flew it so high it was difficult to impossible to see. No problem, I thought. It's directly overhead. Not for long. The other mistake was that I was standing right next to a thick row of trees and brush with no view of the east. So when the Spark went wandering, I had no chance to see it visually in order to know where it headed to.

Mirek6 Posted at 2018-7-12 18:57
Marlowe,

OK - you learn something new every day. This is why I am such a big fan of this forum :-). Thank you.

Something bizarre happened last night. A small and somewhat expensive looking small plane showed up at my location, at low altitude. He pretty much put on a small airshow for maybe ten minutes. Never have seen anything close to it in 20 years of living here.

S-e-ven Posted at 2018-7-12 22:35
I just think, it is time to open a own topic for Atti, with all the findings and the excamples and videos linked to it.
It is just more possible, that people read about it, BEFORE it happens to them, if it is a own
topic, maybe eben "digest" (DJI, are you with us ;-).

S-e-ven,

This is very good idea. I will think about how to do it. Not sure how to keep it up though. This finding and summary where to look for Spark when signal is lost in ATTI should be permanently on top of Discussion lists.  For example, you had an excellent thread regarding HP settings and RTH behaviour in various situations. Excellent research and findings thanks to you and JJB and yet this thread is bound to get buried somewhere down the list.

Any advice on how to keep our findings readily available to everybody and in a structured form so people do not need to get hit and miss results with their searches?

Perhaps we should compile a book in .pdf format titled: "Secrets of Spark - what manual and DJI will not tell you and what you may regret that you did not know - bye bye Spark" :-).

Mirek

Flighty Posted at 2018-7-13 00:12
Great news,, where about was it,, ?

Not far from where the red line ended on the phantomhelp log analysis.

Mirek6 Posted at 2018-7-12 10:18
Hddeuce,

Perhaps it is common misconception. I see experienced drone  vlogers and forum members claiming 400 ft as legal limit in USA. I am from Canada so I am even more restricted (90 m).

Mirek6,

The real statement that I'd counter with is "post appropriate links confirming that there IS a 400' height restriction".  As I'm sure you know, Wiki is not an official or even accurate site.  I will however show a table directly from the FAA web site summarizing both 336 and 107.  Notice there is no mention of a 400' limitation for 336.  Also, a very good website by drone lawyers goes into detail about specifics of the 336 rules vs. guidelines.  That can be found here:  http://dronelawjournal.com/model-aircraft-operations/

N2QLT Posted at 2018-7-11 11:42
Got it! Intact!

Many people suspect that the 000.900 firmware is the cause of the errors. Maybe 00.1000 resolves. Once appeared but withdrawn ... maybe it's still not perfect ... Congratulations for finding the bird. Punish why he ran away ...

S-e-ven Posted at 2018-7-12 22:23
"Try to fly the Spark back to me either VLOS or with the compass display in the app."

I am just not sure, that the Compass (radar view) display will give you the right feedback.

From my personal experience, the radar view compass display did appear to provide correct or mostly correct direction information.

I had an incident where all sensors etc appeared normal and I was literally in the middle of the wilderness staying in a tiny cabin, so took off and started to fly a simple loop flight path gaining some altitude and aiming to circle out away from my location then back to it. About halfway through this flight path my spark dropped GPS and switched to ATTI, began to drift, and also lost video. I tried to correct and bring it back visually, but it was far enough away/high enough that determining its orientation was difficult. At that point I looked down at my phone and used the only aid I had left, the compass radar display, to get it facing in the right direction and flying back toward me. Shortly after that it switched back to GPS mode and I was able to fly it back normally.

Not sure what caused the interference in this case, as it was in the middle of nowhere. Maybe some strange local magnetic deposit in the rock?

I can show you the flight record and corresponding video clip if you like.

Firehawk989 Posted at 2018-7-13 11:18
From my personal experience, the radar view compass display did appear to provide correct or mostly correct direction information.

I had an incident where all sensors etc appeared normal and I was literally in the middle of the wilderness staying in a tiny cabin, so took off and started to fly a simple loop flight path gaining some altitude and aiming to circle out away from my location then back to it. About halfway through this flight path my spark dropped GPS and switched to ATTI, began to drift, and also lost video. I tried to correct and bring it back visually, but it was far enough away/high enough that determining its orientation was difficult. At that point I looked down at my phone and used the only aid I had left, the compass radar display, to get it facing in the right direction and flying back toward me. Shortly after that it switched back to GPS mode and I was able to fly it back normally.

Great story with happy end.

I always fly with that little compass active and checked at takeof for correct headings!
Its a life-saver, uh  drone saver!

cheers
JJB

Jakab Gipsz Posted at 2018-7-13 06:34
Many people suspect that the 000.900 firmware is the cause of the errors. Maybe 00.1000 resolves. Once appeared but withdrawn ... maybe it's still not perfect ... Congratulations for finding the bird. Punish why he ran away ...

I plucked off one of his propellers as punishment.

N2QLT
Very glad for you that you got your DJI Spark back.
"I plucked off one of his propellers as punishment."
very funny

Mirek6 wrote
Perhaps we should compile a book in .pdf format titled: "Secrets of Spark - what manual and DJI will not tell you and what you may regret that you did not know - bye bye Spark" :-)."
This is a very good idea.

The Spark Life-Saving Guide.....  I like it.  Seems like most of the challenges that a Spark can present are avoidable, so new users take note...  get the checklist.....  

An ATTI mode topic with all is warts and dimples would be a good read.

Mirek for President!  (and Prime Minister!)

Northwood Posted at 2018-7-15 11:38
The Spark Life-Saving Guide.....  I like it.  Seems like most of the challenges that a Spark can present are avoidable, so new users take note...  get the checklist.....  

An ATTI mode topic with all is warts and dimples would be a good read.

Northwood,

Jeez - pressure is building...
I am not quite sure I can stand the stress of these distinguished positions but thank you for your vote anyways :-)

MIrek

Mirek6 Posted at 2018-7-15 13:58
Northwood,

Jeez - pressure is building...

My best advice, run like hell!  (and fly your drone   )