Hi guys,

I bought new Spark some time ago and here are my experices I want to share with you.

After unpacking brand new toy and making some short test flights I began to testing all possible settings, options etc. I also did new IMU calibration (without any obvious reason, I was not prompted by app, just wanted to have it calibrated after long way from manufacturer to me). I also calibrated compas every 5-10 flights. And what happend? I began receiving famous YAW errors and AC randomly switched to ATTI mode for short time periods. Fortunately no fly-away as I was able to control AC in Atti mode.  

So I began to look for reason of this compass/yaw error and I found this forum. After reading tons of similar cases, watching many videos and doing my own research, tests and flight logs analysis I realised my novice mistakes. (Thanks a lot for deep and great posts especially from JJBspark , Mirek6, KlooGee and others..)

Here are procedures witch I made and are working for me:

       
• Calibrate your IMU on sturdy flat         desk, without any vibrations from enviroment around, away from any         metal objects. I was literally holding my breath during calibration         to ensure no other forces are interfering)
• Calibrate your compass away from         any magnetic interfering, as written here many times.
• After this initial calibration do         not re-calibrate compass, until you are prompted by app or traveling         long distance(btw. did someone tested how many km is this range?)
• Do not recalibrate IMU also, only         if prompted or have some kind of crash
  
  

After this rules all Yaw errors dissapered and Spark is flying great, even in windy weather, rough maneuvers in Sport mode etc..Did more than 100 flights without any sign of compass error.

But this is not all, after reading many post about fly aways I found most of them have one common issue. It is this sequence: Pilot lost GPS mode (from various reasons), AC switches to Atti. Pilot doesnt know how to react, presses RTH button, switches to Sport mode, nothing works and AC drifts away by the wind. After some time of drifting, AC loses connection with RC and dron is gone.

I think most essencial skill in this case is to control AC in Atti mode. Unfortunately Spark does not allow swich Atti mode manualy, but this can be done by very easy and reversible procedure (just google switch to atti mode with Sport mode switch). I personally used this mod to train my Atti flying. After practising in calm condisions I also did test in quite heavy winds, about 800m away, just to simulate worst case scenario.Here are my observations:

       
• Alway keep hands no your sticks,         do not rely on any smart/autonomous function
• After switch to Atti, do not         panic! Nothing horrible is happening, you only need to control your         drone. Similar when driving car, you also must hold wheel and react         to every situation on the road
• First of all ensure you have good         visibility between RC and AC, as this is essencial in this point. If         not, try to gain some altitude, you dont want to lose connection.         
• Together with point 3 try to         estimate wind direction. Move camera little down to see which way is         AC drifting and compesate with right stick. Do not do any harsh and         extreme moves, just slowly, fluently move stick against the drifting         direction.
• Watch your speed to see if the AC         is more or less stabilised. After this, still holding right stick in         position where is AC stable, watch your map in left low corner, you         can see there AC position and also vector to Homepoint. Move left         stick slowly to rotate nose aircraft in direction to HP. Do not         forget holding right stick, if you let it to the cenral position, AC         will start drift again.
• Now you have AC more or less         stable and heading home. All you need to do is add little forward         motion with right stick, but again remeber, you still need to         compensate. Very usefull is using your front camera to lock at your         position and try to hold it in center.
• When you have AC in your vicinity,         free of obstacles, start descending.
• After you can clearly see its         position, rotate AC so it is in direction nose from you. Now you         have very natural control with you right stick and you can land         literally on penny even in strong wind.
  
  

I hope these my observations will help you, as Spark is beautiful piece of technology and it is so much fun flying with it Only man must understand some basic principles how to operate it.

Some good detail there - well done

Great post, almost correct. After you stabilize the drone by compensating movement you say that you will yaw to your direction, still correct but be aware that when you rotate the drone the compensation you did already will start moving the drone in another direction.

Great post, almost correct. After you stabilize the drone by compensating movement you say that you will yaw to your direction, still correct but be aware that when you rotate the drone the compensation you did already will start moving the drone in another direction.

Dan O Posted at 2018-7-6 02:19
Great post, almost correct. After you stabilize the drone by compensating movement you say that you will yaw to your direction, still correct but be aware that when you rotate the drone the compensation you did already will start moving the drone in another direction.

Yes you are right, I did not mention, that after yaw you need to change compensation accordingly to your new direction.

Jenda,  

Great and very thorough post.

Can you let us know which method you used to force ATTI?

As far as your question about compass calibration and “travelling long distances”.  Mavic Air manual says 50 km. I have recently analyzed fly-away in South Africa most probably due to un-calibrated compass when a person moved 400 km without re-calibration (DJI GO app displayed no warning). My personal rule is – few hundred km but I may go with what DJI recommends for Mavic Air – just to be on the safe side.

Having said that, bad calibration (in wrong surroundings with interference) is often worse than no calibration. Just saying :-).

I would add one more thing to your 4 point procedure. Recalibrate both compass and IMU if you see or experience any abnormality in your flight which you cannot explain by any sort of interference.

Thanks again for sharing.

Mirek

Thank you for sharing this informative information.

I would like to fly atti mode.  How do we set this?

Blellow Posted at 2018-7-6 08:55
I would like to fly atti mode.  How do we set this?

"switch to atti mode with Sport mode switch"
Everyone should look for you, because this is not legal.
By modifying the DJI Assistant program little (debug mode), you can access the Spark parameters and translate a parameter number into the Sport button instead of the Sport mode, but into the "ATTI" mode.

There is also a legal method: wrap the aluminum foil over the Spark top, where the GPS antenna is. The disadvantage is that you can not turn it back on panic.

Mirek6 Posted at 2018-7-6 08:11
Jenda,  

Great and very thorough post.

Mirek6

With forced Atti mode, I am not sure if it is legal. You need older  version of Dji assistant, enable debug mode and simply change action of the sport mode switch. So instead of sport mode you will get Atti mode, when you move the switch. There are some tutorial on Youtube about this, I dnt know if I can share some links here on official Dji forum, but it is really easy to find ;) This mod gave me really big confidence when flying in Atti mode, and when I mastered it, I found it also funny.

I agree that better no calibration, than wrong one. I think, that my first IMU calibration was done bad. I recall, I done it in room with runnning children and I also think that I messed some steps. Same with the compass, I didnt care about metal object nearby. No wonder I had yaw/compass error then.

But I must repeat myself again, worst what can pilot do is allow the AC to disconnect from RC. Then you must only rely on RTH function, which is dependng on your compass, which is known to have issues. This sounds like risk to me. Compass error is not big problem, when you have stable connection.
Speaking of connection, I found from flight logs, that when the image transmited from camera to mobile begins to tear, I still have full signal from RC to AC, obviously the commanding signal is stronger than image sent back to RC/mobile. So when my image begins to be crapy, I dont fly further and I maneuver AC so it has nice image again. By this I am getting warned even before the most important RC to AC signal goes week. Importat is also to know, that wifi signal does not go in straight line, but it has shape of cigar or airship and anything obstructing in this shape weakens the signal. For example, if you have tree 50 meters from you, Ac is 2 meters above the tree and 20 meters behind it, you have clear view from RC to AC, but signal is beiing obstructed.

Jakab Gipsz Posted at 2018-7-6 09:23
"switch to atti mode with Sport mode switch"
Everyone should look for you, because this is not legal.
By modifying the DJI Assistant program little (debug mode), you can access the Spark parameters and translate a parameter number into the Sport button instead of the Sport mode, but into the "ATTI" mode.

Jakab,

Exactly this I ment. I also tried to fly indoors with low light, to disable optical stabilization. This is good for begining, as there is no wind. But you must have bigger room to avoid airflow disturbance  between  AC propelers and walls.

It will land where it is if it disconnects from the RC in ATTI mode.  Roughly land there anyway with some wind movements. DJI policy not to have a disconnected drone flying on its own randomly through the nation's airspace?  If it is in GPS-P mode, obviously, it will try to return to home.  Lots of possible problems in those instances.  Hence me being close to my photography subject and therefore the drone.  It will have to travel quite a ways before it is out of my radio range which gives me time to regain control or spike it into the ground (recoverable).

I have never used RTH.  I don't trust it enough to forgo flying back and landing the Spark myself.

Pretty good and detailed how-to

Good Tutorial

Jenda_cz Posted at 2018-7-6 11:59
Mirek6

With forced Atti mode, I am not sure if it is legal. You need older  version of Dji assistant, enable debug mode and simply change action of the sport mode switch. So instead of sport mode you will get Atti mode, when you move the switch. There are some tutorial on Youtube about this, I dnt know if I can share some links here on official Dji forum, but it is really easy to find ;) This mod gave me really big confidence when flying in Atti mode, and when I mastered it, I found it also funny.

Jenda,

All good points outside of your last few sentences.

You write: “. Importat is also to know, that wifi signal does not go in straight line, but it has shape of cigar or airship and anything obstructing in this shape weakens the signal. For example, if you have tree 50 meters from you, Ac is 2 meters above the tree and 20 meters behind it, you have clear view from RC to AC, but signal is beiing obstructed.”

This interpretation is simply not true.

Yes - Wifi waves are omnidirectional and usually spherical or ellipsoidal (cigar-shaped as you said – with the source smack in the middle of the cigar). Spherical or ellipsoidal shape depends on antennas used and their direction.

However, your interpretation of how it affects signal strength is unfortunately incorrect.
In the case of DJI RC Wifi waves are omnidirectional and ellipsoidal (cigar shaped) in the sense that their strength goes furthest distance perpendicular to antennas and is weakest to the sides. This has nothing to do with obstructions. Yes, electromagnetically passive obstructions such as trees weaken and disperse the signal, but only signal which goes through them (electromagnetically active obstructions may wreck havoc all around). In other words, if you have a clear line between AC and antennas, the signal strength depends on direction of antennas and distance – not a (electromagnetically passive) obstruction which may be below or to the side. Of course, there are many other factors affecting signal strength but obstructions which are electro-magnetically passive and which are not in direct line of sight are not one of them.
Again - electromagnetically active sources (trees are not one of them) may wreck havoc whether your line of signt to AC goes through the source or close to it.

Mirek

Jenda_cz Posted at 2018-7-6 11:59
Mirek6

With forced Atti mode, I am not sure if it is legal. You need older  version of Dji assistant, enable debug mode and simply change action of the sport mode switch. So instead of sport mode you will get Atti mode, when you move the switch. There are some tutorial on Youtube about this, I dnt know if I can share some links here on official Dji forum, but it is really easy to find ;) This mod gave me really big confidence when flying in Atti mode, and when I mastered it, I found it also funny.

Jenda,

You write: "With forced Atti mode, I am not sure if it is legal. You need older  version of Dji assistant, enable debug mode and simply change action of the sport mode switch."

Thanks for the explanation. Yeah - I will not try it. I am sure that it is legal (why wouldn't it be?) but I am also sure that it will void Spark's warranty should something happen to it when you fly it in ATTI which was forced by making changes to Spark's firmware.

Anyways - quite interesting :-)

Mirek

Thanks. Going to use a football, baseball, and farm park for the atti test.  The space should be large enough.

Mirek6 Posted at 2018-7-6 17:13
Jenda,

All good points outside of your last few sentences.

Mirek6

thanks for you input. I maybe write it wrong about the signal. What I wanted to say is, that even some passive obstruction(tree) in the path of the signal shape can weaken the signal. It is called Frensel zone, if you want to make detailed study about this topic

Let me share,what I found (not from my head, I am not taking credit for this, I only found it during my research)
As we spoke, signal is being transited in elipsoid shape.Here is picture of clear connection:




But when is something in the way of this Frensel zone, the signal is not 100% even if there is clear view between RC and AC:




And here is top view, with trees on each side:




Table of the diameter of the zone (the middle) and connection lenght:





Signal is getting wrong if more then 40% of this zone is obstructed. And the growth of the diameter from RC to the middle of the zone is quite steep. If you have 1km connection, after 100m from the RC this zone has allready 3,4m diameter.

I experienced this phenomenon, when I was flying AC in the whole between two big trees. Even if the whole was big enough, as the AC went further (Frensel zone diameter began to growth), video signal got crapy, even the plane sill had direct view to the RC.

Sorry, it is not called Frensel zone, but Fresnel zone

Thank You for posting all of these thoughts/findings. I think that everybody has to review these when using the Spark, especially the IMU and compass calibrations details

Jenda_cz Posted at 2018-7-6 21:08
Mirek6

thanks for you input. I maybe write it wrong about the signal. What I wanted to say is, that even some passive obstruction(tree) in the path of the signal shape can weaken the signal. It is called Frensel zone, if you want to make detailed study about this topic

Jenda,

I am impressed by your thoroughness.

Phenomenon described by the concept of Fresnel zone is caused by deflection of the radio waves off the various objects. It is also caused by some bending of waves which may, usually, be caused by atmospheric conditions. This results in signal phase shift interference which may either strengthen or weaken the signal. Your diagram is simplistic but, nevertheless, has some good points.

While I do not doubt that you may have experienced drop in the signal which could be related to Fresnel zone concept, this is more of a problem with directional parabolic RF antennas.
How do you know that flying between two big trees (as you described) was the effect of Fresnel and not the distance to AC, position of your antennas or some other thing? It would be hard to measure in non-laboratory conditions.

Also, as your table indicates, width of Fresnel zone is really minimal compared to the RC to AC distance (at 500 metres, which is max legal flight distance in Canada and many other countries, it is roughly 1 degree of angle – just calculations in my head without taking out my calculator, so do not quote me here).

It is worth keeping in mind (as you obviously know), that it is much more important to keep Spark’s RC antennas directly perpendicular (flat parts of antennas forwards AC) to the AC especially in iffy conditions. If the strongest signal is hitting AC directly, any phase interference (cancelling or strengthening of your main signal) from either wave deflections or wave bending is minimized.

So, you are right with description of this phenomenon although I do question the picture of why the signal is weaker (you seem to indicate that this is because of smaller area of the travelling signal, which is not the root cause of Fresnel-related issues – phase shift interference is).  

However, I believe that effect of wave bending on WiFi signal which is, by definition very short distance, is really negligible (all other things considered). This is confirmed by your table.

In any case, good points and good summary – thank you.

Mirek   

Mirek6 Posted at 2018-7-9 09:26
Jenda,

I am impressed by your thoroughness.

Hi Mirek,

Thank  you for explanation.
I agree that this is not big deal, antennas direction is more important of course. I jus t found this phenomenom, when I did my research about signal transmition and I found it interesting. That is why I shared it here. I am not expert in radio waves, but Ithink man should alwas study new things
About my experience with flyinig between two big trees on longer distance.. maybe it was not this case, but it made sense, together with all other circumstances. But sure, it could be  combination  of more things.
Anyway, thanks again for you comments, I love to learn new details.
I am still convinced, that only solid cnnection and good pilot skills are the only things that will eventually save our little birds from trip to unknown ;)

Thanks for writing this up. In my experience, having suffered from an unstable Spark (e.g. entered Atti mode on every flight), I would also strongly suggest re-calibration of IMU and compass immediately after performing a firmware upgrade on the drone - even if you haven't changed your location. My Spark was stable out the box, but became unstable after a forced firmware update. After re-calibrating IMU and compass, my drone's been well-behaved.

Thanks for useful observations and advices

first of all, you do not have to calibrate your compass just because you travel long distances, not sure why people think you do or why this myth keeps going so strong.   magnetic north (north pole) is the same anywhere on earth, does not change just because your in a different country/nation.  

according to my friend who know RF theory waaay more than ill probably ever know, the biggest weakness with Spark and the Mavic air isn't lower power outputs its that its wifi.  wifi standards (802.11) are fairly strict standards in terms of communication outside of your home (this is why long range wifi based two way radios dont really exist)  wifi really is not designed for outdoor use where you may start sending data outside of frame just because of tiny reflections off of almost anything.  on top of that competing with other networks on wifi, even if you are out in the middle of seemingly nowhere, doesn't mean your spark/mavic air isnt competing with other wifi signal even if you cant see them, spark/mavic air being higher up maybe more susceptible to running into stray signals from miles away, just because it may not be showing up doesn't mean isn't stray 2.4Ghz or 5.8Ghz signals bursting around in the air bouncing around from distance networks and things making the drone have to compete harder.

Most interesting discussion.

Radio frequency behaver can become very complex depending on the frequency and then a lot of mathamatics and atmosperic conditions at play.
Look-up Fresnel zone and Earth-Moon-Earth (EME) communication on the internet.
EME are used by Radio Amatuers (HAMs) with the moon as passive reflector at frequencies as low as 50 Mhz
At fequencies 50Mhz and above the path are mainly line of sight execpt with special and sometimes rare atmospheric conditions.
Any type of thing in it path can affect the communication quality. It is almost like the beam of a flash light being obstructed by an object.

Have a safe fight all.

HereForTheBeer Posted at 2018-7-10 13:55
first of all, you do not have to calibrate your compass just because you travel long distances, not sure why people think you do or why this myth keeps going so strong.   magnetic north (north pole) is the same anywhere on earth, does not change just because your in a different country/nation.  

according to my friend who know RF theory waaay more than ill probably ever know, the biggest weakness with Spark and the Mavic air isn't lower power outputs its that its wifi.  wifi standards (802.11) are fairly strict standards in terms of communication outside of your home (this is why long range wifi based two way radios dont really exist)  wifi really is not designed for outdoor use where you may start sending data outside of frame just because of tiny reflections off of almost anything.  on top of that competing with other networks on wifi, even if you are out in the middle of seemingly nowhere, doesn't mean your spark/mavic air isnt competing with other wifi signal even if you cant see them, spark/mavic air being higher up maybe more susceptible to running into stray signals from miles away, just because it may not be showing up doesn't mean isn't stray 2.4Ghz or 5.8Ghz signals bursting around in the air bouncing around from distance networks and things making the drone have to compete harder.

HereForTheBeer,

You write: “first of all, you do not have to calibrate your compass just because you travel long distances, not sure why people think you do or why this myth keeps going so strong.   magnetic north (north pole) is the same anywhere on earth, does not change just because your in a different country/nation”

Yes – magnetic north does not change its geo position but magnetic declination, inclination, strength of magnetic field and three-dimensional properties of magnetic field certainly do change in different places on Earth and these changes certainly do affect magnetometer (compass).

Please see my post for more detailed explanation of how it works: https://forum.dji.com/forum.php? ... 9&page=2#pid1433039
Also, please note that DJI recommends calibrating compass if you travel more than 50 km (it is in DJI Mavic Air manual – they did miss it in Spark’s manual).
BTW – they also recommend calibrating it if you did not fly for longer than 30 days.

As for the other part of your post. Yes – WiFi is quite susceptible to interference. But this is only if interference is close (not miles away). If you loose signal in the middle of nowhere it will not be from interference (unless you brought your microwave oven with you :-).

LTE towers are dangerous since they may operate within bands very close to Spark’s WiFI frequencies. For example, LTE Band 40 operates dangerously close to 2.4GHz. Also, strong cell tower signals do disrupt Spark’s communication if close enough. There are many other interference sources – urban environment, high voltage transmission lines etc. - to mention just a few.

  
  

Mirek

Mirek6 Posted at 2018-7-10 14:43
HereForTheBeer,

You write: “first of all, you do not have to calibrate your compass just because you travel long distances, not sure why people think you do or why this myth keeps going so strong.   magnetic north (north pole) is the same anywhere on earth, does not change just because your in a different country/nation”

i only recently calibrated my mavic pro's compass and i owned it for a year before i did that and taken it thousands of miles around without any issues even taken off from boats, atv's, top of my car, etc etc even if it freaking out about compass error once its away from interference its happy again, never had an incident.    the reason you probably do not want to calibrate your spark's compass very often and why its missed in the manual is the spark only has 1 single compass, not reference or redundant compasses like there is in pro model drones which can just change which compass its using on the fly if one of them is having trouble.

i flown next to some 5-10Kw TV antennas that taken my GPS and VPS offline on my mavic pro and forced me to have to reboot the drone to get them back, and i flown around cell towers without issues sometimes briefly get a smeary frame but nothing bad, MAvic pro uses Occusync which is in the 2.4Ghz band just like wifi, however it does not comply with 802.11 standards so its far less subject to interference since it can still force data through and can be out of order plus it has frequency hopping, still stuck to 2.4Ghz for that hopping but it can narrow its bandwidth to 10Mhz if it needs to..   and yes wifi can go miles, specially if outdoor antennas which a lot of large places use... doesnt mean by the time it reaches you miles away that its 802.11 compliant or usable or even visible on phone/tablet, but still a stray signal consuming the spectrum.   also have to understand radio horizon, drone being up 50m lets say puts a massive radio horizon on it where it will inevitable pick up stray signals way way more then you 2m off the ground.

i no longer have a spark and im probably not going to get a wifi based drone again.. even the 2nd spark i owned was horrible with signal quality and i heard similar accounts with the much more powerful enhanced wifi of the Mavic air, super inconsistent signal quality and i cannot give up Occusync, been such an amazing wireless system.

I have a question for Mirek6, and I hope I'm not hijacking thread or discussion. Very informative thread overall.

I am a very happy Spark pilot, and I'm running latest firmware, 4.1.15 (Android), FCC and OTG.
My location is Europe.

In April I was on 4.2.8 and had tons of Compass-, Speed- and Yaw errors.
According to logs at Airdata, almost every single log contained some or all of these errors.

Downgraded to 4.1.15 and calibrated IMU and Compass.
After these actions taken I have 60+ logs at Airdata without a single error. EVER!
My flight-pattern or locations has not changed.
I go close to pylons and LTE towers.
I do high wind and over water.
Always on 2.4 Ghz channel.
Every now and then I go nuts in Sport-mode combinated with high wind.
I doesnt take any precautions at all, just enjoying my bird wherever whenever etc.

@Mirek6:
Most of replys is regarding and explaining fly-aways.
I wonder what kind of enviroment or pilot actions is known to be sure these errors occur?

This is an interesting question, if app version can influence these compass error.

I personally think, that these position calculations are made directly by Spark himself and there is no way how mobile app can interfere with this. Did anybody make some tests?

Jenda_cz Posted at 2018-7-11 04:58
This is an interesting question, if app version can influence these compass error.

I personally think, that these position calculations are made directly by Spark himself and there is no way how mobile app can interfere with this. Did anybody make some tests?

Obviously I had a really successful calibration.
But lately I have tried to provoke these errors, but no. Not possible.

Concrete and rebar..
Powerlines..
Close...really close to metal objects..

I've tried all no-gos..except strapping a Neodym magnet on top of it.

Thank you for sharing.

Haggi Posted at 2018-7-11 05:18
Obviously I had a really successful calibration.
But lately I have tried to provoke these errors, but no. Not possible.

Well Haggi...,

My experience is similar.
I am very careful with calibration and my Spark never displays any issues (well, my previous Spark did develop issue with its gimbal not being straight but it was simple h/w failure which was warranty repaired by DJI).

Even if I push it really, really hard. Must be really good piece of technology :-).

But seriously.
Properly calibrated Spark with no h/w issues is really very resilient. By pushing it to its limits close to sources of interference you are increasing chances of errors but not guaranteeing errors.
Did you check your detailed flight logs after doing some of these experiments? Chances are that you did not see messages on DJI GO App but will see errors in logs. But, again, no guarantee. Just higher probability.

I do trust my Spark but I always do a quick pre-flight check. I do not trust DJI GO app status indicators for IMU and compass (they are buggy and I have seen multiple times cases with normal indicators but troubled flight due to calibration issues)  so I do calibrate IMU and compass when I think it is sensible to do so. For example when I experienced something unexpected during flight. I do push my Spark and I do fly in heavy wind but only when return path is with the wind (so I know I can direct it towards me with no problem), not against heavy wind.

What will, for sure, cause Spark failure? Well – strapping Neodym magnet on top of Spark, as you suggested, would be one of my front runners in winner category :-).

I can only advise you, no matter what your experience is, simply avoid sources of potential trouble, if not necessary. One day it may kick you. Or, it may not. It's like smoking. Some die of it and some live to be hundred. You may be lucky one. Or not ...

Mirek

This article really helpful thank you for sharing.