Mirek6 Posted at 2018-7-16 08:22
msinger,

I know it is already past the fact - however, since the thread is active again.

Mirek, do not want to agree with you on compass calibration. Personally have travelled from Latvia (EU) to Florida (US) last decembre and decided not to touch the compass until i will not be asked to. And guess what - i never was asked to calibrate, but the Spark behaved excellent

BudWalker Posted at 2018-7-19 05:36
You've made several posts. I'll respond to them here in just this post.

I wasn't involved in the Nilesh incident. I didn't see the logs. You probably had me confused with msinger. I'll be glad to review the logs but will probably come to the same conclusion msinger did.

Bud Walker,

Yes - I understand all that. And I did not say that magnetometers point to true north. I said true magnetic north. Huge difference.

In any case, I did send you a private message so we can take this discussion - however interesting it is - off line. I found some scientific articles proposing calibration of magnetometers without knowledge of geo-magnetic field characteristics which depend on location. I just don't know which method DJI is using. Location dependent or location independent. Perhaps we are both right - for some magnetometers you need to measure geo-field in a particular location in order to calibrate properly and for others, using different algoritm, you don't.
Mirek

nixuspix Posted at 2018-7-19 05:52
Mirek, do not want to agree with you on compass calibration. Personally have travelled from Latvia (EU) to Florida (US) last decembre and decided not to touch the compass until i will not be asked to. And guess what - i never was asked to calibrate, but the Spark behaved excellent

Nixuspix,

Singular behaviour is not a proof. I am not talking about whether compass works or not when moving long distances - granted - in many cases not a problem. Characteristics of geo-field may be similar or they may not differ that much to matter. I am talking about principles of how calibration works in general and why it is required (for various reasons).

Having said that, I just discovered some articles about methods of magnetometer calibration which do not require physical location knowledge or measurements. I just don't know which algorithm DJI uses. I took this off line with Bud Walker. I am interested in his opinion after reading that.

The question in my mind remains: why DJI recommends compass calibration when moving more than 50km? Perhaps they use calibration mechanisms which do depend on physical location.

Anyways - bottom line - I am no longer certain. I would like to get to the bottom of this. What I know for the fact though is that there are methods which do and do not depend on location. I also do know for the fact that DJI does recommend calibration when changing location (over 50 km). This recommendation is not in Spark's manual but is in Mavic Air manual.

Mirek   

You see, some time ago DJI even recommended to calibrate before every flight! I think that correct, really CORRECT manipulations during calibration will not harm anyway. But for me, in most cases people do not perform calibration process as it should be done - with high precautions to avoid any possible disturbing influences, resulting in worse compass situation, after such inaccurate calibration.
I think that's  why they started to advise not to calibrate every time before flight

nixuspix Posted at 2018-7-19 06:20
You see, some time ago DJI even recommended to calibrate before every flight! I think that correct, really CORRECT manipulations during calibration will not harm anyway. But for me, in most cases people do not perform calibration process as it should be done - with high precautions to avoid any possible disturbing influences, resulting in worse compass situation, after such inaccurate calibration.
I think that's  why they started to advise not to calibrate every time before flight

That's all true. Bad calibration may be worse than no calibration. No doubt about that.
I am just trying hard to get to the bottom of what is the best recommended insurance policy against compass errors (given that you do calibration properly).

Perhaps I will never know without DJI providing definite answer.

And, frankly speaking, I would not trust 100% their answer either, unless I am talking directly with a developer responsible for magnetometer algorithms. Unfortunately DJI Support does provide uninformed answers - depending on who you are talking to.

In a case from few days ago (I do not recall which one - so many of them),  DJI Tony (I believe), said - re-calibrate compass when moving long distance. He confirmed what is written in Mavic Air manual.

Is it the proof that this is how the Spark magnetometer works? No.
Is it likely? Yes.
Can we get to the bottom of it? Not sure.

Cheers.

Mirek

Mirek6 Posted at 2018-7-19 05:58
Bud Walker,

Yes - I understand all that. And I did not say that magnetometers point to true north. I said true magnetic north. Huge difference.

Thank you, but I'll continue here. Your comment that others may become confused by this conversation doesn't make sense to me. Others can always ask questions or ignore this.
The article you sited
ac.els-cdn.com/S1474667015350400/1-s2.0-S1474667015350400-main.pdf?_tid=e777d530-2eaf-4b5b-8254-40a83a2ca165&acdnat=1532007753_714244eb3a8376741804917c14edb90b

doesn't open for me. If you can provide another link I'll take a look.

Since I've already read the previous article you sited maybe you could take a look at the article I sited. It gives a good explanation of how the compass calibration works.


https://www.sensorsmag.com/components/compensating-for-tilt-hard-iron-and-soft-iron-effects

Just to be clear all this discussion comes down to one key statement. A compass calibration does not and can not detect/compensate for any magnetic effects that do not rotate with the AC. I.e., if a magnetic effect is external to the AC it can not be compensated for by the calibration.

BudWalker Posted at 2018-7-19 11:18
Thank you, but I'll continue here. Your comment that others may become confused by this conversation doesn't make sense to me. Others can always ask questions or ignore this.
The article you sited
ac.els-cdn.com/S1474667015350400/1-s2.0-S1474667015350400-main.pdf?_tid=e777d530-2eaf-4b5b-8254-40a83a2ca165&acdnat=1532007753_714244eb3a8376741804917c14edb90b

Bud Walker,  

  

I will answer in two posts – easier to read. Please bear with me. We are getting somewhere.

  

Post 1:

  

You say: “Since I've already read the previous article you sited maybe you could take a look at the article I sited. It gives a good explanation of how the compass calibration works.”

  

You imply something that just isn’t. Can’t just put my finger on how I can make myself clearer to avoid misunderstandings.  

I did read your article. I did reply to you in post 37:

• I did tell you that we both know very well how magnetometer works.
• I did tell you that I do understand calibration to compensate for hard and      soft iron distortion (which is a topic of the article you attached – very clear      and interesting one for that matter)
• I  did tell you that I used the hard/soft iron effect to explain some of the      questions people on this forum had and to help them with resolving their      cases.
  

We cannot continue this discussion in open forum if we do not take time to read our words very carefully and try to understand very hard what you or I meant and where we are coming from. Hence my attempt to take this as a private discussion – which you refused.

  

Yes – misunderstandings and misinterpretation of each other words – which we are obviously excelling in – are confusing for us, not to mention people reading this discussion.  

  

For me this discussion is not an empty argument but a chance to learn and perhaps get to the bottom of some controversial recommendations.

  

In addition to explaining about compensation for hard/soft iron disturbances your article you adds additional information about compensation for tilt – but this is not relevant to magnetometer calibration (it is relevant to correcting AC heading by accounting for magnetometer decreased sensitivity due to elevation and bank angle).

However, this compensation for tilt – I must admit – was new and very interesting.

   

The article you provided is titled: “Compensating for Tilt, Hard-Iron, and Soft-Iron Effects”. And this is exactly what it describes. It is not intended as a full explanation about all factors which need to be taken into account when calibrating compass. It does not talk about location and various properties of geo-magnetic field in different locations. Perhaps because location data is not important to Chris Konvalin (author of this article), or perhaps because it is simply not the topic of the article. The topic is how to compensate for tilt and hard/soft iron effects.

Read on Post 2 :-)

Mirek

BudWalker Posted at 2018-7-19 11:18
Thank you, but I'll continue here. Your comment that others may become confused by this conversation doesn't make sense to me. Others can always ask questions or ignore this.
The article you sited
ac.els-cdn.com/S1474667015350400/1-s2.0-S1474667015350400-main.pdf?_tid=e777d530-2eaf-4b5b-8254-40a83a2ca165&acdnat=1532007753_714244eb3a8376741804917c14edb90b

Bud Walker,

Post 2:

  

You say: “Just to be clear all this discussion comes down to one key statement. A compass calibration does not and can not detect/compensate for any magnetic effects that do not rotate with the AC. I.e., if a magnetic effect is external to the AC it can not be compensated for by the calibration.”

  

Another misunderstanding.

  

No – this argument is not about compensating for magnetic effects that do not rotate with AC. This argument is about measurement of geo-magnetic field and accounting for geo-magnetic field properties in a given location and than compensating for magnetic effects within AC. I do agree with you that magnetic effects or disturbances which are external to AC cannot be compensated for by the calibration. But geo-magnetic field properties metrics, in particular geo-magnetic field strength as measured in all three axis, is knowledge required as a prerequisite to successful compensation to account for soft/hard iron disturbances within Spark.

  

Here is another link to the article from Honywell Technology Solutions I referred to – I hope this one works: https://www.sciencedirect.com/science/article/pii/S1474667015350400 .

This article describes a method of calibrating magnetometer when true geo-magnetic field strength is unknown (field strength is geo-location dependent). Algorithms which are location independent and are commonly used require very elaborate set-up and rotating magnetometer in three axis and are very complex (Spark’s calibration is 2-axis calibration, horizontal and sideways, and is damn simple – can’t use such algorithms). Authors of this article came up with the alternate algorithm which is simpler and is 2-dimensional. This algorithm does not require measurement of the geo-magnetic field strength in a given location to result in accurate calibration.

Please note start of chapter 2.4 “Calibration does not require location information”. The key assumption to have successful calibration is knowledge about exact radius of the circle – it must be known quantity. Radius of the circle is proportional to the total strength of the true magnetic field. True magnetic field is highly geo-location dependent.

This is why most magnetometers should be re-calibrated when the geo-location changes. The compensation for hard/soft iron distortions calculated in one place may be invalidated by different properties and strength of geo-magnetic field in different location.

  

However, authors of this article do propose a method where calibration may be accurate and independent of the geo-location.

  

In summary:

• Widely used algorithms to calibrate magnetometer must take into account magnetic field strength in a given location. This is prerequisite before adjusting for hard/soft iron distortions. Location-dependent strength of geo-magnetic field can be either computed from tables based on location parameters or can be measured on the spot.
• Algorithms which do not require location information are complex and require elaborate set-up and 3-axis measurements (Spark is not using such algorithms for obvious reasons).
• The article I quoted provides simpler method which also does not require data about geo-magnetic field strength as a prerequisite to successful calibration.
  

I do not know which algorithm Spark uses.

I did not know that non-location specific calibration methods existed until I started research based on conversation with you.

I did know that for regular and simple calibration algorithms you do require measurement of geo-magnetic field in calibration location as a prerequisite.

I do suspect that Spark uses simple algorithm and the simple ones do measure geo-magnetic field strength and than provide method of calibration in 2-axis which do account and adjust for hard/soft iron distortions.

  

My suspicion is based on simplicity of Spark design – it is not a laboratory-quality equipment. While advanced technology, it is also fairly cheap.

  

In addition, DJI does recommend re-calibration after moving long distances (the question is – do you believe them here – I do).

  

Mirek  

hey ... good post and question
I've wrote a post here : https://forum.dji.com/thread-152418-2-2.html
post click here
where you can find on #73 one pdf file to download very rich of TIP and check list

just , let me know , what do you think about there ...
thanks Alex

Just my opinion. You will be flying more than 200m away to get good shots. 200m is nothing for the Spark. But I understand not wanting the push it too far because it's new equipment to you. Make sure you have SOLID GPS lock before flying. This includes not flying too close to buildings and metal objects. (This can include metal in concrete like sidewalks) Most of the time when I hear about crashed drones it involves people rushing the shot and flying before they get GREEN GPS lock.

My BIGGEST tip to you is to fly your Spark every day (multiple times a day if you can charge your 3 batteries and fly again) until you go on your trip. Only EXPERIENCE will save your drone when you get into problem situations. Just fly as much as possible and get comfortable with all aspects of your new drone.

Mirek6 Posted at 2018-7-18 17:41
Bud Walker,

One more thing about your graphs which I find interesting. I now understand that Spark's wait to compensate for magnetic declination until GPS signal gets strong enough is necessary. Correcting before GPS is firm could lead to errors and wrong direction of true north as seen by Spark. This is logical and good design.

The compass calibration doesn't detect true north. It detects how much interference the drone emits and it subtracts it from the electromagnetic readings. That's why it doesn't make sense to recalibrate after traveling, the drone is the same and the interference it emits is the same anywhere on the planet.

ssakellaro Posted at 2018-7-19 23:24
The compass calibration doesn't detect true north. It detects how much interference the drone emits and it subtracts it from the electromagnetic readings. That's why it doesn't make sense to recalibrate after traveling, the drone is the same and the interference it emits is the same anywhere on the planet.

ssakellaro,

I am not sure why people skew my words which I never have spoken :-(.
Please do not oversimplify things which are quite complex - just read carefully what was said in post 48 and try to understand why.

Yes - compass calibration does not detect true north. I never said it does. I said that compass (not its calibration) does detect magnetic north - this is what compass is designed to do and this is why it exists in the first place.

Compass calibration makes sure that compass can detect magnetic north correctly and not some other direction.

During compass calibration, magnetometer assesses and measures disturbances (hard and soft iron) - which Bud Walker and I discussed at length above. However, In order to adjust for these disturbances, most calibration algorithms need to know the qualities of the geo-magnetic field at the location of calibration in order calculate and  apply proper adjustments based on disturbances. If the qualities of geo-magnetic field change, adjustments calculated in different place will not be valid any more.

There are algorithms which do not require knowledge of geo-magnetic field properties as a prerequisite to applying adjustment. But these are usually complex and require 3-dimensional axis calibration setup which is not the case in Spark.

I attached a scientific paper which proposes simpler algorithm which would work without location knowledge and with 2D setup.

As I said - I do not know if Spark uses some elaborate location-agnostic algorithm for calibration but I strongly suspect it does not. It is simply too cheap a piece of equipment. My suspicions are confirmed by what DJI recommends. That's all.

If you know that re-calibration is not required (when moving long distance to a new geographical location) because you know exactly how the calibration algorithm works in Spark (perhaps you designed it) - power to you. Do not re-calibrate.

If you don't know for sure - well ...

Mirek

AirButterflyDrone Posted at 2018-7-19 21:57
hey ... good post and question
I've wrote a post here : https://forum.dji.com/thread-152418-2-2.html
post click here

Alex,

I posted this feedback on your original thread as well.

Most excellent file and tips! I saved it for myself as a reference.
This is something which we really do appreciate – a set of tips and cautions in one place for easy reference.

My feedback:

Most of your specific instructions do not apply to Spark (but you do say that it is for Inspire or Phantom). Perhaps adding at the beginning of your tips that some specific instructions may not apply to Spark or Mavic may be good. Or perhaps say that your screens are Inspire or Phantom specific.

Chapter 1

You did confirm what I am saying about need for calibration when travelling long distances and you eluded in simple terms why: “ it stores the data and assumes that these forces are normal for the current position and will be consistent throughout the flight.”
Your wording is not scientific, simple and the recommendation summarizing them is correct.

I would add the following to make your description and recommendation complete:

Due to the delicate nature of Spark’s compass, it can loose its calibration if Spark is exposed to strong magnetic interference (for example if you put Spark in the vicinity of your car subwoofer when driving). DJI GO app may or may not detect problems with compass calibration. Be vigilant for any “compass error” or “yaw error” warning messages. If you see such errors during flight and there is no obvious source of interference, re-calibrate your compass (you may want to re-calibrate IMU as well) – just to be on the safe side.

When you calibrate your compass, turn Spark slowly – do not rush. Let the magnetometer do its work correctly and precisely.

Chapter 2

In your recommendations for IMU calibration you say: “ Each foot can be adjusted by placing pieces of paper underneath”.
This is not necessary (but will not hurt either). Yes – I know the people asked and discussed this. However, when you calibrate IMU you calibrate it in each of 3 axis. When you put it on the side and the prop arms position cause the Spark to angle a bit it is OK because IMU, at that moment, measures different plane – horizontal plane along Spark’s axis and not the vertical plane which is slightly skewed.

Chapter 4

You say: “It is important to land with reserve power, because if something goes wrong and you are forced to delay the landing it is necessary enough time to find a new landing spot and solve the problem.
“
I would add for completeness – “Remember to account for the wind. If your return home is against the wind it may take longer time, use more battery power and you may not make it back. RTH sequence does account for wind, but be extra cautious.”

Chapter 6

For Spark I would recommend to set “RTH at current altitude”. It prevents Spark from landing on the spot when less than 20 metres from HP. I have seen cases on this forum where Spark landed in water when RTH was invoked because people did not understand how RTH works. This is just extra insurance policy.

Chapter 7 does not apply to Spark

Chapter 8

When you set mobile in air mode you have no access to map. When you do so, cache the map before you fly.

Chapter 9 does not apply to Spark

Chapter 9 (again – mistake in chapter numbering) – talking about propellers

I advise against using third-party propellers. Propellers are too important a part of Spark. I do err on the side of caution.

Chapter on other useful things

I like UAV Forecast application (you did not mention it) – highly recommended.
In your travel with a Drone on a plane chapter I would add that some airlines do require that lithium-ion polymer batteries are discharged to 20-30% of their capacity before you can take them on the plane.

Mirek

Mirek6 Posted at 2018-7-20 05:54
Alex,

I posted this feedback on your original thread as well.

Hey Mirek
Thanks a lot for your detailed analysis...
the important thing is , have a proper checks to make sure do not have any problem during your flight
Planning your flights and try to take account of everything... like interference, higt , wind , weather, distance and others...
think about what and how you want to shoot
Then .... again.... checks list
It’s true ... in pdf file I have try to put in the same place tip and advise in order to remember my points...
(BTW Thanks again to improved it )
And then... flying and flying again, try , have fun , shoot , take pictures and fun
I’m only a drone appassionate ... but I think that in this play/work , my way is have fun and  to try to improve again and again... for exaple right now I’m trying a new app with some new functions...

Mirek6 Posted at 2018-7-17 08:58
Tyr76,

Does you question have anything to do with what was described above? I do not see any connection.

I was referring to the original post, in his checklist he mentions not bringing protective covers, so I wondered why not.

Tyr76 Posted at 2018-7-20 08:39
I was referring to the original post, in his checklist he mentions not bringing protective covers, so I wondered why not.

Ahh, OK - sorry.
Yes - your point is very valid.
I always have protective gimbal cover which I use for transporting Spark  (as a matter of fact I also have a spare one since plastic gimbal cover has a very weak notch which tends to break off). I would highly recommend that.

I have RC cover to protect stick during transport and I do take propellers off during transport and put a small rubber plugs on the ends of propeller arms for protection.

Mirek