Fixing DJI's Compass Problem

[ianwood]

Quit while you're ahead. This has nothing to do with the precision of the GPS. Notihing. You make it sound like adjusting for declination is some sort of next generation rocket science. It's literally 3 lines of code and a table. Give me the Naza M source and I'll do it. In about an hour.

If GPS home location is set:
1. Use GPS coordinates to find nearest declination reference from lookup table
2. Offset compass reading by declination for this flight

That's not the problem though. The problem is they did put it in and the code is defective and it actually makes it twice as bad.

So expecatiations? I can fly my P2 up in the air, send it completely out of sight, turn off my controller, sit down and drink a beer while it comes back to me all by itself. That's just about the coolest thing I've seen in a while. Yet that same thing can't fly straight because a few simple lines of code were implemented incorrectly.

It ruins a number of tracking shots because I have to intercede with corrective action for more than half of their flight.

 

[sar104]

I suppose people just want what the thing is supposed to deliver. If the company said all the GPS is for is to bring it home safely then your argument would be valid but they promise it will fly itself.. A lot of people including myself wouldn't buy a machine like this if it wasn't extremely easy to fly.

I get what you are saying, but a part of me thinks perhaps we are expecting too much from these things. They are consumer grade devices. They are not professional level equipment. For what they are, they are amazing. THe ease of flying allows you to focus on what they are intended to do, which is obtain aerial images (video/still). But even there, the camera is good, but far from professional level.

I work with surveyors who use GPS location equipment that blows away my Garmin. It also costs 2 orders of magnitude more than my little Garmin. So for me, I see some of this as just the limits of how well a relatively low cost system can be expected to perform. Add to that the fact that the "limits" can be minimized by doing just a little active piloting.

It is great that DJI seems to want to improve their product, but I wonder if they will be able to overcome the inherent limitations of the device itself. So is this really a "defect" or is it just the best the system can produce??

You may be right that expectations are sometimes unrealistic, but it doesn't change the fact that something is amiss with the way that the FC is handling declination. It apparently flies fine in areas of low declination and so, since correcting for declination is trivial and since the FC is clearly capable of stable and accurate flight, it is not unreasonable to expect it to be done correctly. It should not be necessary to use manual flight corrections to prevent track errors of 30 degrees and instabilities in hover.

 

[SilentAV8R]

You make it sound like adjusting for declination is some sort of next generation rocket science. It's literally 3 lines of code and a table. Give me the Naza M source and I'll do it. In about an hour.

I'll defer to your expertise on this. So you are saying that there is no difference in the hardware, it is an entirely software driven phenomenon that is easily corrected with 3 lines of code?? Makes you wonder how the guys at DJI missed that. They were smart enough to get it to where it is now, but stopped just 3 lines short of perfection?

If it really is that trivial, then I hope they can fix this soon.

 

[SilentAV8R]

since correcting for declination is trivial and since the FC is clearly capable of stable and accurate flight, it is not unreasonable to expect it to be done correctly. It should not be necessary to use manual flight corrections to prevent track errors of 30 degrees and instabilities in hover.

As I noted above, I have no idea how trivial or complex it is to make this correction and I do not know if it is 100% software driven. I do know that all measurement and electronic devices have an inherent accuracy and I do not know if this will also require a change in hardware. It will be interesting to see what DJI comes up with in the way of a fix.

 

[SilentAV8R]

Quit while you're ahead.

I suppose I should. You've made it clear that any dissenting opinions that are not in lockstep of yours are not welcome. I'll leave you to it then.

 

[sar104]

since correcting for declination is trivial and since the FC is clearly capable of stable and accurate flight, it is not unreasonable to expect it to be done correctly. It should not be necessary to use manual flight corrections to prevent track errors of 30 degrees and instabilities in hover.

As I noted above, I have no idea how trivial or complex it is to make this correction and I do not know if it is 100% software driven. I do know that all measurement and electronic devices have an inherent accuracy and I do not know if this will also require a change in hardware. It will be interesting to see what DJI comes up with in the way of a fix.

Noted, but it really is simple, since it is an approximately constant (in any given location) angular offset between the horizontal component of the magnetic field vector and the direction of true north (as used by the GPS earth model).

Regular compasses, which only measure the direction of the horizontal component, have traditionally corrected for this by a simple rotation of the bezel, and an electronic compass can do the same. The complication with a flight compass is that during flight it does not remain horizontal. Even so, it can be corrected by rotation (as with the Naza pucks), provided that the FC expects that and makes the 3D corrections, or simply by adding the declination value for the location to the measured bearing. As ianwood mentioned - that need only be a couple of lines of code.

 

[landonkk]

As I noted above, I have no idea how trivial or complex it is to make this correction and I do not know if it is 100% software driven.

Well then, with all due respect, without fully educating yourself on the issue, as ianwood and many others on this thread have been doing for the past 6 weeks, maybe you shouldn't just 'pop in' and essentially call all of us whiners that are expecting too much from our toy.

All I am expecting is the same performance as the guy flying one in Kansas.

 

[obiwan_pierogi]

As I noted above, I have no idea how trivial or complex it is to make this correction and I do not know if it is 100% software driven.

Well then, with all due respect, without fully educating yourself on the issue, as ianwood any many others on this thread have been doing for the past 6 weeks, maybe you shouldn't just 'pop in' and essentially call all of us whiners that are expecting too much from our toy.

All I am expecting is the same performance as the guy flying one in Kansas.

Exactly!!

 

[varix]

Hi Everyone

First post here. Just wanted to comment on this issue.
I got my Phantom 2 last week and was very excited. Had been doing heaps of reading on all the features and pre flight checks so I would be ready to go when it arrived.
Had a few flights to start practicing and one of the things I noticed quite quickly was the Toilet Bowl effect when I let go of the sticks. Didnt really think much of it and thought perhaps I had just not calibrated the compass. It doesnt really seem to do it when its low to the ground but almost always does some form of it when its up 10-20m or more.

Had a chat to my mate who has a bigger multi rotor and he mentioned he had been reading about magnetic declination. So I did some reading and found ianwoods amazing document. Sure enough everything in his write up made sense.
I live in Christchurch, New Zealand and so when I looked up the declination charts I realised that I am unlucky enough to have some of the worst declination in the world!

Anyways, put me down as another user that has all the issues mentioned in ianwoods write up. I really do hope I haven't forked out over 2 grand for a device that isn't going to work properly in my part of the world. Lets hope there is a firmware update coming our way soon. I am happy to do any testing if needed being that my declination is huge.

 

[d4ddyo]

Quit while you're ahead. This has nothing to do with the precision of the GPS. Notihing. You make it sound like adjusting for declination is some sort of next generation rocket science. It's literally 3 lines of code and a table. Give me the Naza M source and I'll do it. In about an hour.

If GPS home location is set:
1. Use GPS coordinates to find nearest declination reference from lookup table
2. Offset compass reading by declination for this flight

That's not the problem though. The problem is they did put it in and the code is defective and it actually makes it twice as bad.

So expecatiations? I can fly my P2 up in the air, send it completely out of sight, turn off my controller, sit down and drink a beer while it comes back to me all by itself. That's just about the coolest thing I've seen in a while. Yet that same thing can't fly straight because a few simple lines of code were implemented incorrectly.

It ruins a number of tracking shots because I have to intercede with corrective action for more than half of their flight.

I was ok with the way DJI handling this until I found out that it's taking a back seat to other "pressing" problem. Thus is frustrating now.

 

[sar104]

Quit while you're ahead. This has nothing to do with the precision of the GPS. Notihing. You make it sound like adjusting for declination is some sort of next generation rocket science. It's literally 3 lines of code and a table. Give me the Naza M source and I'll do it. In about an hour.

If GPS home location is set:
1. Use GPS coordinates to find nearest declination reference from lookup table
2. Offset compass reading by declination for this flight

That's not the problem though. The problem is they did put it in and the code is defective and it actually makes it twice as bad.

So expecatiations? I can fly my P2 up in the air, send it completely out of sight, turn off my controller, sit down and drink a beer while it comes back to me all by itself. That's just about the coolest thing I've seen in a while. Yet that same thing can't fly straight because a few simple lines of code were implemented incorrectly.

It ruins a number of tracking shots because I have to intercede with corrective action for more than half of their flight.

I was ok with the way DJI handling this until I found out that it's taking a back seat to other "pressing" problem. Thus is frustrating now.

Mu guess would be that the more "pressing" problem has the potential to be expensive for them - such as replacing lost or damaged defective equipment. This issue just annoys a subset of their customers.

 

[nzvideoguys]

Had a chat to my mate who has a bigger multi rotor and he mentioned he had been reading about magnetic declination. So I did some reading and found ianwoods amazing document. Sure enough everything in his write up made sense.
I live in Christchurch, New Zealand and so when I looked up the declination charts I realised that I am unlucky enough to have some of the worst declination in the world!

Hey there Varix. Myself and two friends are also here in ChCh NZ and having the same problems. Hoping DJI gets it sorted asap. Which part of ChCh do you live in? I'm in Spencerville.
Cheers Paul

 

[varix]

Hey there Varix. Myself and two friends are also here in ChCh NZ and having the same problems. Hoping DJI gets it sorted asap. Which part of ChCh do you live in? I'm in Spencerville.
Cheers Paul

Hi Paul

I am in Casebrook, so not all that far from you. Surely a fix cant be far. DJI seem to be pretty onto it. I am still learning so it doesn't really affect me all that much as I am not doing fast straight line runs or anything yet, so no J turn problems.

 

[d4ddyo]

Hey there Varix. Myself and two friends are also here in ChCh NZ and having the same problems. Hoping DJI gets it sorted asap. Which part of ChCh do you live in? I'm in Spencerville.
Cheers Paul

Hi Paul

I am in Casebrook, so not all that far from you. Surely a fix cant be far. DJI seem to be pretty onto it. I am still learning so it doesn't really affect me all that much as I am not doing fast straight line runs or anything yet, so no J turn problems.

Dji is not on top of it. They have put it off for more "urgent " issues

 

[bignick243]

Does this particular problem only happen with the phantom's? What about the flame wheels?

 

[tom3holer]

What is confusing is that this issue is present in their most popular Naza-M V2. Having flown that controller in my Discovery Pro I am familiar with it. Simply rotating the GPS/compass "Puck", which DJI recommends, will eliminate the problem.

Having no experience with the A2 lind from DJI it would be interesting to see it the issue is the same on that controller.

Tom

 

[Alib]

whats the downside to rotating the compass?

 

[tom3holer]

Well on the Naza-M equipped platforms it is recommended by DJI to do just that.

DJI says not to do it on the P2 and indeed it is difficult the way it is mounted. I did do it on mine and it seemed to fix the issue. It has been reported that after flying a bit with a rotated compass it will sort try and self correct and messes things up. I did not see that but only had 5 or so flights till it was lost.

Tom

 

[obiwan_pierogi]

whats the downside to rotating the compass?

I can only fly my pv2 with the compass rotated about30 degrees, otherwise, I have problems controlling the copter... The only downside I have found is the swings at the end of a straight run are still there, and unpredictable... Fly in a Southerly direction, and the copter drifts left.. Turn it around and fly in a Northerly direction, and it drifts right... The only good things: It has a more stable hover.. And the toilet bowl effect is gone. ...Oh, and it's hard to rotate the compass less than 30 degrees, because of the way the leg is designed-- half round.. I secure it with some latex silicone, as it tends to loosen, given the leg design.

 

[bignick243]

Has anybody tried to put a naza m in a phantom 2? From what i under stand you lose the led's but you could probably rigg something up. Then drill a hole in the top for the a mast and then put the gps/compass on it and rotate it to make up for the declination.

Would this work or is this crazy talk?