Fixing DJI's Compass Problem

[landonkk]

It's possible that I have done as much reading as anyone regarding these quad control issues. In addition, I've used 4 different versions of firmware on my Phantom 2 quads, including both 2.0 and 3.0.

I firmly believe that Phantom 2 Assistant versions 2.0 and 3.0 have a programming error, and that version 1.08 does not have this error.
My suggestion regarding version 1.08 is: try it! You just might like it! If not, you can always go back to version 2.0 or 3.0 if your experience is different from mine.

The above is my honest opinion. I could be wrong. I'm not trying to claim that I know more than anyone else. I just want to share that I've found a solution that works for me.

Thanks for the tip but for those of us with the new H3-3D gimbal reverting back to 1.08 kills it. Everyone appreciates the suggestions to rotate the compass and to revert to earlier firmware but lets be clear...

The purpose of this thread is to get DJI to fix the problem at it's root cause. It is intended to confirm the issue and narrow down the cause. These "slap some tape on it" fixes don't really solve the issue.

 

[obiwan_pierogi]

[quoThanks for the tip but for those of us with the new H3-3D gimbal reverting back to 1.08 kills it. Everyone appreciates the suggestions to rotate the compass and to revert to earlier firmware but lets be clear...

The purpose of this thread is to get DJI to fix the problem at it's root cause. It is intended to confirm the issue and narrow down the cause. These "slap some tape on it" fixes don't really solve the issuete][/quote]


Exactly.. As I said: I'd buy the vision + in an instant.. But not until they get this problem fixed.. I bought my vision 2 the end of January, and have been fighting with problems with it ever since.. From major toilet bowl effect, to huge left drift.. It's been awful.. Wish DJI would come clean on all this, and fix the **** problem.. Surely, it can't be that difficult.. Other than that.. Great copter!!

 

[JPSLV]

Is the problem fixed???

 

[ianwood]

Is the problem fixed???

Not yet but DJI is looking at it as are we.

As a general update for everyone, a small group of us are sending some test data to DJI for them to analyze and see if they can pinpoint the problem. I've been traveling so will be finally getting mine done tomorrow.

In the meantime, I have been busily dissecting the contents of the CAN bus communications to pull raw telemetry off of the Phantom. I know just enough C/C++ to be dangerous (mostly to myself) but I am no developer. With some significant help from others who've done much to figure out the binary structure of the messages for other needs, I've made some good progress.

So far, I've been able to build a table of raw compass XYZ values, raw gyro XYZ, raw accelerometer XYZ, GPS lat/lon, GPS N/E velocity, GPS altitude and computed barometric altitude. From that I can figure out compass heading and compare it to a forward adjusted GPS course.

My intention is to prove the Phantom compass heading is incorrect and that correlates to hook made by the Phantom. There's more CAN structures to process but when it's done, they'll be useful for much more than demonstrating a problem with the compass math!

 

[landonkk]

Is the problem fixed???

Not yet but DJI is looking at it as are we.

As a general update for everyone, a small group of us are sending some test data to DJI for them to analyze and see if they can pinpoint the problem. I've been traveling so will be finally getting mine done tomorrow.

In the meantime, I have been busily dissecting the contents of the CAN bus communications to pull raw telemetry off of the Phantom. I know just enough C/C++ to be dangerous (mostly to myself) but I am no developer. With some significant help from others who've done much to figure out the binary structure of the messages for other needs, I've made some good progress.

So far, I've been able to build a table of raw compass XYZ values, raw gyro XYZ, raw accelerometer XYZ, GPS lat/lon, GPS N/E velocity, GPS altitude and computed barometric altitude. From that I can figure out compass heading and compare it to a forward adjusted GPS course.

My intention is to prove the Phantom compass heading is incorrect and that correlates to hook made by the Phantom. There's more CAN structures to process but when it's done, they'll be useful for much more than demonstrating a problem with the compass math!

This is encouraging ianwood. I dream of the day my Phantom's fly straight. After flying so long with right stick adjustment I think I might fly crooked the other way because of crooked thumb syndrome!

 

[tom3holer]

I have mentioned this before but for some update.
I rotated the compass about 20-25 deg and it eliminated the bad toiletbowl effect. I have had several flights now and don not see it changing during a flight. It did eliminate the jhooking also.

Tom

yeah. and when the software self corrects you are back to the same mess.

I did not see any correction during several flights.

Is it possible that it does not self correct on each flight?

Tom

 

[Lenny7118]

I have no faith in DJI , their fix will be coming out with a new Phantom 2.1.1
DJI should not have consumers doing their work for them as gathering data etc... It's DJI product and they should fix it. I've read from a post that someone called the DJI office in CA in the United States and they had no knowledge of the problem. That's not encouraging to hear.

 

[ianwood]

I have no faith in DJI , their fix will be coming out with a new Phantom 2.1.1
DJI should not have consumers doing their work for them as gathering data etc... It's DJI product and they should fix it. I've read from a post that someone called the DJI office in CA in the United States and they had no knowledge of the problem. That's not encouraging to hear.

Don't lose hope. The reason I am doing the extra work on it is an insurance policy in case they balk. Hopefully they find what they need to fix it from the data we've sent them. Mine is just in case. As for their support, that's a different matter. Abandon all hope ye who call DJI support!

 

[tom3holer]

I hear from two DJI reps that frequent another forum that they are indeed aware of the problem and are working on it.

Tom

 

[onfourblades]

I hear from two DJI reps that frequent another forum that they are indeed aware of the problem and are working on it.

Tom

which fourms?

 

[Deltafox]

" Auto calibration. You do nothing, and the code figures out all the offsets and declination by comparing the compass readings with the GPS and IMU readings over time in flight. Pro: no user effort. Con: it takes a few minutes of flying to get right, so the compass is inaccurate at first launch. "

Explains why the problem lessens after a few minutes of flying.

 

[Bigbells]

Thanks for the tip but for those of us with the new H3-3D gimbal reverting back to 1.08 kills it. Everyone appreciates the suggestions to rotate the compass and to revert to earlier firmware but lets be clear...

The purpose of this thread is to get DJI to fix the problem at it's root cause. It is intended to confirm the issue and narrow down the cause. These "slap some tape on it" fixes don't really solve the issue.

Yes, you're right. The goal is to get DJI to correct the issues. I also wasn't aware that 1.08 didn't work with the H3-3D gimbal.

For my Vision+ and my Vision, reverting to 1.08 is like restoring a malfunctioning computer to its last known good configuration. It sure is nice to again be able to yaw and have the Phantom hold position instead of wandering all over the place.

 

[ianwood]

I hear from two DJI reps that frequent another forum that they are indeed aware of the problem and are working on it.

Tom

which fourms?

rcgroups.com. The same guys are on here as well just not as often.

 

[ianwood]

" Auto calibration. You do nothing, and the code figures out all the offsets and declination by comparing the compass readings with the GPS and IMU readings over time in flight. Pro: no user effort. Con: it takes a few minutes of flying to get right, so the compass is inaccurate at first launch. "

Explains why the problem lessens after a few minutes of flying.

Where did you read that? It should and does calibrate throughout the flight but is supposed to do auto-declination which means you start out with an accurate true north heading based on your GPS position.

 

[Deltafox]

Where did you read that? It should and does calibrate throughout the flight but is supposed to do auto-declination which means you start out with an accurate true north heading based on your GPS position.

DIYDRONES forum .... interesting to read all the comments.
http://diydrones.com/profiles/blogs/com ... le-does-it

 

[landonkk]

I am going to self-test the manual compass correction. I am planning on it overcorrecting as the flight goes on because everything I know about this issue leads me to believe that result, but i would like to see it for myself. Since I have three P2's it is easy to 'fiddle' with my P2Vision and document the results. I shall report back my findings.....

 

[d4ddyo]

Why is everyone trying to re-inventing the wheel? We been there done that. Read up. It is not a hardware issue as much a software issue because the Phantom autocorrects.

 

[landonkk]

Why is everyone trying to re-inventing the wheel? We been there done that. Read up. It is not a hardware issue as much a software issue because the Phantom autocorrects.

I don't think anyone is trying to do anything other than try to help pinpoint the problem. Yes, it is a software issue, but if we can give the engineers more info it can only help.

That being said I am going to throw a wrench in things and report that with a manual compass relocation clockwise (to offset a left j-hook) my P2V did fly perfectly straight with no hook and I flew it to 40% battery and did not notice an overcorrection.

 

[d4ddyo]

Why is everyone trying to re-inventing the wheel? We been there done that. Read up. It is not a hardware issue as much a software issue because the Phantom autocorrects.

I don't think anyone is trying to do anything other than try to help pinpoint the problem. Yes, it is a software issue, but if we can give the engineers more info it can only help.

That being said I am going to throw a wrench in things and report that with a manual compass relocation clockwise (to offset a left j-hook) my P2V did fly perfectly straight with no hook and I flew it to 40% battery and did not notice an overcorrection.

It's already been done. Takes off no problem but auto corrects and j hooks opposite direction.

 

[landonkk]

Why is everyone trying to re-inventing the wheel? We been there done that. Read up. It is not a hardware issue as much a software issue because the Phantom autocorrects.

I don't think anyone is trying to do anything other than try to help pinpoint the problem. Yes, it is a software issue, but if we can give the engineers more info it can only help.

That being said I am going to throw a wrench in things and report that with a manual compass relocation clockwise (to offset a left j-hook) my P2V did fly perfectly straight with no hook and I flew it to 40% battery and did not notice an overcorrection.

It's already been done. Takes off no problem but auto corrects and j hooks opposite direction.

Ok.. so what part of my post that said that is exactly what I expected and wanted to see it for myself confused you?