Beta Testers requested to fix Compass Issue

[d4ddyo]

Crap I'm here in North Jersey and I am noticing this effect.

 

[landonkk]

We did it this afternoon, turned the compass for about 17' CW, it worked like a charm! P2 now hovers very steady with a small or zero drifting, moves forward and backward on a pretty straight line. Will do some fine tuning tomorrow.

[/quote]

Can you please advise on effect towards the end of your flight? Others have tried this fix but noticed that as the Naza adjusts (as it seems to do so and correct the j-hook issue) it ends up "over-correcting" and the hook shows up in the opposite direction.

 

[d4ddyo]

has beta started?

 

[onfourblades]

yes in testing phase now

 

[SeanAxiom]

We did it this afternoon, turned the compass for about 17' CW, it worked like a charm! P2 now hovers very steady with a small or zero drifting, moves forward and backward on a pretty straight line. Will do some fine tuning tomorrow.

Can you please advise on effect towards the end of your flight? Others have tried this fix but noticed that as the Naza adjusts (as it seems to do so and correct the j-hook issue) it ends up "over-correcting" and the hook shows up in the opposite direction.[/quote]


We've had 4 flights so far after rotating the compass, still doing some fine tuning. I've seen others' concern about the j-hooking at the end of the path but in my case it wasn't noticeable at all. I was alone today and couldn't operate two cams, just had the gopro on the bird. it was a pretty windy day but the flight was pretty stable, the bird could stay solid on its position whenever I released the sticks. This is a clip from today:

http://youtu.be/VdUxjakaZsU

I'll have another flight tomorrow if whether permits, will have a ground camera to capture the flight.

 

[ianwood]

Beta testers,

It appears the tests we've been asked to do might not reveal the full extent of the issue. We may need to do some more tests to demonstrate the full extent of the issue. I have contacted Terry and Zach about it and waiting to hear back. In the meantime, I am going to suggest the following test to run. The .DAT file will be about 45Mb so put it on Dropbox or Google Drive and send me a link. Note the file will reveal your flying location. I will try not to stalk you!

1. Use firmware 2.00 or 3.00. Make sure it works before running this test.
2. Perform IMU advanced calibration, make sure compass mod value is OK.
3. Identify a flying area free of granite or other potential magnetic material.
4. Perform compass calibration.
5. Start up and hover for 15 or more seconds.
6. Fly forward 20-30ft at a reasonable pace and then release. Allow Phantom to hook, drift and/or TBE. Try to let it come to a stop on its own if possible.
7. Fly backward the same distance and again allow the Phantom to settle on it's own if possible.
8. Fly the Phantom as you would normally for 4 minutes.
9. Repeat steps 6 and 7.
10. Fly the Phantom as you would normally for another 4 minutes.
11. Repeat steps 6 and 7.
12. Land, shut down and then download the data from the Mark II OSD.

Thanks,

Ian

 

[xtonex]

if they need more people for this and are still willing to send out a mark II let me know... im having some real odd issues my last few flights. tbe jhooks and now very unstable flying with or without anything attached to the phantom

 

[Xrover]

Is everyone that is testing calibrating their compass in the same manner?
ie. Rotating it on it's axis rather then holding out in front of you and rotating your body.

It appears to me that depending on how I do a compass calibration and the speed at which I do it results in slightly different behavior on the flights. It's hard to tell which method and speed of the calibration procedure yields the best performance on the first flight. I think the P2 behaves differently on the first flight as the IMU might still be warming up. I have 5 batteries and have noticed that if I fly with a short amount of time in between flights, the P2 fly's a little better each time. Not substantially better but enough that i can feel a difference.

 

[ianwood]

Is everyone that is testing calibrating their compass in the same manner?
ie. Rotating it on it's axis rather then holding out in front of you and rotating your body.

It appears to me that depending on how I do a compass calibration and the speed at which I do it results in slightly different behavior on the flights. It's hard to tell which method and speed of the calibration procedure yields the best performance on the first flight. I think the P2 behaves differently on the first flight as the IMU might still be warming up. I have 5 batteries and have noticed that if I fly with a short amount of time in between flights, the P2 fly's a little better each time. Not substantially better but enough that i can feel a difference.

It's a good question. I would suggest not doing the dance and instead turning the Phantom in place as is depicted in DJI videos.

As for IMUs warming up, etc., it's quite possible that a number of things settle in on the first flight including the pilot! It could be like driving a car with the engine cold. However, I don't think there is any persistent data (adaptions or what have you) between flights beyond the basic configuration data. I have no data to back that up. Yet.

 

[BenDronePilot]

With a regular DJI quad you would move the GPS puck to the proper magnetic declination. I wonder if anyone has tried moving the Phantoms puck which is stuck to the upper body shell to compensate.

 

[droneranger]

With a regular DJI quad you would move the GPS puck to the proper magnetic declination. I wonder if anyone has tried moving the Phantoms puck which is stuck to the upper body shell to compensate.

Unlike the Wookong/Naza/DJI puck which houses both the GPS receiver and the electronic compass, the Phantom separates these 2 functions. The GPS is indeed adhesively attached to the upper body shell, but the electronic compass is mounted to one of the landing legs. Since this is a declination issue, moving the GPS inside of the shell would have no effect since it provides position information rather than heading data.

 

[ianwood]

Terry at DJI emailed me back and confirmed he would like us to do the tests I outlined so they can see the hook in forward and backward flight. This important as it demonstrates the issue much better than the previous test.

1. Use firmware 2.00 or 3.00. Make sure it works before running this test.
2. Perform IMU advanced calibration, make sure compass mod value is OK.
3. Identify a flying area free of granite or other potential magnetic material.
4. Perform compass calibration.
5. Start up and hover for 15 or more seconds.
6. Fly forward 20-30ft at a reasonable pace and then release. Allow Phantom to hook, drift and/or TBE. Try to let it come to a stop on its own if possible.
7. Fly backward the same distance and again allow the Phantom to settle on it's own if possible.
8. Fly the Phantom as you would normally for 4 minutes.
9. Repeat steps 6 and 7.
10. Fly the Phantom as you would normally for another 4 minutes.
11. Repeat steps 6 and 7.
12. Land, shut down and then download the data from the Mark II OSD.

Please try to maintain the same heading. Find a distant point and aim at that each time from the same spot for the forward / backward part. It’ll make highlighting the difference between course flown vs. compass heading easier. Also, don’t use any aileron, rudder or throttle when doing the forward / backward part of the test, just elevator. This is important as the data will show zero actual stick input other than elevator and the Naza will be adding aileron due to the compass issue. If you need to recover from TBE or hooking into a tree obviously do what you need to do.

The .DAT file will be about 45Mb so put it on Dropbox or Google Drive and send a link to me and Terry and Zach. Note the file will reveal your flying location. I will try not to stalk you!

EDIT: Zach from DJI just emailed as well and said they want a test in ATTI mode as well from the same location. Take off in ATTI, 100m forward, then back 100m, hover for a minute and then left 20 and then right 20.

 

[Sanababit]

Its great you are doing this guys, hopefully you figure things out for us

Sana

 

[Xrover]

I've noticed that ever since i upgrade from firmware 2.0 to 3.0, i now have the TBE! I didi not have that prob. With 2.0!
Something must have been effected when DJI made the change and i think the testers should all be runnig 3.0 to keep the testing exactly the same. My P2 does behave in a different manner since upgrading to 3.0. This was just a observation i have made and think it would help maintaing uniformity in the test procedures.

Just my thoughts!

 

[John Shaw]

I have upgraded to 3.0 and Android .40 and the TBE, JHook, and Course Lock offset have remained the same. I believe for the initial flights I was focused on more basic aspects of flight but after a few flights my son and I commented to each other about the course lock misalignment - but didn't think much about it. After reading ianwoods initial videos and analysis it rang the bell that that was exactly what we were seeing. As with all things you notice it more when your looking for it.

I do think the Course Lock Offset is the most dramatic effect and may lead to easier analysis as it does not appear to be dynamic. It seems once the Course is set it doesn't change like the JHook does. I assume if you reset the course periodically you would see the offset decrease each time it was reset. I have been planning to test this but our weather has not been very cooperative.

My declination is +17

 

[ianwood]

I do think the Course Lock Offset is the most dramatic effect and may lead to easier analysis as it does not appear to be dynamic. It seems once the Course is set it doesn't change like the JHook does. I assume if you reset the course periodically you would see the offset decrease each time it was reset. I have been planning to test this but our weather has not been very cooperative.

My declination is +17

Yes, this is because the course lock is set at the very beginning of the flight. I am searching pretty hard in the data for where it is stored. If I can find the number, all I need to do is compare it to the raw compass values form the same time and adjust them for declination and that becomes our smoking gun!

 

[cruiser607]

DJI was obviously able to reproduce the issue with my data from NZ.

They asked for a retest in ATTI mode. Hopefully I can deliver this data this week.

 

[onfourblades]

DJI was obviously able to reproduce the issue with my data from NZ.

They asked for a retest in ATTI mode. Hopefully I can deliver this data this week.

The asked for the same "atti test" from me as well.
I tested today and sent them the data.
New York here

 

[ianwood]

DJI was obviously able to reproduce the issue with my data from NZ.

They asked for a retest in ATTI mode. Hopefully I can deliver this data this week.

Please do the test I outlined above as well. They want both tests.

 

[cruiser607]

Please do the test I outlined above as well. They want both tests.

They already have these results.