Had a crash and now when connected to mission planner yaw is continuously changing. It starts at 360 and counts down. I created a short video showing this in mission planner. Flying is of course impossible, the copter just starts to yaw out of control.
Any suggestions to fully diagnose this or correct it? I suspect the gyro is messed up. I read about reflowing but would appreciate more details on how to do this. I do have a heat gun.
I am having a similar issue, I noticed that if I reload the code, setup the radio and everything without turning the APM off, it works fine, i can even fly. But as soon as I restart the APM the yaw gyro isnt taken into account it seems. The IMU test works great, but the APM doesent incorporate the gyro data into flight.
I am working on it now, I'll let you know what I find.
Sounds like you don't have a magnetometer, it's not enabled or perhaps it's broken. Gyros drift so you normally need the magnetometer to counteract that.
I couldn't actually see the heading in the video :-(. Isn't it right at the top but just out of view?
Gyro's drift a bit normally, but not so much that its impossible to fly. In my case, I have the mag disabled and still get the drift in planner and in flight. After testing both the gyro's and mag sensor both work in my case.
duane, try running the "tests" in the cli menu. Try testing the IMU and compass, both values should change. If the compass stays at 365, or any other value, try disabling it and see if that helps.
I've had similar problems ever since we updated to 2.055, with the problems persisting in 2.1, and seen other users with similar yaw problems (an maybe related roll or pitch problems).
I was able to get the problem to go away in 2.055 by resetting and erasing eeprom, reinstalling firmware, and re-doing setup. However, I'm having problems getting the same to work in 2.1. Based on my observations, it seems to me that the problem has to do with the compass offsets. It seems that bogus numbers occasionally get created and then a permanent yaw develops, which can only be solved by erasing the offsets and starting over. I don't know if Jason changed the code related to how the offsets are created in these newer firmware versions, but that might be a place to check.
I could be way off base, but I've seen enough similar problems from others to think that it might be a bug in the code.
I have met the similar problem, I think if it oscillates with in 1 deg should be OK to fly, but the performance is not quite well.
By the way, my board was purchased quite long time ago, I don't know whether the latest code can work well on it....
I suggest u to go CLI test to test the compass and imu, if that is OK, then try to use some old version code...
I am not quite sure whether PID settings will help..
I've checked the video again and the yaw is slowly drifting by about 1 degree every few seconds. This is perfectly normal and consistent with what I see on my perfectly flyable quad when I disable the mag.
Next time you see this, please ping me and I can send you a tiny sketch to pull out the mag offsets and we confirm that's the issue.
But if you disable mag and then spin the copter by hand, does the hud recognize this movement?
In my experiences, the yaw slowly drifts, but the mag does seem to be working in that the hud does display the correct orientation (at least approximately), it just doesn't maintain it.
My problem is more than just what is shown in Mission Planner. When I attempt to fly, the copter yaws immediately in a fast spin. I had been flying for a few months without a GPS and without a mag with no issues and I never noticed the forever yawing in MP. So I assumed what I saw in MP and the spinning during flight was related.
I tried with 2.0.48 and saw the same yaw in MP. Though I haven't tried to fly with this build.
The drift is about 1 degree a second but always in the same direction. I guess if left long enough it would go all the way from 360 to 0. Not really a drift... just a countdown.
I will add a magnetometer this weekend... but I doubt that will change anything. I really think something is busted from the crash. I actually have tried a second oilpan yet still have the same results.
As Todd mentioned... how do I do this: "erasing the offsets and starting over".
Wait! You don't have a magnetometer? That creates a whole different situation. If my magnetometer is disabled, I get constant yaw drift no matter what I do. I'm surprised that you had the absence of yaw drift in the past.
In the absence of a magnetometer, the first thing I would check is that all of your props are perfect perpendicular. If the problem started after a crash, a change in propeller angle could explain the drift (especially in the absence of a mag).
In answer to your question, in order erase the offsets, go into CLI, 'setup', 'erase', and 'reset'. Then you'll have to reinstall firmware and re-setup your copter.
You dont need a mag sensor to fly the arducopter... Its to compensate yaw drift in the long run. Duane's problem is his copter is unfly-able.
Do my a favor duane, in the cli, can you run an IMU test and verify that your yaw gyro is outputting good data?
Propellers offset or something wouldnt cause yaw drift, maybe a little resistance one way or the other. I, like many others have flown mag-less for months without issue until this version. Now even after downgrading and erasing my CLI I cant get it back to how it was. I would hate to have to re order an APM while I wait for my APM 2.0 but hopefully I can cancel my order and just get another oilpan/2560.
When I run IMU from the CLI test (using 2.0.48 currently) I see:
Gyro offsets: -15.00, -34.00, 28.00
Accel offsets: 2.00, -15.00, 16.20
Hit enter to exit
-46, 47, 35810
and then the 35810 number just keeps slowly increasing.
Just a guess but I expect your problem is not a change in the code but rather that the mag offsets were erased after you upgraded the code. This is normal but they should be recalculated and resaved if you fly (or wave your quad around) for a couple of minutes and then land, ensure you're in stable mode and put your throttle to zero.