I am trying to tune the ArduPilot for Bevrc X-5 plane. At the moment I am playing with roll P gain and trying to achieve acceptable stability in FBW mode. Also, I am trying to overcome the plane peculiarities: he has tendency to roll in positive direction (left wing down) and because of this, rates of left and right turns are not equal.
In order to get an idea what is going on during FBW flights, I converted flight logs into charts which display desired (navigation) roll, and actual roll detected by IMU.
I started from P = 0.003. It seemed to be too much for my plane, because oscillation amplitude is pretty high. The plane even was unable to turn right due to the oscillation. Attached LOG00220 demonstrates how it looked.
Then I tried to reduce P gain down to 0.002. This gain value worked much better, oscillation reduced significantly, but sometimes the plane could not keep navigation roll during left turn and tried to fall into left spiral. LOG00221 shows this test.
Then I reduced P gain to 0.001 (LOG00222). The tendency to fall into left spiral becomes even more visible, although oscillation almost disappeared.
Now I am going to stay with P=0.001, and will try increase I gain in order to achieve better stability. I appreciate any hint how to tune P and I gains correctly.
I'm interested in how your tuning goes, but your logs are blank for me. Maybe you could post them as a screenshot or something?
Oh, sorry for the mess, I have converted logs to pdf.
I keep IMU leveled during initialization, and oscillation in case of LOG220 were clearly visible, so I think that roll readings provided by th IMU are close to truth.
I did a test with increased I gain, and it looks quite positive: I keep P gain the same as for LOG00222 (0.0015) but set I gain to 0.001. LOG00223 shows quite good match of the roll value with navigation roll for first 10 minutes of the flight.
However, after that time, the tendency to fall into the left spiral have returned. I guess it can be explained by a decrease of air speed due to battery voltage drop (I keep throttle level the same during all the flight).
To check whether this guess is true, I will try to draw the log of the air spped.
how did it go with your project. it has been a while.
thanks for the interest. I had achieved acceptable level of roll and pitch authority in FBW mode, and RTL mode worked acceptable too. I even created a some sort of diy OSD to visualize the AP output (startd text GCS) during fpv flights, see A screens shot:
However, the end of the story is a bit sad: I lost the plane and all equipment due to UBEC failure: I have totally lost control over the plane and OSD frost (due to absence of AP output) about 3 km avay from the start point. Camera and VTX keep working, because they were powered directly from the battery, and I had a chance to see the final spiral of my X-5. Unfortunately, I failed to found it.
A good part of the story is that I made a step forward after it: now I am playing with X-8 and APM 2.5 :)
aslong as the imu is facing forward any angle it is calibrated will be its level point and change from there after and reflect the movements the same
you can verify this by attaching it directly to the ftdl port and by using arduimutest.exe
the plane will always be level after inital calibration .
however the sides and front will be swapped if not facing forward atleast