Hi,
I was reading Jason Short's post on gyro drift / soldering the IMU pads, and in his last post he says that a lot of the gyro drift issues should be resolved in the latest firmware.
I just updated the firmware to 2.0.35 and was testing drift due to vibration by monitoring where it thought it was pointing via the latest mission planner software (which is what I used to update the firmware). I also cleared the eeprom and ran the AC initial setup. Data is streamed via an xbee to the laptop.
With the props off and sitting level on the ground, as I start to increase the throttle the directional feedback I was getting on screen seemed to indicate that the quad was rotating on the pitch, roll and yaw reasonably rapidly. As soon as I lifted the platform which the APM/IMU is attached to from the posts it was resting on so that it was free from the center hub, the readings were showing that it was stable and level with no yaw drift. As soon as I touched the platform back onto the center hub while the motors were running the random directional drift started up again. The APM is stuck to the platform with two sided foam tape and the platform attaches to the hub via the rubber grommets you get with large servo's. There is only soft rubber between the platform and the hub connection points. Holding the hub with my hand the vibration from the motors is really minimal so it is quite surprising to see such a large amount of fast drift from this little amount of vibration. I also have tag ties wrapped around the motors to balance them (this also reduces vibration from unbalanced motors).
There is no way in the world I would try to fly it with drift like that. You can hear the motors ramping up as the drift takes it all over the place.
I will go and solder the three pads now to see if it eliminates the drift, however it does look like 2.0.35 beta does not settle the drift down any more than the 2.0.33 I had on the board previously. I will also get some 10mm soft foam and place that below the APM board to see if that also makes a difference as well.
Replies
On my quad the only somewhat rigid contact between the APM and the rest are the cables; it is only wrapped in a layer of foam which is then tied with a cable tie. This sandwhich then fits inside a small cd spindle case. I'm waiting for a replacement compass so I've been testing Simple mode without it, and I'm only seeing rather small drift in the heading. I have soldered the gyro filter pads. I haven't measured the drift though and I wouldn't want to fly far with the possibility of the yaw changing, but a working compass should solve that.
My only problem is that I have to field calibrate the level because the APM only stays put when ready for flight; when I interface via USB I have to open the case and nothing holds it in place. Any ideas on how to construct a hub that can hold a foam-wrapped APM in the same attitude every time?
Kári Davíðsson and Jose Julio are doing some great analysis on the issues which seem to point at high frequency aliasing affecting the accelerometers. Even though this is a specific HW issue with the resonant freq of the accels, I'd expect at least one more SW update to attempt address the issue, but that may be a ways away.
The IMU seems to be affected at only certain motor frequencies which don't include hover. You can protect yourself easily by soldering the filter pads, mounting the IMU on a foam rubber cushion, and not flying with bent shafts or out of balance props. I, however fly unbalanced props all the time with no issue at all. This is with a board that drifted up to 15° with vibration before the SW and HW fixes. For me, these fixes have made the issue go away in practice.
The SW fixes included upping the DCM gains recover from drift faster and narrowing the limits of the acceptable Accelerometer values.