3DM-G Analysis

A work in progress... jwg

I wanted to throw together a quick page to detail my experiments with the 3DMG. These experiments stem out of the obvious error in the Euler angle data coming off of the 3DMG that we experienced with yesterdays test flights. For now I will ignore Accels and AngRates and focus just on the Euler angles. I am using my jig which suspends the MAV in a level postion for these experiments. I am trying to identify possible interferences with the sensor due to the environment within the airframe. The MAV is packed with the electronics suite (just as it would be during a real test flight). Mujahid and I have placed the 3DMG at the rear of the aircraft (to avoid EMFs) and I am using this setup for these tests as well.

The test procedure:
1. Setup aircraft in jig fully packed and balanced.
2. Orient airframe until 3DMG outputs gave ~0 degrees for Roll, Pitch, and Yaw.
3. Start Datalog.
(Each one of the following tests were done, and then the plane was allowed to sit with no commanded inputs)
4. Motor 50%
5. Motor 100%
6. Pull Up, Dive, Left, Right (No Motor)
7. Pull Up, Dive, Left, Right (50% Motor)
8. Pull Up, Dive, Left, Right (100% Motor)
(the stick motions may have been performed more than once, i.e. PU, D, L, R x2... it doesn't really matter)
9. On one or two tests I believe at the end of the last experiment I bled the motor off slowly...

Standard In-plane Benchtest

This is what I consider the baseline data, and very similar to what we were experiencing in the test flight data. You can clearly identify my above sequence of events, and you can clearly identify the noise (+/-5 degrees in the Euler data). At times in the flight data these errors were much higher, but this is explained below.
After looking at this initial benchmark I believed that I had clear evidence that the motor was responcible for the errors in the Euler data. I wanted to do another check to be sure, a mechanical check. The errors in this data could be directly correlated with EMFs or electronic noise, but it could also be correlated with mechanical noise (motion). Keep in mind that the jig I am using keeps the planes motion very restricted, so any disturbances in the motion of the plane will be very small.

No Motor (Tapping Vibration Test)

This next test is basically me leaving the plane in the jig and doing a slight amount of finger tapping or drumming on different parts of the plane to simulate small disturbances. In this data there are 5 clearly visible sets of disturbances. The first one is very light tapping (or drumming) on the side of the nose. The next 3 sets correspond to my attempted uniform tapping disturbances on the roll, pitch, and yaw. The last set is unique because I wanted to look at what kind of disturance a piece of metal in close proximity to the sensor does. The last set is me waving a small screwdriver about an inch over the sensor.
This data is significant because it shows how much the Euler data is affected by motion, and how it is possible that EMF is not causing this interferance at all. I was going to write a small filter, but before doing that I decided to try the gryo-stabilized values that the 3DMG can provide.

Standard In-plane Benchtest with Gyro Stabilization

Here is the exact same benchtest as in Figure 1, the only difference here is that we are getting the gyro-stabilized data from the 3dmg, instead of the instantaneous packet. We are still able to get this data at 30Hz on the ground. One thing to note is towards the end of this data after the motor is turned off, I rotate the jig ~180 degrees and back. There still seems to be a little inconsistancy in the yaw data, but at this point it could be due to the fact that we are sitting inside a building.
It would seem that the issues we have been having with the Euler angles disappear with the gyro-stabilized values. I am very interested in seeing the data from a test flight (which we can hopefully get tomorrow). I will put it up here once we fly.

Ground Test with Gyro Stabilization

Test Procedure:
1. 90 deg Turn
2. 20 deg continued back to original position.
3. Lift plane about 2 feet.
4. Complete Right Roll back to stable.
5. Complete Left Roll back to stable.
6. Complete Pitch Roll forward.
7. Complete Pitch Roll forward.
8. Complete yaw roll left.
9. Place back on ground.
10. Wave keys over 3DMG, and to the sides.
11. Tapping on the nose, pitch noise.
12. Tapping on the tail, yaw noise.
13. Tapping on the wings, roll noise.
14. Lift Plane again, about 3 feet.
15. Complete yaw rotation to left.
16. Complete roll right.
17. Complete pitch forward.
18. Forward accel.
19. Back accel.
20. Right accel.
21. Left accel.
22. Lift plane fast.
23. Push plane towards ground.
24. Lift plane about 3 feet.
25. Quick 90 left bank, and back.
26. Quick 90 right bank, and back.
27. Quick 90 pitch up, and back.
28. Quick 90 pitch down, and back.
29. Quick 90 yaw right, and back.
30. Quick 90 yaw left, and back.
31. Quick 90 pitch up, and back down slowly until pitch -90, then back up.
32. Slow yaw 90 left, and back.
33. Slow yaw 90 right, and back.
34. Slow yaw 90 left, and back.
35. Slow yaw 90 right, and back.
36. Slow roll 90 left, and back.
37. Slow roll 90 right, and back.

Accels during Ground Test with GS

Roll Rates during Ground Test with GS

Euler Angles during Ground Test with GS

Last updated June, 2, 2003 by Jason W. Grzywna