Tuesday, November 4, 2014

Troubleshooting Wobble of Death (WoD)

With the help of the multi-rotor pilots in the RCGroups forums, it appears my WoD might be related to excessive airframe vibrations affecting the APM FlightController. While not much, as I've been adding extra weight to Nova (Camera Mount, Mobius Camera, Telemetry Radio, and finally Prop Guards) it seems to be getting worse and more frequent during flights. While I originally thought the WoD spells were a new development, as I review my flight-log-posts ... it has been occurring as far back as Flight #4.

Here is an IMU graph from Flight #7.


According to: http://copter.ardupilot.com/wiki/ac_measuringvibration/

The X-Y (red & green) Acceptable Range is -3~+3. The Z (blue) Acceptable Range is -15 ~ -5. Even ignoring what appear to be "landing spikes" ... all vibrations are beyond acceptable levels.

The following items will be checked:
  1. Flight Controller mounting location
  2. Flight Controller mounting material and isolation from airframe
  3. Propeller Vibration and Balancing
1. PASSED. Currently, the Nova's FC is mounted Aft of CenterOfGravity or CenterOfFrame by about 10mm, and Port about 20mm (away from CenterLine). The FC still has the indicated side Forward. That side is parallel to the front side of quad, and is perpendicular in the aircraft. This was done in the beginning to facilitate mounting everything (FC, RX, and GPS) to PowerBoard without them touching each other, or any raised components on PowerBoard. While nothing is declared in official docs, according to some knowledgeable pilots in APM forum, there is no reason to re-mount FC in exact CG middle of quadcopter. After normal calibrations (including Accelerometer/Gyro) due to quadcopter's size and speed, there should not be a performance or dependability penalty for this slightly off-set mounting location.

Edit 2015-01: This turned out to be a FAIL. While it doesn't show as good on above graph (as others on this blog) ... X and Y should originate at 0. I think this can only happen when FC's gyro/accelerometer chip is in true middle of quad.

2a. PASSED - FAILED. Mounting needs to be around 10mm thick so that the side USB plug clears the RX, yet FC does not touch the top of body shell. As I examine the current layers of thin white foam tape, it seems very rigid. I have replaced it with a more appropriate material. I used two 1-inch sections (one on each end) of thick/plush (5mm) foam gyro mounting tape, two layers for 10mm thick. The FC mount is now plush and flexible (but still secure) and should be much better at isolating vibrations from FC and it's on-board instrumentation.



2b. PASSED - FAILED. Since the FC boards are not actually screwed down inside plastic case, there is some movement. I'm not sure, but I think the PCBs (and attached sensors) might have been vibrating inside case. Added small pieces of gyro foam-tape inside case (on sides ... not on top where barometer-foam piece is). Seems to work fine.

2c. PASSED. Inside the FC, (with it's two-PCB construction) I noticed the grounding metal case of the Micro-USB plug (on upper-PCB) is dangerously close to pins hanging-down from main power-header (from lower-PCB). Added a small piece of electrical-tape (3 layers thick) to prevent a short circuit or arcing.



3. PASSED. Balanced stock-white CX-20 8045 props with Dubro Prop Balancer. They needed little if any balancing so I really doubt they were causing any significant vibrations.

Other things I have done to Nova in this maintenance cycle:
  • Did another Accelerometer calibration since FC was moved. Hold quad very still and level/perpendicular at each stage.
  • Motor Test via MP/Terminal. They all run and sound ok.
  • Upgraded MissionPlanner to v1.3.10 and then to v1.3.11 a few days later
  • Set THR_MID parameter to 550.

4 comments:

  1. Hi,
    my new Nova (still on stock firmware) tell me
    Test: IMU Mismatch = UNKNOWN - No IMU log data
    when i try to analyze log (or empty chart in manual review).
    I've to enable something to collect IMU data?

    Thank you for all the info you share! :)

    ReplyDelete
    Replies
    1. Try LOG_BITMASK = 958
      This is Default + IMU (for vibration logging)
      Gives you:
      ATTITUDE_MED GPS PM CTUN NTUN IMU CMD CURRENT

      Search my blog for imu or vibrations. Also see
      http://copter.ardupilot.com/wiki/common-diagnosing-problems-using-logs/

      Delete
  2. Not sure if I understand your article correctly. So FC should or should not be in center of gravity of the quad? Is that the main reason for quad wobble?

    ReplyDelete
  3. I think so. I only know what works for me (centered, perpendicular, and anti-vibration Zeal). http://quadcopter-robotics.blogspot.com/2015/01/nova-rebuild-flight-test-winter-2015.html

    ReplyDelete

The stupid spammers have now forced me to approve each Comment before it appears (but I am usually pretty quick about it).