Possible strategy for finding motor causing vibration

Discussion in 'Cinestar 8' started by Andy Johnson-Laird, Oct 7, 2012.

  1. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    After a recent "arboreal adventure" with my C8, I was back out testing all the systems today and noticed that the FPV camera on boom #1 was shaking like a leaf in a gale -- it seemed to be a vibration at, oh, around 15 - 20 Hz.

    There appear to be three possible sources for vibration:

    1. The propellers are not balanced. (But they are. I checked them after the flight.)
    2. There are one or unbalanced motors. (Bent out of shape by said arboreal adventure, perhaps?).
    3. The gimbal is oscillating (But it was powered off and the vibration was present with and without a CX760 which I would have changed the vibration by virtue of its mass.)

    Testing the prop. balance was easy enough using a prop balancing jig.

    The I started how I could test the motors and find out whether one or more of them was damaged. So I removed the props. and used MK Tool's motor test to spin up each motor in turn.

    With a motor running at "80" (in the MK Tool window), I tried using the iPhone app. "Vibration" -- but none of the motors seemed to generate enough vibration to really register a significant waveform on the app. Similarly, just placing my fingertips lightly on the boom with the spinning motor didn't reveal any motor that was different from the others.

    So I decided to use a small lavalier clip-on microphone, clipped to the same place on each motor mount to record the acoustic signature (a.k.a. sound of) the motor starting up and running so I could both look at the waveform and also get a spectrum analysis to see if any one of the motors was substantively different from all the others.

    Attached is a composite image of screen grabs for motors #1 and #3 -- it shows the spectrum analysis and the waveform. I would have uploaded the .WAV files, but, sadly they're all slightly larger than the 1MB limit for uploads.

    If you are interested in looking at the files I collected, you can download them from my public Dropbox folder. These include the waveform screen shot, the actual .WAV file so you can hear the motor, and the spectrum analysis screen shot. There's also a .sfk file that is created by SoundForge (the audio recording software I used). Ignore those.

    I did a couple of "takes" for Motors #1, #2, and #3, because I realized I wasn't recording the motor start up.

    What struck me was that Motor #3 seems to be very different. If you look at the image below, it's much louder than motor #1 and, if you listen to the .WAV file has a very raspy sound. You can see this in the spectrum around the 3 Khz to 5Khz. There are couple of spikes there. There's also a curious attenuation at 62 Hz.

    What I don't yet know is whether Motor #3 is bad and is causing the vibration. I'm going to have to replace it with a new one and see what changes.

    For those of you with the time and energy to download the files from my public Dropbox account, I'd welcome your thoughts on this technique and also the results -- is Motor #3 bad? Are there other bad motors?

    Thanks
    Andy
    Motor #1 and #3.png
     
  2. William Johnston

    Joined:
    Oct 2, 2012
    Messages:
    140
    Likes Received:
    5
    LOL I landed in a tree once. If you have altitude on when your ESC goes into BL limiting mode the copter will drift sideways even if you have position hold on.

    Anyway, neat idea. Maybe you have a bad bearing. Or some other noise in the background. What about sticking an accelerometer on the boom?

    If it was an unbalanced motor I would think that the vibration would be at the same frequency as the motor. For a motor test value of 80 I measured a speed of 52 Hz (3120 RPM).
     
  3. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    Actually, tree landings are preferable to "uncontrolled flight into terrain." The tree is a pretty good shock absorber/crash barrier. The downside is that one or more motors get shock-stopped when the props. break. Oh, and you need a ladder....

    There was no ambient noise of significance -- what there was was showing on the VU meter at -52 dB so not a factor in the recording.

    I'm suspecting bad bearing -- I've ordered up a one and a three access accelerometer (vernier.com -- neat equipment, not lab grade, but certainly good enough for this kind of work).

    Interesting that you measured 52 Hz for a motor spinning at 3120 rpm? That seems faster than the vibration I was seeing on the FPV camera, but then it's probably oscillating at its natural frequency. Quite a low harmonic, I presume? I'm also struggling to correlate that to the spectrum -- of course, it could be that your motor is spinning faster/slower than mine -- especially if the bearing is naffed up.

    In infer that you're at a .edu and happen to have access to a tachometer/strobe?

    Thanks again
    Andy.
     
  4. William Johnston

    Joined:
    Oct 2, 2012
    Messages:
    140
    Likes Received:
    5
    Yeah I did one of those too. Broke a boom and a leg. That's way more expensive than landing in a tree. The tree landing broke all the propellers. And luckly I have a nice ladder.

    I'm at an .edu, but I didn't use a strobe. I used a infrared diode and phototransistor. I fed the voltage drop across a sense resistor attached to the phototransitor to an oscilloscope and measured the frequency there. I get two pulses per rotation (two prop blades) so I divide the measured frequency by two to get the motor frequency.
     

    Attached Files:

  5. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    Nice setup -- a good way to get an rpm under the load of a prop. I note the tie-down going through the landing gear leg.....
    Thanks for doing the experiment.

    There's a fair amount of vibration that feeds back down the booms. I seem to remember seeing someone (was it on mikrokopter.de -- I don't remember) who was experimenting with vibration isolators as part of the motor mounts to try and kill the vibration at source rather than deal with the consequences of it.

    Thanks
    Andy.
     
  6. William Johnston

    Joined:
    Oct 2, 2012
    Messages:
    140
    Likes Received:
    5
    If your interested here is my data. I measured two different motors:

    Code:
    MTvalue  Hz*2    Hz*2    AvgHz
    3        24      24      12
    9        38      38      19
    22        56      56      28
    31        64      64      32
    40        72      72      36
    49        79      79      39.5
    61        88      88      44
    71        96      96      48
    80        103    104    51.75
    89        111    112    55.75
    101      120    121    60.25
    111      127    129    64
    120      134    135    67.25
    129      138    140    69.5
    141      146    147    73.25
    151      151    151    75.5
     
  7. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    Thanks, William.
    The AvgHz seems quite bounded regardless of the motor RPM. I presume that'll be because of the natural resonant frequencies of the stator, rotor, and boom?

    Thanks
    Andy.
     
  8. William Johnston

    Joined:
    Oct 2, 2012
    Messages:
    140
    Likes Received:
    5
    That's the motor speed. I'm not measuring any vibrations.
     
  9. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    Oh. Understood, I think.
    MT Value is the value shown on the Motor Test slider, then Hz*2 is the direct frequency (two blades), so the motor speed is then the right-hand column.

    Thanks
    Andy.
     
  10. William Johnston

    Joined:
    Oct 2, 2012
    Messages:
    140
    Likes Received:
    5
    Right. The right hand column is average motor speed in cycles per second (Hz). Multiply by 60 to get rotation per minute (RPM). The two middle colums are the values I measured on the oscilloscope which are double the motor speed.
     
  11. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    Thanks!
    Well, I've just swapped out motor #3 with a new one so I'll record the audio from it and then do a quick test flight. The good news is that the problem is repeatable. The bad news is I've got seven more motors that could be the culprit!

    Andy.
     
  12. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    Hmm. I've now swapped out motors #3 and #2 -- I'm still seeing the vibration through the FPV camera that is hard mounted on boom #1. To my eye it appears almost like a beat frequency -- with a period of about a second -- in between which there is little vibration, then it shakes, then it is steady and so on.

    Made me wonder what the MK FC board might do to avoid beat frequencies between the various motors.

    Anyone seen this kind of vibration before?

    Thanks
    Andy
     
  13. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    Hmm. I've now swapped out motors #3 and #2 -- I'm still seeing the vibration through the FPV camera that is hard mounted on boom #1. To my eye it appears almost like a beat frequency -- with a period of about a second -- in between which there is little vibration, then it shakes, then it is steady and so on.

    Made me wonder what the MK FC board might do to avoid beat frequencies between the various motors.

    Anyone seen this kind of vibration before?

    Thanks
    Andy
     
  14. William Johnston

    Joined:
    Oct 2, 2012
    Messages:
    140
    Likes Received:
    5
    I don't think there is anything that the flight controller could do. It has to change the speeds of the motors all the time to keep it flying right. I don't think it could also take over the task of precisely adjusting the motors.

    You might be able to have the ESC do it with disturbance accomodating control (DAC). The basic idea is for the motor controler to inject a signal into the motor control input that would generate in the motor an oscillation that exactly cancels out the undesired vibrations. Easier to say then to do though.
     
  15. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    I agree, William. I'm beginning to think that the source of the vibration may be some, as yet undiscovered, damage to the booms/hub plates caused by the "arboreal adventure." But, to preserve the present evidence, I'm going to video record the FPV camera footage and also the C8 on hover out of ground effect -- there is a long antenna that is part of the emergency locator beacon (Walston Retrieval Systems) that's part of the "find the C8 if it flies away). The Tx for that is in the bug between two booms and the antenna is suspended horizontally between two booms by a a piece of 40lb test monofilament fishing line. That antenna appears to be a sensitive vibration indicator and I can see it vibrating at what I suspect is the same rate as the FPV video.

    I'm also planning to duct tape (what else?) my iPhone to Boom #1 (where the FPV camera is) while it's running the Vibration app. and doing a short test hover out of ground effect. I want to see if I can capture the vibrations so I can see their periodicity.

    When I've done the above, I'm going to take apart the entire boom/hub arrangement and do a visual inspection to see if I can see any cracks or notice anything amiss. Then I'll reassemble and test.

    I'll post what results I can get just in case others have similar problems. At this stage, given the acoustic signatures of the other motors, and the lack of discernible runout on the propeller shafts (I've also checked the prop blade tips, aligning them with their nearest neighbor's blades), I cannot quite bring myself to replace all of the motors. I may end up there, of course -- that's usually the way it works.

    Thanks
    Andy.
     
  16. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    I captured FPV video of a test flight, manually hovering just out of ground effect and with the FPV camera pointed at some relatively close objects. If you look to the right of the screen you can get a better sense of the vibrations I'm seeing.

    I don't recall getting anything like as much vibration from the C8 before the aforementioned "arboreal adventure."

    I'd appreciate any comments you guys with FPV cameras have in terms of whether the vibration is better, worse, or pretty much the same that you're seeing.

    I've also uploaded the output from the Vibration app on the iPhone that I strapped against Boom #1 where the camera is mounted. I'm not sure it's giving me valid enough data as (a) it appears to only sample once a second and (b) if you scroll down the frequency analysis only goes up to 4 Hz. (Note: Remove the .txt suffix from the file before trying to use Excel on it.)

    Comments and wisdom appreciated. I guess I'll just have to keep swapping motors out....
    Andy.
     

    Attached Files:

  17. William Johnston

    Joined:
    Oct 2, 2012
    Messages:
    140
    Likes Received:
    5
    The jitter seems mostly left to right. What if it's not a beat frequency? What if your motors are mounted, on average, with a slight tilt to the right? That would give your CS8 a tendency to turn counter clockwise. After turning a little bit the flight controller would notice it's change in facing and correct it with a little turn clockwise. That would be repeated over and over again and a camera mounter on the boom would see a left and right jitter. Check your average currents and see if your counter clockwise spinning motors are pulling more current than your clockwise spinning motors. Try getting your motors perfectly vertical.

    just my two cents
     
  18. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,372
    Likes Received:
    1,162
    William: Considering your day job, your $0.02 are worth at least $1.50 in my book! :)

    I set the motors up with the hub flat (using a spirit level) and then using a small spirit level held against the motor's rotor to get them vertical both on the side 90 to the boom axis and along the boom axis. But, since I treated the motors to a shock stop (the aforementioned arboreal adventure), I could also see that perhaps one or more of them is sick. I'll check the GPX files to see what's going on there -- all good thoughts, for which thanks.

    I've also ordered up new motors so that I can systematically swap them all out.... good job it's not a duodecimocopter!
    Andy.
     
  19. William Johnston

    Joined:
    Oct 2, 2012
    Messages:
    140
    Likes Received:
    5
    LOL
     
  20. William Johnston

    Joined:
    Oct 2, 2012
    Messages:
    140
    Likes Received:
    5
    One thing to try to see if it is your flight controller adjusting the yaw is to adjust the "Yaw P" and "Yaw I" on the "Gyro" panel in MKTools. If I'm right, lowering P and I will slow down the vibration. If you set P and I to zero the CS8 should spin.

    It might be a little risky doing that so this experiment should only be tried by people with deep pockets.
     

Share This Page