I was browsing apps with my phone and found some cool Seismographs... I was wondering if we could have a seismograph in order to detect the vibration frequency on our drones using the APMx.x within the mission planner and get some accurate oscillation frequencies and find out what we have to deal with. Then we can just jump to the relevant dampeners or other procedures to attenuate those oscillating frequencies... . 

Views: 1226

Reply to This

Replies to This Discussion

you should have a read of this post on ArduPlane 2.8 release. http://diydrones.com/forum/topics/arduplane-2-68-released

It gives you an idea of the noise that vibration an generate, detected via the gyros etc... and what filtering is need to keep the plane flying

On my last flight we pulled the sound track from the HD camera into Matlab and did a PSD.  There are likely free tools to do this.  Visible were the 1st and likely the second wing bending modes as well as the pitch oscillations that the plane was experiencing.  There are likely other ways to instrument a plane using a video cam and mic input that you could look into.  High rate data (kHz data) is really needed to do a good job, but that really depends on what you're looking for.

Most navigation grade IMU's do use mechanical attenuation. And, you can characterize different materials for attenuators just as easily as you can measure the vibration source. 

I am generally a multirotor flyer where props most of the times will generate vibrations which will most of the times will affect the video rather the flight... I tried to record some vibrations with my phone app which seems to be very accurate regarding a single light finger touch on the table will generate some be picked up by the phone but the export file is a bunch of numbers each reffers to one of the 3 axis X Y Z. I don't know how to translate them into usable data... An example file from one of my Seismograph recordings is attached can be viewed with matlab or excel... 


Notice that the data is 30 Hz. You can determine magnitudes of low frequency vibrations, maybe up to 10 Hz.  What I was getting at earlier was connection of an external mic or other analog device to measure relative vibration levels.  You should be able to hard mound on any component you're interested in.  A sound track is cool because the sampling rate for MP3 (for example) is usually 44.1 kHz.  The fast loop on APM may run close to 50 Hz and you could get data this fast, but if there's an issue with vibration it's imposible to tell what's really going on.

What most people do (including me) is float the IMU since frequencies important to an autopilot are pretty low.  i.e. a very "soft" platform for the IMU.

Of course you could guess at transfer functions for structure, the isolator.. measure some stuff and optimize the isolator.  What's cool is that you can actually do this as a hobbiest since we have inexpesive analog devices, sound track recording and software to help.  I was using this technique since I was flying a plane into the atmosphere at 350+ mph and interested in flutter.  The flight control surfaces didn't flutter. 

Thank you for that long answer Larry. How did you manage to calculate the values and figure that there is oscillation at 30hz ? It was a simple test i was slightly holding the device with my hand to prevent a fall from the arm but I could make tests a bit more carefully and end up with some cleaner results... Not these values where recorded without any probs on the quad... Just the motors spinning... 

Nadar,  I didn't suggest a frequency for vibration content.  I did compute the timestep, determine the frequency as 1/dt and say that it's impossible to determine meaningful vibration content since the Niquist frequency is only 15 Hz.  My other suggestion was to "guess" at an isolator or use analog sampling so that you can measure the vibration content.  If you really want to do this you should look at SW that does a PSD on a sound track.  Then look at about 1 second of any sound track to demo the SW.   At that point you can look at how to record the vibrations and you're off and running.  Alternately, you can guess at an isolator and move on.

Recall I was only doing this because I had significant dynamics in my last high altitude attempt.  I a had also cut down the frequency of my telemetry to 3.3 Hz to save space.  Autopilot oscillations were right on top of this frequency and my data was all aliased. 

Larry Thanks again for the answer. I am not sure i want to get into such detail. Sounds a bit complicated for my mind :D I was just just suggesting that an app or a feature could be develop for the APM so that you get a spectrograph of how vibrations from the frame are passed down to the Flight Control and then try with random or specific dampening to eal with the vibrations. Since a simple samsung can do this i thought the APM can also do this since it has the gyros and accelerometers... 

Another thing that would be useful in diagnosing vibration is a CLI test mode where you can spin each motor in turn, either at a constant rate or being driven by the throttle stick.

Otherwise you need to go unplugging and re-plugging ESCs to isolate each motor/prop, which is a pain, and the throttle won't be constant as even in Acro mode there seems to be some smarts in between the transmitter's throttle position and the power going to the engines, it doesn't seem to be constant, which makes vibration testing tough.

Reply to Discussion



Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2020   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service