Comment by ctech4285 on November 26, 2011 at 1:21pm

what does it look like wihtout the motors running?

Comment by Jack Crossfire on November 26, 2011 at 1:23pm

The trick is to make the IMU heavier.

Comment by Johann Van Niekerk on November 26, 2011 at 1:24pm

no motors running looks like this ! how do you make the IMU heavier ?

Comment by ctech4285 on November 26, 2011 at 1:48pm

hmm,....intresting

so i would think you would see a difference in vibration depending on the mounting method, you dont...

theories;

vibration transmitted throw the cables

could be picking up sound waves

electrical noise

how about structurally isolating the sensors so there is no physical contact except the cables with the motors?

also one could glue a heatsink or any mass on top of the sensor chip...

do you have the units for these measurements?

Comment by ctech4285 on November 26, 2011 at 1:51pm

how do you make the IMU heavier ?



just make the sensor heavier, glue metal on top it, for high frequency low amplitude vibration doubling the sensors mass would have the same effect then doubling the boards mass

Distributor Comment by Martint BuildYourOwnDrone.co.uk on November 26, 2011 at 2:02pm

How well balanced are the props on this "test" rig?

 

Regards

 

Martin

Comment by Johann Van Niekerk on November 26, 2011 at 2:10pm

Ctech

Can you show me a example of what you mean ? @ Martin the props being balanced or unbalanced is not really relevant for this test as I want more vibration to see if different mounts make a difference.

Comment by Iain Peet on November 26, 2011 at 2:37pm

Honestly, this is a pretty poor presentation of the data for noise comparison.  It's difficult to compare magnitudes, and nearly impossible to tell if a perceived difference is statistically significant.

 

At the very least, I would suggest looking at the measurement means and standard deviations, rather than straight time domain.  Evaluating full covariances and checking for correlations may be illuminating.  You should certainly do some basic difference-of-means statistical tests before drawing conclusions.  It might be useful to look at histograms, both to validate a normal distribution assumption, and to get an intuitive grasp into what the statistics are telling you.

 

You have to be really careful with damping.  Switching a bolt for velcro or velvet will increase your damping, yes, but it also reduces rigidity, which will decrease resonant frequencies, which can easily result in higher amplitudes in the part of the spectrum you care about.  I've heard reliable reports from people working on high end quadrotors that they got best results by making the frame and mount as rigid as possible (shifting the vibration spectrum into higher frequencies) and then removing that noise with a low-pass filter.  Also, apparently, with loose mountings, you have to watch out for an effect where the vibratory motion adds up to a circular trajectory, which does all sorts of evil things to your gyro measurements.  

 

It might also be illuminating to run your results through a FFT, and see what your various mountains are doing in the frequency domain.

Comment by ctech4285 on November 26, 2011 at 2:38pm

Can you show me a example of what you mean?

assuming you mean structurally isolation:

put the sensor board on the floor and hold the copter in the air, or suspend the board from the ceiling with a rope of sorts while the copter is on the ground. any which way you can keep the sensor board from making physical contact with the frame will do. the point here is to eliminate the possible travel routes of the vibration. so you would be able to rule out/in different routes...

 

 

 

 

 

Comment by Sebastian Gralla on November 26, 2011 at 3:02pm

I don't think sound waves could do ANY effect to these sensors (at least not in range we are working in)

as well you could ignore electrical noise, vibration transmitted through cables should be a very small amount with normal wires, not relevant in comparsion to mechanical vibrations caused by mounting.

have you tried vibration absorbers? or just normal thick dubble sided tape? works very well on my coper.

Comment

• ‹ Previous
• 1
• 2
• 3
• …
• 8
• Next ›
• Page