Start this thread to have a place for focused discussion of the new vibration logging function of AC3.3.  I'll start by copy-pasting my initial discussion from the AC3.3 beta testing thread.

Views: 4675

Reply to This

Replies to This Discussion

Just as a point of discussion, the vibe logs from my gas helicopter are pretty interesting, as they show the opposite situation from what many multirotor pilots are experiencing.  My IMU Accel logs show fairly heavy vibration, actually over the old guidance number of +/-2 m/s, particularly in Y axis.

However, the Vibe logs, show quite good levels RMS.  I don't fully understand it yet.  I think the issue is that the board has quite good vibration isolation, which is removing much of the high frequency vibes (gas motor at 14,000 rpm or 233 Hz) but might be resonating at it's own natural frequency of the damper system which is much lower.  The Vibe RMS calc does an LPF which might be removing this lower frequency vibration.  Hover it's still affecting the accelerometers as seen in the IMU.Accel logs.

To me, this indicates the Vibe logging might be missing the plot somewhat.  But I don't understand it well enough. We're very focused on clipping, which obviously is a serious problem.  But there's this assumption the EKF can integrate right through heavy vibration, as long as it's low enough frequency to not cause aliasing.  But this seems to ignore the concept of very basic signal-to-noise-ratio.  The SNR on my Y Accel looks really terrible. Yet the Vibe logs show the Y vibration levels are actually very good.

This seems wrong to me.

So, wondering if maybe the isolation system is too soft, and causing the IMU to be "rocking" around on the dampers, causing vibration at the natural frequency of the damper system, (aka the "Roly Poly" problem)

I checked the Gyro data from the Helix, and compared it to the vibe logs from my small electric helicopter, which has really minimal vibration isolation (1mm foam tape). 

I'm not sure if there's a problem here or not.  When zoomed out, both look like "noise" at about the same amplitude.  But on zooming in like this, it's really obvious that the gyros on the electric machine (top) are showing really great Signal.  You can actually clearly see what appears to be real airframe movement, and maybe some angular vibration.  But on the gasser (bottom), it just appears to be noise.

It's a bit misleading looking at this however, as the small electric machine is very maneuverable, and it's just moving around more.  The big gasser is really stable, it's not moving, so there's not so much signal to see in the first place.  But the noise/vibration seems to be much more "jaggy", whereas you're seeing actual waves from the electric heli.

Ah, nevermind, that was a bad graph.  The gasser was sitting on the ground with the engine idling there, so that really is just pure noise/vibration.  Here's the graph again when it's flying.

Here, you can see some motion.  And also noise.  When I zoom right in, the noise is quite jaggy/random compared to the sine waves on the electric..  I think this means the vibration on the gasser is very high frequency?  In any case, no roly-poly going on here I don't think...

And here's the IMU.Accel and Vibe logging from the electric.  So this is really interesting.  I know for sure this helicopter has WAY less vibration, it's very smooth.  It does have really minimal vibe isolation on the IMU as I pointed out.  So the measured vibration is higher than it would be with equivalent isolation as the gasser has.

But what's really interesting, is the Vibe RMS log is quite a bit higher.

And then here's the really interesting thing. Here's a really zoomed in graph of the electric vs gas accel data. Electric on top, gas on the bottom.  The electric is showing clear sine waves, I believe you can actually see the rotor going around.  You can actually see a Sine and Cosine wave on X and Y, always exactly 90 degrees out of phase as you'd expect, which is what tells me this it the rotor going around.

The gasser however, looks like random noise.  There's not much signal here.  I mean, it almost looks like breaks in the data, (it doesn't have a fast card) but I'm not sure.  My understanding is that shows up as bigger gaps in the data, which is not what we have here.  So assuming that's not the problem, it's clear this data quality is much lower than the electric.  But the electric shows much higher Vibe RMS data.

Ok, so new content now.

I test flew the gas heli again last night, it was a little closer to myself so I could see it better (I normally keep it quite far away) and I noticed something.  The Auxilliary fuel tanks are vibrating very badly.  These are 800cc tanks full of gas, weighing something like 1lb each, mounted on cantilevered G10 slabs.  It's not a great system and I was aware of that but throught I'd try the "cheap" system first.  Anyway, you can see them vibrating badly, amplitude had to be about 5mm, and probably at the main rotor frequency of about 25Hz.  Obviously, this is super bad.

I then tried emptying those tanks so they weigh very little.  This drive up their natural frequency, such that they don't resonate at the same frequency as the main rotor.  They aren't visibly shaking anymore.  Here is the comparison between these two flights.

So, a fair improvement.  

Also very interesting, is I've changed nothing with the electronics, except updating from RC6 to RC7, and the logging seems to be working much better.  The data is much cleaner, and you can now see the sine waves from the main rotor.    Very odd.  I can't explain that.  Same card, same everything.

I still find it troubling that, with two 1lb masses visibly shaking at ~25Hz, the Vibe message shows no problem at all.  I do wonder if you had something like a very large Octocopter, with large, low RPM props, if the structure was not sound or the rotors unbalanced, and the motor booms were shaking at the 30-40Hz frequency of the rotors, would it notice this?  I think it won't.  This vibration frequency might not be enough to cause a problem for the control loops, but I think the Vibe message should be showing this problem to the user, because it's not OK.

Hi Rob,

thanks for opening this thread! 
Have you tried a FFT on your rc7 data?

Get Raw logging and we will compare what is really happening to the IMU and VIBE outputs. We may have simply set the filtering a little high.

Do you want any of us other users to post logs for comparison? Or just if we're noticing something strange?

I use my T-rex500 Helicopter AC3.3rc7 FW tested different frequencies INS_ACCEL_FILTER and INS_GYRO_FILTER, showing the different curves characteristic vibrations.

Respectively 20Hz and 15Hz, the obvious can be seen, at 15Hz parameter value AccY shock is much smaller!

Here's a flight where you can see the gas tanks shaking.  The video was originally shot at 60p, and it was hard to see.  When I downsampled to 30p for Youtube, it became super obvious how much they were shaking.

Wow! :-)

Seems EKF is doing a perfect job.

Reply to Discussion

RSS

Groups

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

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service