paparazzi-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Paparazzi-devel] imu sensor filtering?


From: Eduardo lavratti
Subject: Re: [Paparazzi-devel] imu sensor filtering?
Date: Sun, 19 May 2013 18:21:00 -0300

Chris, where in the wiki i can find the mechanical box filter ?
I am making a 2.60m gasser flywing and i will need some dumper for IMU.

Regards.

Date: Sun, 19 May 2013 14:01:26 -0600
From: address@hidden
To: address@hidden
Subject: Re: [Paparazzi-devel] imu sensor filtering?

Hi Chris,

When we worked on dampening vibrations in our gas-powered plane, we found it was definitely necessary to take the time and measure the environment to properly design the mount.

For this, we mounted the aspirin 1.5 in the aircraft and took accelerometer samples in all three axes at a 1kHz sample rate. We only did this in one location, but ideally we would do this in a number of different areas in the airframe. After collecting this data, we did some very basic analysis (namely, ran it through an FFT for both individual axes and in different planes by summing accel vectors) to determine the major problem frequencies (should be done for many throttle settings across whole range). Unsurprisingly, the worst vibrations were in the plane perpendicular to the crankshaft axis.

Following this, we prototyped a few mounts and then tested the frequency response in different axes. For this we used the same 1kHz measurements, and then looked at the IMU/AHRS output, but applied known frequencies using a big shaker table. Probably, one would need to try the mount and measure the acceleration right in the aircraft. Also, we had to try a number of different filter window sizes to "tune" it for the specific mount and airframe.

You could probably write some accelerometer/gyro measurement firmware for your specific autopilot, stripping out all unnecessary code, and push the sample/periodic rate as high as possible. I was using an Arduino because I had some existing code, then pushed out 6 bytes (2 per accel axis) to a small C ivybus script, which I could then log with the regular server and plot with the realtime plotter.

In any case, it is probably worthwhile setting up a test suite as it may save you time and frustration in the future.

Thanks,
-Stephen Dwyer


On Sun, May 19, 2013 at 12:57 PM, Chris <address@hidden> wrote:
Hi
I have read the wiki page about vibration dampening about 100 times and Monday i will use 2 more imu mounts, one with ear buds (memory foam) and one with latex foam.
I have already started constructing a box similar to the one in the wiki and i added a moving
average  filter in the Accelerator raw values like the wiki says.
I will also try to set the mpu6050 to use a 2g range and see what happens.
So far the problem is minimized and i think i can fly (that was the main reason that i selected a
twin cylinder engine and not a single.
Btw with IFR do you mean a FIR filter or a Instantaneous Frequency Rate filter?
I have been thinking of measuring the Engine's rpm and construct a digital notch filter where i move
the notch frequency according to the measured engine rpm.
Chris


On 05/19/2013 07:00 PM, paparazzi-devel-request@nongnu.org wrote:
Hi Chris,
the problem is vibration generated by motor and propellers.
I read in a new dissertation  that a 3 th. order  IFR filter with 50 Hz
for the raw accelerations values
before calling the ASH  can help. Currently only a first order is used.
For the MPU6000 the range is 16g . You can set it down to 2g  an clip
all valuse over 1.5g .
There is in the wiki mechanic  damping with a wood box and additional
mass to fly combustion engines.
If you fy multirotors you can use only backstepping and no differential
(D) of the angle.
This will reduce noise of the motor values set point about 30%.
Heinrich


_______________________________________________
Paparazzi-devel mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/paparazzi-devel



_______________________________________________ Paparazzi-devel mailing list address@hidden https://lists.nongnu.org/mailman/listinfo/paparazzi-devel

reply via email to

[Prev in Thread] Current Thread [Next in Thread]