How we do autopilot vibration-testing at 3DR labs

This is a quick demo of how we vibration-test our next-gen autopilots at 3D Robotics labs. Two big bass speakers are programmed to generate vibrations at a range of frequencies, and the autopilots are mounted on a structure that's driven by them in two dimensions. We then measure the output from the autopilots to look for aliasing and harmonics, which helps us design the internal vibration-insulating foam, brass masses around the sensors, physical shape and software filters.

This particular setup compares two autopilot modules with different vibration-damping, each with two redundant sensor packages. The aim is to design the vibration-damping so that all the normal vibrations from motors and props are handled well.

Views: 3065

Comment by Rob_Lefebvre on April 18, 2014 at 6:20pm

Love to see the subwoofers being used.  I did an internship at a division of General Dynamics about 20 years ago, and we had a massive shaker table you could put an entire PC on.  I thought it was funny that when we ran the profile, it sounded exactly like a fighter jet.... probably shouldn't be surprising because that's exactly what we were testing. ;)

But I used to joke that it was the worlds biggest subwoofer. :)

Comment by keeyen pang on April 18, 2014 at 8:31pm

This is absolutely the right direction to go. I always wonder why our APM and even Pixhawk is hard mounted in their case. The vibration and magnetic interference issue may have cause lots of  "mystery" crash or bad behaviour. Some of them may have gave up on the DIY drone project. 

Can we have the vibration proof for APM and Pixhawk soonest possible rather than the "Next Generation" board ?

Comment by Jack Crossfire on April 18, 2014 at 9:22pm

Takes a lot of weight to dampen vibration. That's why they only hire PhD's in subwoofers.

3D Robotics
Comment by Chris Anderson on April 18, 2014 at 9:46pm

Keeyen: the external vibration-mounting foam that ships with Pixhawk is fine and does the same thing. Nobody should be hard-mounting it, as the instructions make clear. 

Comment by Harry on April 18, 2014 at 10:12pm

You guys don't need that test rig, I'll sell you my tricopter with the slightly twisted arm and out of balance props.  It looks like it shakes about the right amount.  It flies the APM like a champ.

Comment by keeyen pang on April 18, 2014 at 10:12pm

Hi Chris, Thanks for the info. I always use gyro foam to mount my APM and will use the 3M foam tape come with Pixhawk. Great to know the simple foam dose the same thing as expensive vibration damping casing. 

Comment by Harry on April 18, 2014 at 11:21pm

I use a 1/2 inch thick slab of modeling clay wrapped in masking tape for the mass and the damping. 

Comment by Quadzimodo on April 19, 2014 at 8:23am

Very interesting.

I have a whole carton of linear motors sitting in a box which I have been saving for a home cinema project (which I will likely never complete) which would be well suited to a bench rig like this.  These were originally manufactured for US company Aura for their 'Aura Interactor' video game peripheral.  I was workin' in purchasing at Jaycar Electronics when we picked up around 30,000 of the complete gaming packs from a liquidator for less than 1USD per unit. We also bought up the last of the amplifiers and raw motors surplus.  Everything sold like hotcakes, but the prices were cheap cheap cheap - we actually had so many amplifiers we used to stuff them into $5 1kg electronic component bargain bags (big bag full of caps, resistors, ICs, connectors, fasteners, etc).

These might be even better suited to the task, and much more compact than playing with large and cumbersome 15s. They also offer the added convenience of having a nice solid bolt for easy attachment.

An image of one of these linear motor is pictured below (CHRIS - If you want me to send you 3 of these for an improved rig please don't hesitate to ask).

Could someone explain to me why testing is conducted at these frequencies.  12-250Hz seems incredibly low (even going subsonic).  Wouldn't the frequency range we need to model be much higher, so to be closer to the frequency at which our props are rotating? More upper-mid and treble than earth shaking bottom end. Say around 2-10kHz?

Also, I must admit seeing subwoofers on the test bench brings back so many memories!  Below is a scan of one of my earliest published works.  Please excuse the poor standard of my writing in this editorial, I was a total greenhorn when I wrote this way back in 2002 (fresh outa high school with zero writing experience).  Back then, all the best gear was still made in the USA or Japan, with China was starting to find it's feet.

Comment by Morli on April 19, 2014 at 10:18am

Interesting  topic.

Monroe ,thanks for sharing the video link. I haven't seen this effect before.

Comment by Quadzimodo on April 19, 2014 at 10:45am
Monroe - Yeah, that is the same sorta thing, just smaller. The Aura shakers are basically just a bare motor structure. They are engineered much like an inside out speaker, with the voice coil mounted stationary on the frame with the magnet able to move inside the gap. They are quite efficient and can handle a 50Hz sine wave up to ~50WRMS before running into difficulties.
Running from a decent amplifier (they have an impedance of 6-ohm) they are nice and clean and surprisingly linear, but the very heavy moving mass gives them a particularly steep impedance curve, with a natural roll off of what seems like >18db per octave. As a result, they are the very definition of low fidelity.
Admittedly, attached to a suitably live surface or object (able to provide as much constructive distortion as it does destructive distortion) and with a powerful narrow-Q 32-band Equaliser and pre-amp stage capable of an unusually high dynamic range, you could get a pair of these things to deliver an acceptable image. But it would hardly be worth the effort. Always wanted to do it with a pair of stylish Frank Gehry inspired curved plywood chairs, but it is just too great (and expensive) a task (I am no acoustic engineer and my CAD skills are limited to sketchup).
The best way to make a wall become a speaker is with a line array or electrostats. Both can be DIY'd with relative ease.
BTW - Anyway wanting to get hold of one or more of these Aura shakers cheaply to play around with, just search for Aura Interactor, Aura Vest or Aura Cushion. You can find them on eBay for next to nothing, and they will run from any 4-ohm stable amplifier (preferably one with an adjustable active low pass crossover.


You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2020   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service