Isolating IMU's in high vibration environments

[Jack Crossfire]

So what methods for mounting IMU's are giving the best vibration isolation? Have found no matter how much padding is used, the wire interconnects still carry vibration to the IMU but have not found any suppliers of flex wire that might reduce the problem. If the IMU is not tight enough on the airframe, low frequency noise arises from wobbling.

[hg1]

How about estimating the frequency of the vibration and using a bandpass filter in software to knock it out ?

[Jack Crossfire]

The gyros & accelerometers are skewed by high frequency vibration at the analog stage since they are based on oscillating mass themselves. Documented here http://www.rcgroups.com/forums/showthread.php?t=814381

[hg1]

I don't know anything about your setup, but those look like DC biases. Are you using a common ground for all of your electronics, tied to the negative side of your main battery, or are there separate supplies for electronics and motor power without a common ground ?

[small_rcer]

I know this sounds strange but have you considered mounting a lead plate to the bottom of the IMU and then wrapping the whole in very soft foam. The thought is that by increasing the mass, you would reduce the vibration passed to the unit by the wires. I think that 5-10x the mass of the imu board should be sufficient. So if it is about 5 grams start with a 20 g piece of lead or other mass to damp the vibrations.

Just a thought and I know how we never want to add weight to a plane unless absolutely needed.

Jim H

[kd7ost]

I know this sounds strange but have you considered mounting a lead plate to the bottom of the IMU and then wrapping the whole in very soft foam. The thought is that by increasing the mass, you would reduce the vibration passed to the unit by the wires. I think that 5-10x the mass of the imu board should be sufficient. So if it is about 5 grams start with a 20 g piece of lead or other mass to damp the vibrations.

Just a thought and I know how we never want to add weight to a plane unless absolutely needed.

Jim H

Doesn't sound strange at all.

How much vibration are you talking about? Frequency and amplitude wise I mean. Maybe easier, what is causing the vibration? Glow or gas powered plane? Electric motor and prop combination. Here is a good read in vibration isolation for cameras.

http://www.hicam.com.au/art_vib.htm

Fast shutter speeds dictate less care than an IMU might need. But just employ Newtons Laws of motion. In this case I agree with the statement from small rcer. Padding the device won't help much if the padding isn't right and the inertia is lacking in the IMU mount.

I found the bast padding is the latex foam sold in hobby stores. It has visco elastic or memory propertied that are good for us. Not like the memory pads we put on beds. That material reacts too slowly. But the latex foam has what I like to call good gush qualities. It's almost like a gel. It will absorb motion in one direction, but doesn't rebound back and overshoot it's static position. Elastomeric. Like a shock absorber. Some foams are just crap to the vibration cause.

Next is how much is used to absorb what amount of mass or weight. Using small pieces all around, but not loose nor pre-compressed is critical. The mounted piece should just touch the mount points (all four corners and all directions of travel) and without compressing the foam. This keeps the foam at optimum compression for light loads. It also provides range of motion.

Finally the mass. If the IMU and mount is so light that it can't compress the foam as it tries to stay still, it lacks the inertia to remain at rest. Weight in this case is our friend as it serves a critical purpose. The mass will allow the foam to compress around the vibrating structure as it resists motion because of its higher inertia. If a little weight doesn't do the trick, add more weight and/or remove more foam. (One at a time is best) The above link discusses this in great terms with the brick and sugar cube analogy. You just have to experiment and strike a balance.

Dan

[yuri_base]

Jim, this is a good idea - and it seems at first thought that the benefit is proportional to mass. That would be true if we had the same magnitude of force acting on the sensor (F = const), increasing mass would then decrease parasite accelerations in direct proportionality to mass: a = F/m->0 as m->infinity.

But since aircraft's mass is much greater than sensor's (even with additional weight), it's more the question of resonance coupling between "infinite" mass vibrating at certain frequency and smaller mass attached to it. By changing sensor's mass and the springiness of the foam or cotton supporting it, you change the frequency. If frequency resonates with the main frequency, you get more noise.

Perhaps, we can learn from earthquake resistant high-rise buildings how to negate the vibrations?

Yuri

[kd7ost]

After a little pondering I had an "Ah Ha" moment. I answered above with respect to soft mounting my camera on a gasoline engine plane. You don't have to add lead. This would be tough for you in your application because an IMU is so small and light. The wiring transfers the vibration by your comment. It would take an awful lot of extreme mass compared to the weight of the IMU to soft mount it.

You already have the mass on board. If you made a mount like I described but custom made it for a battery pack, them mounted the IMU over, under, whatever works in your application to the battery mount plate you would have a bunch of mass in place to dampen the IMU. Creating some C channel around the battery mount plate would allow you to add or remove some foam as needed to decrease compression requirements.

Anyway, it would be a pretty broad frequency range dampener and I'm sure the amplitude can't be too extreme. Probably much less than 1/64th inch of motion. (Unless your flying a single thumper lawn mower :D ) Then just put a gentle loop in the wiring as it gets close to the mount to allow motion without the wires being a factor. Just strain relief the battery pack end.

My .02

Dan

[workshop]

Airborne Systems makes closed cell polyurethane foam pads in different densities for protecting R/C equipment. I use the blue and pink (don’t have the ratings handy) for all sorts of protection and dampening needs. There is only an email address on the package I have here in the shop:
airbornesystems@cox.net

[air]

I've used the Carvec helicopter IMU autopilot and the method that JohnC recommends is to use small silicon squares (about 5mm squre) on 3 sides of the IMU enclosure. It works pretty well and maintains the orientation of the IMU perfectly. Using soft foams etc doesnt work because there's potential for the IMU unit to shift over time which will affect it's attitude reading. He told me that he'd tried a lot of different materials and that gave the best results, I tried a few options myself and had to concur with him.

[wadiprawita]

Dear Air,

Where can we buy this small silicon square ?

Regards

-doni-

[Ceros]

Jack, the only time mechaical insulation will make any differance to your IMU output is if the vibration is so bad it is damaging the IMU.

Otherwise, the only realistic solution to unwanted IMU outputs caused by vibration is through the use of Kalman Filtering.

[Jack Crossfire]

The problem was solved using a sandwich enclosure, making the IMU heavier & removing the flybar. Video testing showed high frequency vibration was causing accelerometer skew. Could see the IMU wasn't moving, yet the analog outputs changed as the engine spooled up & the IMU padding was varied.

Flight testing showed the accelerometers & gyros were sensitive to different modes of high frequency vibration. More vibration was actually better in some cases because the resonant frequencies changed.

The accelerometer is based on a spring mounted mass & the springs have a resonant frequency. The corialis masses in the gyros have their own resonant frequencies. Removing all those frequencies from the environment was like tuning an instrument.

[solarsystems]

I can make flex tape circuits if anyone is interested...
If the community agrees, we can make a batch of them and solder on some connectors. Might run ~$20 apeice or so.

Thom

[air]

wadiprawita - It was just standard silicon in a flat sheet but cut into squares with self adhesive backing. You could make some yourself with some silicon sealant and two soapy pieces of glass.
It worked well at reducing the frequency of vibration that was relevant in the helicopter that I was using.

Thom - those flex circuits sound interesting, could you share some info on how they're made & what type of connectors you use with them?

[solarsystems]

Thom - those flex circuits sound interesting, could you share some info on how they're made & what type of connectors you use with them?

Air,
I have a stock of 2oz copper on 1 mil Kapton I use for building lightweight solar modules. I run a process similiar to making printed circuit boards and can make smallish traces (>=10mil) up to 14 inches long.

As for a connector, I figured a single row of 10mil pins on 100mil pitch like a ribbon cable would use might work for you guys. That way you can easily get parts without having to find some goofy connector somewhere... :eek:

In another life I build stuff for space applications and vibration isolation is a big deal. We use large captured loops of cables to isolate cameras and stuff from the spacecraft for the exact same reasons you guys want to (imaging and IMU isolation). Since your distances are so much shorter, you can use something simple like my flex tape instead of wires! :cool:


Thom

[air]

Hi Thom,
That sounds very useful!
Have you any pics of the tape and / or connectors?
Thanks.

[solarsystems]

Hi Thom,
That sounds very useful!
Have you any pics of the tape and / or connectors?
Thanks.

Air,
I have pictures of diode assemblies for solar modules. I haven't made anything for this application that I can show.

Could you tell me how many traces, how long, and what connector you'd like to see? I can build a prototype in a couple days without the connectors. Since I can get just about any connector, just tell me what you need.

As for length, I would suggest it be long enough to have at least a small 'S' formed in it.

Thom

[dmgoedde]

I know this sounds strange but have you considered mounting a lead plate to the bottom of the IMU and then wrapping the whole in very soft foam.
I had to re-read this.... by "lead" you mean Pb the soft and dense element, not electrical leads. I think this is a brilliant idea, it is used in lithography equipment in the semiconductor industry. Vibration is not compatible with trying to pattern sub-micron features on a silicon wafer.

Considering the MEMS accel and gyros PCB could weigh but a few grams (the IC's weigh mere milligrams), even using 1 ounce of Pb would be a significant improvement.

[dmgoedde]

wadiprawita - It was just standard silicon in a flat sheet but cut into squares with self adhesive backing.
Do you mean Silicon (hard and brittle element Si atomic mass 29) or silicone rubber?

[dnile]

What kind of magnitudes are you seeing in terms of the noise on the accelerometers? From the pictures you posted it looks like you're getting perhaps 0.1g peak-peak?

I'm interested to know because I've been running the dynamics model integration and Kalman filtering code for my new navigation system in a MATLAB testbench I made, and I'm interested in putting in some realistic values so I can get a good idea of the accuracies.

In terms of reducing vibrations, I'd try recording a decent chunk of data from the accelerometers and then viewing the Fourier transform of it using some signal analysis software, and then removing any resonant frequencies with bandpass filters. Once the noise is more or less Gaussian, any Kalman filtering you use will be significantly more effective.

[air]

Sorry, silicone rubber, it seemed to work well at reducing the transfer of the vibrations at the particular frequency that was causing issues on a maxi joker.