accelerometers

[fatpo]

Is it possible to calculate the angular velocity using only accelerometers? It can be done with gyros, but i want to implement it with accelerometers, maybe combine 2-3 or them i same application...is the any way?

 

[itsmoked]

Sure, why not? You might need to know the radius or add more sensors to compute it.

 

[fatpo]

Hmm...could you be more specific? (post some links btw...)

 

[itsmoked]

Hi,
Check the links in on this page.

http://search.analog.com/search/default.aspx?query=angular+velocity&local=en

 

[rjgoebel]

You could definitely do something like this. We just got done researching using accelerometers for use in a stylus device, as a PC input device.

The key would be knowing what point your accelerometer was rotating around. For example, if you mounted a 2 axis accelerometer on a carosel (sp?), you could orientate one of the axis to measure angular acceleration and the other axis to measure radial acceleration. Both of these signals could then be integrated to determine an angular and radial velocity.

I am not sure if saying radial velocity is the correct way of wording that. But regardless, accelerometers are very powerful devices and yes, you could do something like what you mentioned with them. Most of them output either an analog voltage proportional to the acceleration, or a PWM voltage thats also proportional.

But to give you an idea of what accelerometers can do on the high end, they use an array of 9 of them (single axis) in fighter cockpits to determine exact rotation and position data (in 3d, so 6 degrees of freedome - translation in x,y,z. And rotation in x,y,z) after someone has ejected from a plane. These things are really neat.

Jim Goebel,
Electrical Engineer
Mid-West Forensics, Inc.

 

[fatpo]

Jim, do you have anything that you can share with me (anything)? Actually, i'm more intrested in the "yaw" rate. It seems that, as you mentioned, these things are used in military purpose...its too bad that they dont put it on the net

 

[rjgoebel]

fatpo,

As far as actually generating the angular velocity from accelerometer output, I need to know more about what application you are using the accelerometer for. If you want to put the things on a plane to measure yaw, which I understand as rotation about the center of gravity, you can do one of a couple of things.

The easiest thing I can think of is to place one accelerometer axis parallel with the vector that points straight in the direction that the plane travels with no yaw, ON the center of gravity of the plane (with respect to the wings). This accelerometer axis will output a linear acceleration when the plane accelerates, but will not have any component of that acceleration that accounts for rotation or an angular acceleration. Now once you have that accelerometer placed, you are going to want to place another one, pointing in the exact same direction, but out on a wing, at a certain known distance. This accelerometer will output an acceleration vector that has both a linear acceleration (same as the other one) as well as an acceleration from rotation (angular acceleration).

Subtrating the two waveforms will give you *just* the angular acceleration.

And remember, you still need to integrate this result to get back to the angular velocity that you want.

Now, consider that most analog accelerometers output 2.5V @ 0g, or no acceleration. Each axis' output varies from 0V up to 5V, with 0V being the negative most acceleration and 5V being the positive most acceleration. If you orientate your two accelerometer axes correctly, you can simply continuously do an operation like 2.89V-2.7V = 0.19V which will correspond to a certain acceleration measured in g (9.8m/s^2). From this you can calculate an angular acceleration, because your component of acceleration due to angular acceleration is measured, and you have the distance from the COG that the plane is rotating about.

And once you do all that, you can use an analog integrator, which can be made out of an opamp and an RC circuit pretty easily, although whenever doing analog integration, something called DC drift is a big issue.

And keep in mind that all along the way if you want to go from analog to digital so that maybe you can do some DSP processing or more in depth signal processing of those acceleration signals, you can always just use a good ADC and program a DSP.

Sounds like a really fun project you're working on, does your company have any openings
Good luck!

Jim Goebel,
Electrical Engineer
Mid-West Forensics, Inc.

 

[fatpo]

Hi jim,

I'd be great ?f i had a job at all ... I'm on my 22 and just graduated from KTH in Sweden (E.E)...and looking for intresting jobs...right now I'm in turkey and enjoying the summer...the project will be continued when i'm back...

however thanks for your answers. be sure to get back here cause i'll post some more questions...

see yaa!

 

[itsmoked]

ahhhh... the insouciance of youth...