Trouble sensing rotational freefall MMA8451

[davidgroff]

Hi,

I have two questions:

1.  I am looking at document AN4070 and it seems like it is possible to detect rotational freefall but I can't get it to work.  Could somebody help dumb it down for me?  I was able to get linear freefall to work great.

2.  As I said I was able to get linear freefall to work.  However it works on both positive and negative acceleration.  I want to make it so that it only triggers on negative (when the object drops, not when it is thrown in the air).  Is this possible?

Thanks

[michaelestanley]

David,

I'm going to respectfully disagree with the author of AN4070 on the topic of "rotational freefall".  I would like to refer you to a blog posting I created some time back at Accelerometer placement - where and why.  Equation 15 at the end of that posting is:

A_r' =A + A_r +  <-- Inertial Acceleration

2ω X V_r + <-- Coriolis Acceleration

α X r +    <-- Euler Acceleration

ω X (ω X r))  <-- Centripetal Acceleration

I'll refer you to the original blog posting for definitions of terms.  The important point is that AN4070 was attempting a Heuristic approach to detecting rotational free fall that does not consider the full implications of the equation above.  A single accelerometer cannot by itself separate all the effects shown in the equation.  You need to either add additional sensors or restrict the types and/or ranges of motion you are considering.  I would not spend any more time pursuing the AN4070 rotational free fall approach, and I'll be recommending that AN4070 be updated and/or removed from the website.

As to your second question, since there is no force, and therefore no acceleration other than gravity, on your sensor once it leaves your hands, I don't see an easy way to differentiate the peak from the rest of the trajectory.  You would have to compute the initial vertical velocity & acceleration of the object based upon accelerometer data prior to release, then figure out how long in free fall you need to reach 0m/s at the top of the arc.  Very tough to do for the general case.  Maybe someone else is more clever than me...

Mike

12/31/2014 4:44PM Addition: David,check out AN4446, which addresses determining the high point of the trajectory.

[davidgroff]

Mike,

Thank you for the detailed reply.

Your blog post was very informative.  It was also somewhat advanced for me.  From what I gather, I might be able to detect rotational freefall and as well static freefall if I had two acceloromters and a microprocessor to make computations based on the data from both sensors?  Is that the only way?  Would a gyro sensor help?

Document AN4446 was helpful in understanding how to detect positive and negative freefall.  I don't know if I want to take that approach or not.  Do you think I could use a barometer/altimeter to detect altitude in order to distinguish positive or negative freefall?  - I see that the MPL3115A2 has a resolution of 0.1m

Thanks for your help/ideas,

David