Displacement using accelerometer?

[MatB]

Hello,

I've been set the task of calculating the displacement of a moving (in two dimensions - surge and sway) platform relative to its fixed base. The platform will move approx +-2m from the base over a period of a few seconds. I need to measure the displacement to within 5cm over several hours upto possibly a couple of days.

I have read http://www.freescale.com/files/sensors/doc/app_note/AN3397.pdf and understand the calculations involved but i've never used an accelerometer before and am not sure whether in a practical situation the drift/noise will render the system unusable too quickly.

Would something like the MMA7260QT accelerometer with MC9S08QG8 microcontroller fit the bill? I expect we could recalibrate the system every few hours but would it be accurate over even that sort of timespan?

Any suggestions are welcome.

Thank you,
Mat

[HIWAYSTAR]

What kind of mechanical geometry are you measuring?  The frequencies and displacement you mentioned might be better measured with something simple like a string pot if it will fit your geometry and environment.

[MatB]

Thanks for your replies.

I think i will go ahead with the accelerometer and see what the errors are like before trying some other system of measurement if necessary.

Thanks again!

[jindal]

Were u finally able to find distance by using only acclerometer IC?

[mke_et]

Well, realize that with the resolution of your A/D converter, and the swing of the part you use, errors can compund quickly.

I'm assuming you're using an accelerometer with an analog voltage output. Try to limit the errors, at least from the sensor inputs, but amplifying the signal so that at your expected max acceleration, you are 'bumping the rails', to get the best resolution you can.

Also, if you pick up a math package, use the highest precision math you can.

You won't be perfect, but it may do the job.

[bigmac]

Hello Mat,

My gut feel is that the accelerometer solution will not maintain sufficient accuracy for position calculation.  The double integration process is likely to introduce significant errors, exacerbated by very low levels of acceleration (2 metres movement over a few seconds), and therefore very low output levels.

You might wish to consider alternatives that directly measure position/displacement.  One possibility might be an ultrasonic method, measuring reflected pulse delay.  I guess the limiting factor here is that the amount of movement of the platform over the measurement delay period must not be large.

The measurement delay period will also limit the sampling rate.  You may also find that you need to alternate the measurements for each plane, to minimize cross interference.

Just a few thoughts.

Regards,

Mac