[Qbranch]

Yes, I resurrected an old thread... but I'd rather do that than start a new one on essentially the same subject.

I know doing double integration blows any error in a sampling of the acceleration up very fast over time (squaring and accumulating the error each time) but, has anyone done the double integration and can share some findings?

How fast does error show up? In what way? How accurate is it?

One application I'm looking at would require the double-integrated accelerometer to remain accurate for about 100ms.

Thanks for any data you can offer,

q