I'm working on a project that requires logging attitude data from high-powered rockets (amateur types, not military). I'll need to monitor all 3 axis. These rockets can see g-forces upwards of 20+ g's, but for relatively short periods of time (2 to 10 seconds), in addition to the deceleration up to apogee (perhaps -5g to -10g). Overall, the measurements will need to be done over a total time of 10 to 30 seconds.
I've been looking at the ADXRS646 single-axis gyro, which shows great performance. The spec sheet shows a Linear Acceleration Effect of 0.015°/sec/g, which in comparison to other MEMS, seems very good. From what I understand (for example), at say, 20 g's, for 10 seconds (a fairly extreme case), I could expect a 3 degree error (0.015 * 20.0 * 10.0s = 3.0°), from each of the three gyros. While this really isn't THAT significant, can this be calibrated out? I'll have a rate table I can use, which I believe I can also simulate linear acceleration with as well (via centrifugal force).