I have a sample, but I haven't had time to play with it yet. It is a rate gyro (measures angular rate rather than directly measuring attitude). To get attitude, you have to integrate rate. Rate noise density (max spec) is quoted as 0.05 degrees/sec/sqrt(Hz). In other words, in a 100 Hz bandwidth (reasonable for flight control on small objects) you'll see 0.5 degrees/sec of random rate error. You can reduce that by filtering the bandwidth down to, say, 1Hz, but you can't use a slow signal for flight control.
There are also temperature and acceleration effects. The summary specs say "null drift over temperature" is up to 24 degrees/sec (max) and linear acceleration effects are 0.2 deg/sec/g max. The discussion in the application note gives much better numbers. Under "temperature calibration" it says "overall accuracy of 70 degrees an hour or better is possible" using an elaborate temperature calibration algorithm. The chip provides a temperature output to assist calibration. Under "acceleration sensitivity" it says the first-order null shift has zero mean and 0.02 deg/sec standard deviation, over a large number of samples. Presumably the acceleration constants can be calibrated per unit, in a centrifuge. More expensive inertial nav systems use parts which are individually calibrated for first- and second- order acceleration effects in all three axes, and run in a constant-temperature environment. This calibration goes a long way toward explaining why they are more expensive. It's probably possible to build a lower cost IMU using parts like this, but they still need to be calibrated to get decent performance. (I wonder if there's a market for that? Hmmm...) Cheers! --Stu _______________________________________________ ERPS-list mailing list [EMAIL PROTECTED] http://lists.erps.org/mailman/listinfo/erps-list
