[aprssig] Cool new sensor
steve at dimse.com
Wed Mar 25 00:07:22 CDT 2015
On Mar 25, 2015, at 12:25 AM, Scott Miller <scott at opentrac.org> wrote:
> My money would be on the magnetometer for linearity and long-term stability, since you're just counting zero crossings over time. You don't need to worry about offset, non-linearity, or drift. Or the mounting position of the sensor.
I'm thinking off the top of my head, but if an accelerometer were mounted in an anemometer perfectly square and at rest, the vertical axis has 1 g, the other two 0 g. As it spins up the 1 g axis does not change, one horizontal axis has the centrifugal force and the other varies depending on whether the anemometer is speeding up or slowing down. Since the centrifugal axis directly indicates the speed it seems to me an easier option. It does require a zero calibration in a real, imperfect mounting, but as long as the mounting were close enough that the zero angle approximation applies it is simple subtraction of the value at rest. Each measured centrifugal acceleration is directly convertible to a speed, so you get windspeed readings at the rate of the sensor, faster than the zero crossing magnetic method where the windspeed rate will be available once per rotation (or would it twice?), and the sensor rate, rather than being the data rate becomes the resolution limit.
Imagine the anemometer rotating at a rate one-fourth the sensor rate. Using the accelerometer you get a wind reading four times per revolution, and the error on each is the sensor error plus any uncorrected level error. If you use zero crossing magnetometer you get one or two readings per revolution, and the error is the sensor error +/- 12.5%. Faster sensor rates will lower the 25% error, but you are still stuck at one or two readings per revolution, whereas increasing the accelerometer sensor rate gives you more frequent data, with more opportunity to average out noise.
Not to mention magnetic north drifts but 1g has been 1g for a very long, long time!
Wanna have an anemometer throwdown? ;-)
> And yeah, the fact that handheld anemometers are dirt cheap now is why I didn't bother with the phone thing.
Damn progress ruins another cool idea!
More information about the aprssig