>The project sounds like a fun hack -- I would be curious as to the >resolution
>you achieve with these modules.-------
The best description on the net about building a sonic anemometer is one by
Hardy Lau:http://www.technik.dhbw-ravensburg.de/~lau/ultrasonic-anemometer.html
I have also built one that I call the "Cheap Ass Sonic Anemometer" It is part
of a portable weather sensor for use at launches of large model rockets (some
reach over 100,000 feet). It is built out of very inexpensive, off-the-shelf
components. The only wiring involved is hooking power, ground, clock, and data
lines to small off-the-shelf circuit boards (mostly from China off of Ebay). I
did lay out a custom circuit board so I don't even have to do that. A
commercial sonic anememeter can set you back $5000.
It uses 4 HC-SR04 ultrasonic distance sensors ($1 each!) as the wind sensors.
These are small modules that you drive with a 10 microsecond "ping" signal and
they output a pulse width dependent upon the distance to an object. I use two
modules facing each other on each axis and fire them in pairs. Instead of
picking up an echo, they receive the ping from the opposite module. You can
fire them around 30 times per second.
I use an $5 ATMEGA328 chip based CPU board (you could use an Arduino) to
measure the pulses and calculate the wind direction, speed, and temperature.
I was originally using a more powerful processor (ATMEGA128) that has two 16
bit counters with input capture capability to measure the pulses with 65
nanosecond resolution, but to make the device more accessible for other people
to build, I switched to the ATMEGA328 and time the pulses in software to
around 1 microsecond resolution. I was measuring all four directions at the
same time, but started seeing some cross-axis interference with some of the
SR04 modules and switched to simultaneously measuring the N-S and S-N times
followed 3 milliseconds later by the E-W and W-E times. Surprisingly the
downgraded pulse measurement capability did not affect the results to any
significant degree.
The structure is built out of 1 inch inside diameter PVC plumbing pipe and
couplers/caps... around $5 total. The ultrasonic sensors are spaced around
0.65 meters. The D/L ratio is around 20. The larger the D/L ratio, the
smaller the errors caused by the structure interfering with wind flow.
The device also contains a 10 degree-of-freedom inertial measurement unit board
($11) that is used as a tilt compensated magnetic compass and highly accurate
barometer. It also has a DHT-22 humidity sensor ($5) and compensates the sound
measurements for humidity and air pressure (which can be quite high a high
temperatures and humidity levels). With the compass measurements you can
deploy the sensor in any orientation and it adjusts the wind readings for
however it is aligned. Also, you can buy the IMU boards for less than the
price of the pressure sensor alone! The software auto-detects and supports 5
different pressure sensors, 4 different accelerometers, and the popular
HMC5883 magnetometer chip (also used in the popular ST LSM303 integrated
magnetometer/accelerometer.
It also has an AS3935 based lightning sensor board ($20)... but they don't
seem to work very well. I live next to a golf course that has an elaborate
lightning warning system and the AS3935 has never triggered even though the
golf course waring sirens have gone off several times. It also has a UV light
sensor ($13 from Sparkfun) for reporting a UV index.
I get temperature measurements with around 0.5 degree F accuracy and 0.25 F
noise (averaging readings over 3 seconds). I am still characterizing the wind
speed/direction measurements, but they seem to agree quite well with a
cup/vane anemometer... basically if your temperature results are accurate
your wind measurements will be spot on. The nice thing about measuring
temperature via sonic measurements is that the measurements are unaffected by
solar heating of the apparatus... it does not need to be in the shade. Also
_______________________________________________
time-nuts mailing list -- [email protected]
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
