I am looking to create a small wind vane for measuring the wind direction on my arduino. I have been searching around a bit and found a few possibilities. Some electrical potentiometers (usually with a small dead-zone) and some sensors sensing a magnetic field over an IC. I would like it to be as small as possible, hopefully also quite cheap. Since it should be small, a low torque solution would be preferred. I am not looking for a full wind vane, but rather just the sensor to create my own.
So I was wondering what kind of experience you guys have with such sensors and what you would suggest?
Interesting that you brought this up, as I am seeking a similar sensor. However, it has to be extremely light, and ideally have no moving parts. What I am thinking of is some kind of differential pressure technique using a few pressure sensors arrayed inside of a cylinder. Think of a model yacht mast.
I had an earlier idea that used a computer mouse quadrature encoder at the base of a typical vane, and that has some promise, but it is more easily damaged and wants a little more hardware up top. I haven't completely ruled it out, though.
But how sensitive and tiny can I get air pressure sensors, and how much hardware do I need to get good data out? Has anyone tried this?
A few months ago there was a program on the TV of a device that was a circle of simple temp sensor with a small heating element in the middle. If the wind blew over it one (or two) of the sensors got slightly warmer. The difference in temperature gave a good indication of the wind direction. MInd you no moving parts! You could build one yourself with a number of NTC's or so?
Similar you could make a circle of hall effect sensors and use a vane with a tiny magnet, or use the magnet itself as vane.
robtillaart:
A few months ago there was a program on the TV of a device that was a circle of simple temp sensor with a small heating element in the middle. If the wind blew over it one (or two) of the sensors got slightly warmer.
Optical daisywheels can easily be printed on the pc, Using successive rings with half as many spokes as the prior ring you can make an optical encoder that will give you 360 degree read in binary format. The more rings the higher precision. an alternate approach is also to phase shift each ring and XOR each ring against a reference ring, once again giving you a binary numerical output. EG ring 1 had 720 spokes no phase inversions, this would be the reference ring, ring 2 would represent the MSB and would have 2 phase inversions, the next ring 4, so on and so forth until the last ring would invert every other spoke. I have some optical encoder wheels already for printing, I made them for speed sensor pickups.
ajofscott:
Optical daisywheels can easily be printed on the pc, Using successive rings with half as many spokes as the prior ring you can make an optical encoder that will give you 360 degree read in binary format. The more rings the higher precision. an alternate approach is also to phase shift each ring and XOR each ring against a reference ring, once again giving you a binary numerical output. EG ring 1 had 720 spokes no phase inversions, this would be the reference ring, ring 2 would represent the MSB and would have 2 phase inversions, the next ring 4, so on and so forth until the last ring would invert every other spoke. I have some optical encoder wheels already for printing, I made them for speed sensor pickups.
Such encoder wheels (and more importantly the sensor device itself) can become tricky to build past a certain number of bits, depending on how small you want the device.
A simpler approach (which may or may not be as accurate) would be to print a gradiated radial circle wheel (going from clear to black), with a slotted phototransistor sensor reading the light level (one could also print onto a white surface and use a reflectance sensor). The accuracy would mainly depend on the sensitivity of the sensor to changes in density of the disc pattern...
I recall reading of one method that used four ultrasonic sensor/emitters placed 90 degrees apart and facing each others opposite member, that could determine both wind speed and direction just by measuring time of flight of the four paths and crunching the calculations. So no moving parts and two key weather measurements, simple interfacing, and arduino friendly. Sounds like an interesting project.
However as Bob Dylan sung " you don't need a weatherman to know which way the wind blows"
I recall reading of one method that used four ultrasonic sensor/emitters placed 90 degrees apart and facing each others opposite member, that could determine both wind speed and direction just by measuring time of flight of the four paths and crunching the calculations.
Any idea how the math looks like? I assume sound travels faster in the direction of the wind instead of against it. Do you have any formulas for that?
I recall some thoughts about the effect of humidity & temperature on the speed of Sound.
Couldn't you use the Modern Device approach to put a couple of RTDs on a compass rose with tunnels to determine the vector and the speed of the wind? You'd need a custom PCB...
I recall reading of one method that used four ultrasonic sensor/emitters placed 90 degrees apart and facing each others opposite member, that could determine both wind speed and direction just by measuring time of flight of the four paths and crunching the calculations.
Any idea how the math looks like? I assume sound travels faster in the direction of the wind instead of against it. Do you have any formulas for that?
I recall some thoughts about the effect of humidity & temperature on the speed of Sound.
For a bit more money, the inspeed directional vane looks really tasty. A sealed hall effect sensor, no deadband, a 5VDC input, and a 0.25-4.75V output with 12 bits of resolution. I'd oversample the sucker, decimate, and enjoy that 0.25 degree resolution... see the data sheet at inspeed.com.
Rob, thanks for sharing that link. What an amazing array of instruments, sensor systems, etc. this gentleman has built. Some day, I may begin to understand 1/2 of what this gentlemen does in his spare time. Using a laser to characterize the quantity of precipitation is pretty amazing. Distinguishes among the many different types of precipitation is over the top! Certainly inspiring. Some day when the kids demand less attention perhaps I'll have time...
maybe late :P, but the older systems in windturbines use a quite simple system. You take a small lightsensor that is stationary, place it under a disk, with a slit cut in it, that starts at its widest at 0° and gradually gets smaller untill fully closed when you reach +-360°again. Install another stationery LED oposite to the lightsensor, for the lightbeam. So the lightsensor will measure more light at 0° or a lower voltage(depending on the senor), and the voltage will change as the gap in the disk gets smaller untill no light can pass. Ofcourse you will need to put a closed cylinder around it so no foreign light can hit the sensor. Now just use the voltage difference of the lightsensor and translate in winddirections.
Indeed it needs some good finetuning. And the gap needs to be cut quite precise. They use the same principle for windspeed, a disk with holes in it and a IR sensor. Count the amount of pulses in a second, and translate it into speed. Ofcourse this system needs to be calibrated as well, but is not that difficult. Measure the circumference, and count the amount of holes (mostely little squares). For example 20 holes is 1 tour, and 1 tour is 5 cm. So 200 pulses a sec. is 0.5 m/s. I think the most difficult is the programming( for me at least a newbie with arduino ).
).