Anemometer

[squarepie]

I'm building a raspberry pi weather station. Does anyone know how to make an anemometer? The only premade ones I can find online are handheld ones which are pretty useless for my project. Also I am interfacing it through i2c and Dallas 1 wire.

[aTao]

All you need is to count the revolutions on just about any windmill. 4 plastic cups on 2 dowels fixed in a cross would be an extremely basic example.
At first "Google" I2C frequency counters seem rare, so you could make one, use I2C gpio expander to read a counter that is clocked by an encoder on the windmill shaft and reset every second.

[klricks]

Here are 2 ways:

1. Magnet and reed switch -. A magnet mounted to a shaft or cupset spins around the shaft. The reed switch is mounted so that the magnet passes over it and opens and closes the reed switch contacts as the cupset turns. Could be interfaced to the RPi with little or no additional electronics.

2. Chopper Wheel - Pulses are generated by shining a light source such as an LED through a slotted wheel mounted on the cupset shaft. A detector on the other side of the wheel detects the light / dark pulses as the wheel spins. Some additional circuits would be needed to run the detector and generate usable pulses.

[vk4tec]

Computer mouse

LED wheel and RX - older ones

[Burngate]

You could always try for a non-mechanical one - ultrasonic transmitter and three receivers, measure the delays to work out wind speed and direction

[aTao]

You could always try for a non-mechanical one - ultrasonic transmitter and three receivers, measure the delays to work out wind speed and direction

Nice!
If you also want to measure wind direction the ultrasonics could be mounted on a weather vane.

[VENTS]

Hi,
You can see that ultrasonic anemometer :
http://www.lcjcapteurs.com/?lang=en

[Burngate]

Hi,
You can see that ultrasonic anemometer :
http://www.lcjcapteurs.com/?lang=en

That's the one!

Actually it was a long time ago that I saw an article about it, so LCJ Capteurs may have commercialised it in the mean time.

Solid state, no moving parts, brilliant.

[klricks]

Hi,
You can see that ultrasonic anemometer :
http://www.lcjcapteurs.com/?lang=en

That's the one!

Actually it was a long time ago that I saw an article about it, so LCJ Capteurs may have commercialised it in the mean time.

Solid state, no moving parts, brilliant.

Sonic anemometers are not a new technology (several decades). Many companies make them including the company I work for. http://www.metone.com/
Most Sonic wind sensors can measure wind direction as well as wind speed. (By definition anemometer is wind speed only). Some even have temperature measurement. Sonic wind sensors are fairly complex and require digital processing electronics (CPU) embedded in the sensor.
Sonic wind sensors are expensive as well..... several hundred to well over 1K$ US.

http://en.wikipedia.org/wiki/Anemometer

[trevj]

I found what look to be some useful links. Apologies if they've already been noted elsewhere in this forum.

http://blog.makezine.com/2010/06/04/how ... ter-to-an/
http://www.thingiverse.com/thing:41367
http://www.yoctopuce.com/EN/article/how ... ind-part-1

[richardp]

http://www.maplin.co.uk/n25fr-wind-speed-sensor-676620

I can't build it for less than £2.50 - So I would buy one and hack it up.

[wstein25]

I've been messing around getting an RPI set up as a web server showing real-time wind data. I'm using a (relatively) inexpensive 3 cup anemometer with a reed switch that generates pulses - the rate dependent upon the wind speed. Instead of trying to count the pulses with the RPI directly, I have TI's MSP430 Launchpad (US$ 4.30) set up to do the pulse counting, and make the data available at its serial port.

This then is fed to the RPI serially, and saved as a single line file on the RPI. Just a bit of PHP code can put this on the web page that the RPI serves. I will get around eventually to writing this all up if anyone is interested. (A bit of shameless promotion - you can see the data at our facility here: http://174.63.120.250/

I have limited the updating to once every 10 seconds, as I had concerns about the reliability of the SD card memory. It doesn't push the data, you have to refresh the page to see current data.

[wstein25]

I have changed the Raspberry pi web server address to the alternate port 8080 i.e.
http://174.63.120.250:8080

[scruss]

Care to do a short writeup, wstein25? That's a decent solution to a complex problem, and pulse counting are what little µc boards are made for.

[wstein25]

I can't guarantee when I'll get around to writing this up. I'll do my best over the
next month. Perhaps I can post bits of it as I get it done. Do you have any particular
aspect of the project that you would like to see first? HW / SW , MSP430 code etc?

Bill

[scruss]

The MSP430 bit would be perfect, thanks. It's a nice cheap way to interface weather sensors, and might help to isolate the Raspberry Pi from the bad things that weather sensors have to deal with. No rush at all.

[pumpkinpi]

Has anybody tried interfacing this anemometer to the Pi?
http://moderndevice.com/product/wind-sensor/

I like that there are no moving parts. I am not clear on how to read the analog output with the Pi.

Thanks!

[scruss]

Has anybody tried interfacing this anemometer to the Pi?
http://moderndevice.com/product/wind-sensor/

Ooh, that's rather clever … a cheaper variant on the hot wire anemometer.

From the blog post attached to the bottom of the link, it seems it needs a very steady ≥ 5 V power supply, and then will output an analogue voltage between 0 and the supply voltage. So you'll need an A2D able to handle higher voltages.

The output is non-linear, and the sensor would be best in very low wind areas. I don't think it would last too long outside.

[pumpkinpi]

Not having used an analog to digital converter, what would work with the Pi?

The technology is well known. I have been buying this type of instrument for 15+ years for use in industrial plants to measure air flow. Why do you feel it would not last long outside? It just needs an appropriately designed enclosure. My feeling is that one with moving parts is more likely to fail.

pumpkin.

[scruss]

Not having used an analog to digital converter, what would work with the Pi?

Not had that much experience with ADC on the Raspberry Pi. Something like the Microchip MCP3424 might do, as used in the ADC Pi board. I don't know if you can configure it for differential inputs. You'll need a separate power supply for the sensor than the Raspberry Pi.

The technology is well known. I have been buying this type of instrument for 15+ years for use in industrial plants to measure air flow. Why do you feel it would not last long outside? It just needs an appropriately designed enclosure. My feeling is that one with moving parts is more likely to fail.

Maybe my comment came across as dismissive; I'm sorry if it did. My experience is in anemometry for wind power in northern climates: if it can crack, freeze, get struck by lightning, shot at, get gummed up with bird poop/nesting bees, invaded by ants or generally go south in amusing and expensive-to-replace ways, I've seen it happen. I guess if you enclosed it carefully (potted it?) and recalibrated it regularly, it would work well. Its non-linear output and upper limit of ~19 m/s are the main issues for me. You'd likely never see 19 m/s in a domestic location, but old habits die hard.

Mechanical units (admittedly, the expensive ones I've used, like from Vector Instruments, Thies and R. M. Young; tip o' the hat to the MetOne dude who posted further up the thread — most wind turbines use MetOne ultrasonics) last pretty well, even in Canada. Replace the bearings every two years, get 'em recalibrated, and you can take the numbers to the bank. We're mostly moving to remote sensing (LIDAR or SoDAR) and I'm frankly glad never to have to build a 110 metre met tower ever again …

[pumpkinpi]

Like the original poster, I am looking for a low cost solution (think $50 or less) that I can use at home just to satisfy my own curiosity. Accuracy is not a prime motivator, but the more accurate the better.

The ultrasonic technology looks pretty cool, but I suspect it will break my budget.

scruss, you obviously have some good experience here and you live around the corner from me in the GTA, so, what would you recommend?

I really wish there was an i2c solution. I have just started working with i2c instruments and they are dead easy to work with.

pumpkin

[pumpkinpi]

This is one option:
http://www.adafruit.com/products/1733

Looks good on screen.

Plus one of these to convert the voltage to something that can be understood by a Pi?:
http://www.adafruit.com/products/1083

pumpkin

[scruss]

This is one option:
http://www.adafruit.com/products/1733

Looks good on screen.

Hmm, not bad. Mil-spec connector means it's probably for boating. 0.4 to 2 V output looks suspiciously like it's a 4–20 mA current loop device internally, dropped through a 100 Ω resistor. I ♥ current loop devices, but they use lots of power …

Analogue output, though, and a surprisingly small range. Given that you won't see much over 4 m/s in a domestic setting, that means you have to get all your readings from a 400–600 mV range. Since the thing won't start until you have ½ m/s wind (bearing stiction; it'll measure down to 0.2 m/s slowing down from higher wind speeds → oh hai, hysteresis …) you're not going to get much of a range of outputs.

So yeah, as a cheapish spinny thing, I've seen worse. I don't know what the bearings are like, though. It'll probably ice up like a good 'un, though the colour helps.

For really cheap, those simple add-on anemometers from home weather stations will do something. All they have is a magnet on the shaft and a reed switch that closes once every revolution. Poor resolution, but easy to read as a slow pulse count value (as long as you remember to debounce the circuit). It's the same way that bicycle computers read your speed from a spoke-mounted magnet — and it allowed the ingenious (and sadly late) Gordon Proven to build really cheap wind run loggers using a pole, a bicycle computer and a reed-switch anemometer.

[boyoh]

I'm building a raspberry pi weather station. Does anyone know how to make an anemometer? The only premade ones I can find online are handheld ones which are pretty useless for my project. Also I am interfacing it through i2c and Dallas 1 wire.

One option for the wind speed indicator
is to use a slotted disc spinning through
slotted Opto Isolator , very little mechanical
drag on the shaft , A noise free clean pulse
to feed in to the GPIO in/put to be counted
and calibrated

[mithras]

There is a robut but simple low-cost anemometer here - http://www.audon.co.uk/weather_sensors/vortex.html that has a reed relay contact that closes 1 per mph and a matching weather vane - http://www.audon.co.uk/weather_sensors/evane.html with a 0-5V output. Both should be easy enough to connect to the GPIO pins.