Anemometer

[pumpkinpi]

That one looks really good. I like the way it can be scaled (or calibrated) already. It is likely not that accurate, but probably close enough for my purposes.

Need to figure out what debouncing means.

pumpkin

[davidcoton]

"Debouncing" is a technique in hardware or software to make sure that contact bounce (on closing or more rarely opening) is not detected as multiple pulses. Search for it to find detailed methods.

[DougieLawson]

Hardware debouncing: http://raspi.tv/2014/rpi-gpio-update-an ... ling-edges
Software debouncing: http://raspi.tv/2013/how-to-use-interru ... pio-part-2

[pumpkinpi]

Awesome. Thanks folks. I will report back in ~5 months when I have this working. (yes, this is a hobby).

pumpkin

[pumpkinpi]

Well, my anemometer arrived in short order from InSpeed. It is very well constructed. Very impressed.

Have written a short python function (as part of my weather station program) that calculates wind speed.

The website says this:

Converting pulses to wind speed is simple using the following formula:
2.5 mph per Hz (1 Hz = 1 pulse/second)

So I converted to km/hr below and I think my math is ok.

I have done some basic desk counting over the 20 seconds that I am taking a measurement and it seems to count reasonably well.

I would be interested in some other ideas to do debouncing. I wanted to detect wind speeds up to 80 km/hr which would be about 20 Hz, so that is where I came up with that sleep timer of 0.05 seconds (1/20). Does this seem reasonable?

I used a For loop that seems to run for about 20 seconds given the little 0.01 sleep timer in there. While doing this, it is looking for switch activations.

Code: Select all

def get_WindSpeed():
    global T, wind_speed
    wind_speed = 0
    try:
        if T == 8:
            count = 0
            time_start = time.time()
            print ("Start wind speed check.")
            for x in range (1000):
                if GPIO.digitalRead(5) == 0:
                    webiopi.sleep(0.05)
                    if GPIO.digitalRead(5) == 1:
                        count += 1
                webiopi.sleep(0.01)
            time_stop = time.time()
            wind_speed = 4.02336 * (count / (time_stop - time_start))
        T +=1
        return wind_speed
    except Exception as e:
        print (e)
        print("Something messed up with wind script")
        pass

Thanks,
pumpkin

[pumpkinpi]

I have another question. The way I wired this up is to simply connect 3.3V to one side of the switch and the other side to GPIO pin 5. I am using webiopi which I like. I allows easy integration to a webpage. I have this line further up in my code to make sure the GPIO is in IN mode:

Code: Select all

GPIO.setFunction(5, GPIO.IN)

Do I need a resistor in the line? Don't want to damage my Pi.

Even when the anemometer is not turning, I get counts. Any idea where those are coming from? I could try hardware debouncing, but I prefer the soft approach.

pumpkin

[Massi]

I have another question. The way I wired this up is to simply connect 3.3V to one side of the switch and the other side to GPIO pin 5. I am using webiopi which I like. I allows easy integration to a webpage. I have this line further up in my code to make sure the GPIO is in IN mode:

Code: Select all

GPIO.setFunction(5, GPIO.IN)

Do I need a resistor in the line? Don't want to damage my Pi.

Even when the anemometer is not turning, I get counts. Any idea where those are coming from? I could try hardware debouncing, but I prefer the soft approach.

pumpkin

i think you need to add a pulldown resistor to the circuit, otherwise the GPIO is floating and you read unset levels.
With a pulldown, when the circuit is open you'll read LOW, when the circuit is closed you'll read HIGH since it will be connected to 3,3V

You can use the integrated pi pullup/down

And you can setup the circuit even in "opposite" mode: gnd to one side, gpio on the other and a pullup on the gpio.

In your case you'll want to sense a rising edge
in the second you'll want to sense a falling edge

Btw, software debouncing when managing interrupts is always well done

[Massi]

Sorry man, i gave a look to you code.
You are waiting for something to happen (a state change of the input gpio).
I don't think a for loop with a sleep inside is the right choice
Please give a look to interrupts managing on gpio.. (for example, the "software debouncing" page posted by Dougie)

[pumpkinpi]

Do I need to be concerned about lightning strikes? Do I need a fuse in the line to protect the Pi?

pumpkin

[pumpkinpi]

Sorry man, i gave a look to you code.
You are waiting for something to happen (a state change of the input gpio).
I don't think a for loop with a sleep inside is the right choice
Please give a look to interrupts managing on gpio.. (for example, the "software debouncing" page posted by Dougie)

I am using the loop with the sleep inside as a form of debouncing the circuit. I know the desired sensing frequency, so I used that as a means of determining how much debounce I could put in there. It seems to work, but maybe I am missing something.

I am not clear on how I would wire in the resistor. I will have to do some more research on that one.

pumpkin.

[davidcoton]

Do I need to be concerned about lightning strikes? Do I need a fuse in the line to protect the Pi?

pumpkin

If lightening strikes your weather station hardware, a fuse won't help. Your Pi and everything connected to it will likely be wrecked.

You can provide some protection with a lightening conductor as the highest part of the weather station. Connect it to earth with a heavy metal strap, as used for building protection. "Earth" in this context is a copper rod sunk into the ground. Keep the lightening connection well away from the RPi connections. Lightening may still cause serious damage. If you are in a high risk area, seek advice from a local protection expert. There are specialist lightening arrestors that protect power lines and phone lines, you may need to explore that route. (I have only been involved on the fringes of providing and inspecting protecting for a church building. There is a lot of conflicting advice around. But if you get a direct strike, your Pi will fry.)

Oh, and NEVER work on your weather station when lightening is anywhere near.

[pumpkinpi]

Oh, and NEVER work on your weather station when lightening is anywhere near.

I have no intention of winning a Darwin award.

http://www.darwinawards.com/

[davidcoton]

I am using the loop with the sleep inside as a form of debouncing the circuit. I know the desired sensing frequency, so I used that as a means of determining how much debounce I could put in there. It seems to work, but maybe I am missing something.

Try it, if it works, it's good. Really you need an oscilloscope to see what you are getting and whether your delay will fix it. If your highest frequency is 20Hz, the debounce period should probably be less than 25ms (less than half a cycle). I seem to remember from engineering days long ago that the debounce period was usually about 1ms, but I may be mistaken.

I am not clear on how I would wire in the resistor. I will have to do some more research on that one.

Use the Pi's internal pull up, activated in software (when your program initialises the hardware). Then connect the anemometer contact between the Pi input and ground. It's a safer way round than taking 3V3 out to the contact. No incoming series resistor is required. (This is what pattagghiu was suggesting above, +1 credit).

[Massi]

(This is what pattagghiu was suggesting above, +1 credit).

i'll make good use of that

[pumpkinpi]

After monkeying around a bit, I think I have a script that includes debounce and that works reliably. I think I need to take the anemometer for a drive to check calibration, but this gives me a reasonable idea of speed, which was the intent of this mini-project for me.

Code: Select all

def get_WindSpeed():
    global T, wind_speed
    wind_speed = 0
    try:
        if T == 8:
            count = 0
            time_start = time.time()
            print ("Start wind speed check.")
            all_ok = True
            while all_ok:
                if GPIO.digitalRead(5) == 1:
                    webiopi.sleep(0.01)
                    if GPIO.digitalRead(5) == 0:
                        count += 1
                if (time.time() - time_start) >= 15:
                    all_ok = False
            time_stop = time.time()
            wind_speed = 4.02336 * (count / (time_stop - time_start))
            print ("Wind speed: " + str(wind_speed) + " km/hr")
            if wind_speed < 1:
                wind_speed = 0
                print ("Wind speed: " + str(wind_speed) + " km/hr")
            print ("Time: " + str(time_stop - time_start) + " s")
            print ("Frequency: " + str(count/(time_stop - time_start)) + " Hz")
            print ("Count: " + str(count))
        T +=1
        return wind_speed
    except Exception as e:
        print (e)
        print("Something messed up with wind script")
        pass

I changed my circuit around so that the reed switch is between the ground and the GPIO #5. I have this line in the setup:

Code: Select all

GPIO.setFunction(5, GPIO.IN, GPIO.PUD_UP)

Thanks for all the help here.

pumpkin

[jbeale]

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

I think this could work well. How long it works outside will depend on how well the enclosure protects the opto detector and whatever bearing you use (simple pivot?) from the effects of rain / snow / dust.

[scruss]

Incidentally, the very nice (read: ££££) Vector Instruments anemometers use photochoppers. They are extremely carefully finished though. Considering they were originally designed to measure and model potentially lethal biohazard plumes from Porton Down, they kind of have to be sensitive.

[grompot]

You can use the sensirion SDP810 125 PA or 500 PA Flow sensors. It is extremly accurate and have no moving parts.

Check my project at https://www.instructables.com/id/Handhe ... r-Station/
Also check python code at https://github.com/JJSlabbert/Sensirion_SDP810