The design
 |
| Source |
The above picture is
from the 1-wire design which is a common picture, I used this to get the
basic design principle. Mine is constructed from tin from an old juice
concentrate can. The design was hand drawn to get the sizing that I
wanted and I could see the life size thing.
The enclosure is made with ABS plastic sheet that I had lying around, but anything that can be cut and glued should work.
The
design of the catchment area of the enclosure is very important as that
plays a role in what the amount of rain is caught and the amount per
tip. What is needed is a catchment area of $100cm ^ 2$ this equates to
$10mm$ of rain being caught by the system, but because this is 100 times
smaller, the $10mm = 1mm$ per $1m ^ 2$. Then the trick is to calibrate
the sensor to tip at a specific point, this can be between .01 to 1mm of
rain, but this is to the needs of the area. Mine by default ended up by
being $0.25 mm$ per tip, this is nice as I can pick up the sometimes
common drizzles where I live.
I was originally going to use the reed switch and magnet method as this uses less power to run the system, but the little glass tubes broke every time I worked with them, I went with a reflective IR sensor after that and ensured its current is 10$mA$ at 3.3 V.
Next we will attach the side pieces to the main bucket. This is done by soldering with a small blow torch.
The next part was to add the supports for the bucket
and the sensor. I again used ABS sheet and hot glue. The actual
dimensions were cut to fit the inside because of the funnel.
From the first pic you can see the threshold adjustment screw that was placed about 33mm away from centre.
The signal is $Active_{HIGH}$ which keeps the current consumption down. Also I used a 330$\Omega$ resistor on the LED at 3.3$V$ as there is no need for a bright IR signal as it is <10mm away from the reflector. The signal is then connected to a Digital pin. I used a comparator to 'Level Shift' the 3.3V signal for the Arduino UNO to understand, this is as its max voltage varies between 1-1.5$V$.
$R5$
and $R4$ are chosen to give a voltage just below the active voltage,
about 0.4V, with a final part being really weird that I only used a
pull-up resistor of 1$k\Omega$ to get the circuit to work.
I was originally going to use the reed switch and magnet method as this uses less power to run the system, but the little glass tubes broke every time I worked with them, I went with a reflective IR sensor after that and ensured its current is 10$mA$ at 3.3 V.
BUILD
 | |||
| The tin |
Start by cutting off the
base that is left on the bottom, then cut up the join mark of the tin,
this gives one a straight line to follow when cutting. Next you will
trim the top and bottom of the tin so that you can open it flat (the
curly bit). Next is to mark out the dimensions on the flat piece.
Next we cut and fold the strip of tin to shape, I used a steel ruler to give me a square edge when bending the tin.
Next you cut the sides of the bucket, this is done using the measurement across the base of 110mm, then cut two of these.
- Rough the sides along which you will be soldering
- Apply some flux
- Tin the the strips along which you roughed
- Place flat and attach one side ONLY
Then
what is needed is to drill the hole for the axle $3.2mm$, this is $5mm$
from the base, You will drill the side that is attached. Then you will
place the other side behind that and aligning the two mark the hole on
the unconnected piece. Then drill and attach it using the above method.
 |
| The drill bit was a bit small, so I used the copper shaft, and a hammer and opened the hole up so it moved freely |
 |
| The finished bucket |
The Enclosure
The enclosure for
the bucket was made using ABS plastic sheet. The internal dimension of
the rectangle is $120mm * 80mm$, this gives a catchment area of the
$100cm ^ 2$. The height of the enclosure is 100mm, and the funnel is
placed 20mm down from the total height, this gives some free board if
there is a lot of rain.
When
cutting $Side l_1$ and $Side l_2$ ensure to add $6mm$ on to the
measurement so that the internal measurement stays $120 * 80$
 |
| Some Hot Glue |
 |
| Bucket inside with space to spare |
Sensor
Next is to attach the sensor components. This is done using a IR LED at $940nm$ and a matched Phototransistor. The reflector is attached to the bucket at centre, this is just a strip of the tin.The signal is $Active_{HIGH}$ which keeps the current consumption down. Also I used a 330$\Omega$ resistor on the LED at 3.3$V$ as there is no need for a bright IR signal as it is <10mm away from the reflector. The signal is then connected to a Digital pin. I used a comparator to 'Level Shift' the 3.3V signal for the Arduino UNO to understand, this is as its max voltage varies between 1-1.5$V$.
 |
| Reflector placed centre |
Catchment area
The catchment part
of the tipping bucket rain gauge was not built off a drawing/design, so it was more of a
process of elimination of fitment and seeing where how it would fit. I
used one tin to make it, the centre square piece is $30 x 15mm$ this is
from centre. The hole was drilled to a 3.2mm diameter and punched again
so as to create a lip downwards for water flow. From the corners of the
smaller rectangle lines were drawn outwards and then cut and soldered.
The
whole thing is hot glued 20mm down from the top. In the third image you
can see that the surface is ruffed for the application of white Hammerite. This was also applied to all metal parts after being ruffed.
The function $tip()$ is something I wrote for another program which became useful in this code, it just checks to see if the variable that you put in. When it is only working with logical 1 and 0, and you place it in an if() statement looking for a logic 1, it only ever triggers on a 0-1 transition. One thing though with this was that I had to set "int previous = 1; //no false reading on boot" because as the comment says, I was getting false readings from the code on the first time the thing ran
The code originally used interrupts, but with one tip it triggered about 100+ counts and gave some seriously weird results. So I just made it loop in the code to check for the change, and because the code never opens serial and also looks for the log button, its pretty much immediate. I do believe though that I will probably add the interrupt functionality for the rain gauge and the anemometer.
When the log button is pressed it stores the $count$ variable into eeprom and flashes th onboard LED, $state$ also makes it so you can only do this once until you open serial and read the value. This was done because when it was setup away from the PC and then powered on and off, data was lost or when it was connected to the PC and the terminal was opened, data was lost because of restarting :(
Test Run
I tested the rain gauge
for a day or two when it was raining and it was actually very accurate. I
left it for a while after it rained to look at it again and saw some
flash rust on some areas which must have been exposed, this was before
painting, this was odd as the tin had some sealant over it. It also come
up on the shaft which was a bummer. This was solved by partial sanding and spraying with silicone spray. Also after the tipping bucket was painted, it was also sprayed with the silicone and the catchment area. This was done to help the water flow better because it was pooling in some areas and causing it to fail.
Some pictures
~Code~
The code is written for an Arduino for testing so as to have access to the serial functionality for debugging and retrieval of the logged rain value. Pin 3 is the button with a pull-up, Pin 2 is the tipping bucket signal.
/* AWS - Rain Gauge test code
#Logs the rain fall to eeprom when a push button is pressed #This version works nicely with no interrupts because of the #function checking the button state for a change and because #the bucket does not move at lightning speed By Reece May Blog: https://www.bariumelectronics.blogspot.com Written: 3/30/2015 */ #include <EEPROM.h> //Variables int count = 0; float mm_const = 0.25; //calibrated mm of rain for a mm/m^2/tip float mm_rain = 0.00; int state = 0; int button_state = 0; void setup(){ pinMode(3, INPUT_PULLUP); pinMode(2, INPUT); pinMode(13, OUTPUT); // DDRB |= (1 << 1); // PORTB |= (1 << 1); Serial.begin(9600); Serial.println("Rain Gauge Test Code"); } int previous = 1; //no false reading on boot int tip(int to_test){ int current = to_test; if(current != previous){ previous = current; return current; } else return 0; } void loop(){ // digitalRead(3);//(PINB & (1 << 0)); button_state = digitalRead(2); delay(10); if(tip(button_state)){ count += 1; // Serial.println(count); } else count = count; //Function to see if log button is pushed and stores tip count once if(digitalRead(3) == 0 && !state){ digitalWrite(13, 1); delay(50); EEPROM.write(0, count); //writes count as no need for float storage delay(50); digitalWrite(13, 0); state = 1; //run only once, this is 0 when Serial is called } if(Serial.available()){ if(Serial.read() == 'k'){ byte rea = EEPROM.read(0); mm_rain = (rea) * mm_const; delay(100); Serial.print("count: "); Serial.println(rea); Serial.print("rain: "); Serial.println(mm_rain); state = 0; } else if(Serial.read() == 'c'){ count = 0; mm_rain = 0; Serial.println(count); Serial.println(mm_rain); } } }
The function $tip()$ is something I wrote for another program which became useful in this code, it just checks to see if the variable that you put in. When it is only working with logical 1 and 0, and you place it in an if() statement looking for a logic 1, it only ever triggers on a 0-1 transition. One thing though with this was that I had to set "int previous = 1; //no false reading on boot" because as the comment says, I was getting false readings from the code on the first time the thing ran
The code originally used interrupts, but with one tip it triggered about 100+ counts and gave some seriously weird results. So I just made it loop in the code to check for the change, and because the code never opens serial and also looks for the log button, its pretty much immediate. I do believe though that I will probably add the interrupt functionality for the rain gauge and the anemometer.
When the log button is pressed it stores the $count$ variable into eeprom and flashes th onboard LED, $state$ also makes it so you can only do this once until you open serial and read the value. This was done because when it was setup away from the PC and then powered on and off, data was lost or when it was connected to the PC and the terminal was opened, data was lost because of restarting :(
Results
The results of the rain gauge were good. I compared the values with a normal rain gauge. The system was tested over about three days, the tipping bucket was within about $0.5mm$ at most from the reading on the normal rain gauge. This also picked up about a $0.75mm$ drizzle. The readings are very accurate and I believe I have made a decent design of this and it looks nice.
The op-amp circuit and the tipping bucket combined draw a total of $9.1mA$ at $3.3V$, this I think is reasonable and should make it excellent for battery operation.
No comments:
Post a Comment