Measuring Rain Fall

May 9th, 2016 • Electronics, ESP8266, Software Development1 Comment »

A new day and a new sensor to look at, the pluviometer.

The pluviometer is a mechanical sensor that looks a little like a see-saw. A small plastic bucket is placed on each side of the fulcrum. The buckets are designed so that when a critical mass of water is in one bucket it tips the see-saw and the bucket on the other side starts filling. The process continues as each bucket is filled and emptied. A picture paints a thousand words, so here is a view of the inside of the pluviometer:

Inside the Pluviometer

Inside the Pluviometer

A small magnet is placed on the see-saw and a reed switch is behind the fulcrum. The tipping motion triggers the reed switch each time the see-saw tips, this in turn generates a single pulse.

Reading the switch becomes a seeming simple repetition of the wind speed problem, namely debouncing a switch and attaching an interrupt.

One long term goal has to take this solution off grid. This will make power consumption a critical factor in the design. Attaching an interrupt does not necessarily become an attractive option as the Oak would be running continuously and with the WiFi running this would consume a fair amount of power.

According to the specification for the pluviometer, a pulse will be generated for every 0.2794mm of rainfall.

Offline Counting

One possible solution to the power problem would be to put the Oak to sleep and wake it up every say 5-15 minutes to take measurements and upload them to the cloud. Doing this would reduce power but would also mean that no measurements would be take during the sleep period if interrupts were used.

A cheap solution in terms of cost is to use the 74HC4040 counter. This could be put into a circuit and kept active while the Oak is sleeping. The output from the debounced switch would then be used as a clock signal for the 74HC4040. This would allow the pulses from the pluviometer to be counted while the Oak is sleeping.

The downside is that eight pins are needed to read the output from the counter. With only a small number of pins, many of which have already been used, this will require some way of adding extra pins to the system. Fortunately there are a number of digital IO expanders on the market. A common series of chips is the MCP23x17 chips. These add an additional 16 inputs/output pins with communication to the chip vis I2C (X=0) or SPI (X=S). The I2S variation will fit the bill nicely.

One final connection that is required for the counter is a reset connection. Left to itself, the rain gauge counter would continue to count pulses until the counter overflowed and started from zero once more. From a design perspective there are a few options:

  1. Reset the counter each time it has been read
  2. Reset the counter once a day
  3. Let the counter overflow and detect the reset to zero

One consideration will be the amount of rainfall the system can detect before it resets to zero.

Maximum number of counts = 212

Which gives 4096 counts. However, for simplicity we will only consider the lower 8 bits, i.e. 0-255 counts. The system can always be expanded later should this be necessary (or if the system has the capacity).

Maximum rainfall = 255 * 0.2794 mm

Giving a maximum rainfall count of 71.247 mm per counting period. It is envisaged that the counting period would be somewhere in the region of 5-15 minutes. This would make an hourly average of 285mm assuming a steady rainfall and a 15-minute interval between counts.

This should be well within expectations for UK weather.

As an aside, the full 12-bits would allow for a rainfall of 1144mm per counting interval, or 4.5 metres of rain per hour.


With a little supporting hardware to be added, the schematic looks as follows:

Weather Station Partial Schematic

Weather Station Partial Schematic

Note that the BME280 and TSL2561 are also on the schematic.

The rain gauge counter bit values have been labeled RXQx, RG = Rain Gauge and Q is the standard notation for a bit in a logic design.


The MCP23017 (I2C version of the IO expander) is sold by Adafruit. They have a standard Ardunio library for this component and so in the interest of code re-use, expediency and idleness this will be used.

So the first thing to do is to create an instance of the MCP23017 IO expander:

Adafruit_MCP23017 _outputExpander;      // MCP23017, I2C controlled 16-port output.

Next up, the output expander needs to be setup. Looking at the schematic, the output from the counter is mapped to port B on the MCP23017. This maps to IO pins 8 to 15 inclusive. These bits should be set to inputs. One final piece of configuration is the reset pin, this needs to be set to output:

//  Set up the pluviometer, zero the counts and then attached and initialise the
//  counter to the output expander.
void WeatherSensors::SetupRainfallSensor()
    _pluviometerPulseCount = 0;
    _pluviometerPulseCountToday = 0;
    //  Attach the rainfall counter to the output expander.
    int index;
    for (index = 8; index < 16; index++)
        _outputExpander.pinMode(index, INPUT);
    //  Attach the counter reset pin to the output expander and reset the counter.
    _outputExpander.pinMode(PIN_RAINFALL_RESET, OUTPUT);

As well as setting up the pulse counter, the above code resets internal counters for rainfall today and also resets the pulse count.

Reading the pulse counter is a simple matter to reading the state of each bit starting at 0:

//  Read the counter from the Rainfall sensor then reset the count.
void WeatherSensors::ReadRainfallSensor()
    byte mask = 1;

    _pluviometerPulseCount = 0;
    for (int index = 8; index < 15; index++)
        if (_outputExpander.digitalRead(index) == 1)
            _pluviometerPulseCount |= mask;
        mask <<= 1;
    _pluviometerPulseCountToday += _pluviometerPulseCount;


The eagle-eyed amongst you will have spotted a small flaw in the above logic, namely that the pulse counter could increment whilst the system is reading the values. It turns out that this matters a great deal in some cases. Consider the following:

The pulse counter has counter 15 pulses (0b00001111). The software is processing these bit values and has processed bit 2, counter value = 7. A pulse is registered by the counter and the pulse count becomes 16 (0b00010000). The system processes bit 3 and finds it is 0 and then moves on to bit 4 and finds this is 1, i.e. 16. The counter then becomes 16 + 7 = 23 instead of 15. A little bit difficult to see, so lets put this in tabular form:

Bit Value Decimal Value Cumulative Sum Counter
0 1 1 1 00001111
1 1 2 3 00001111
2 1 4 7 00001111
Counter increments here 00010000
3 0 0 7 00010000
4 1 16 23 00010000
5 0 0 23 00010000
6 0 0 23 00010000
7 0 0 23 00010000

The counter column contains the value in the 74HC4040 counter chip (lower 8 bits only).

The problem occurs because the reading of the pulses is not instantaneous, the software takes longer to process the data than the counter takes to register a clock pulse and update the counter.

There are two possible solutions to this, both left for later investigation:

  1. Add an AND gate between the incoming rain gauge pulse and the 74HC4040 clock pin. This can be turned off when taking a reading.
  2. Check the documentation for the MCP23017 to see if the values can be latched and then read. Latching the values means that they cannot be changed; once latched and the values being read would be consistent.

Both offer a solution to the problem and will be left as a later refinement.


The pluviometer is a seemingly simple switch but it needs to run constantly. This presents a challenge when considering the requirement to save power where possible. This article presented a possible solution using a small number of components bringing powered on permanently.

The solution does introduce a possible error in the reading, although this should be reasonably small. A full analysis will be presented in a future article.

Weather Station Proof of Concept Software

April 3rd, 2016 • Electronics, ESP8266, Internet of Things, Software Development1 Comment »

The weather station project is still at the proof of concept stage but the last few articles have run through the concepts for connecting several sensors to the Digistump Oak microcontroller. In this article we will look at the basic (and I stress basic) software required to connect the sensors discussed so far with the microcontroller and start to collect data.

Hardware – So Far

In the previous articles the following sensors have been discussed:

These sensors are currently connected to a piece of breadboard along with the Oak.

Now we have the sensors connected we need to add some software goodness.

Software Requirements

The first and most obvious requirement is to be able to collect data from the above sensors. In addition the proof of concept software should also permit the following:

  1. Test logging data to the cloud
  2. Serial debugging
  3. Setting the system time from the Internet

Data logging to the cloud will initially be to the Sparkfun Data Service as this is a simple enough service to use. One of the first things to do is to create an account / data stream in order to permit this.

Development Environment

The Arduino development environment can be used to program the Oak and it has the advantage that it is available across multiple environments. Digistump recommend using version 1.6.5 of the Arduino environment as there are known issues with more current versions.


In previous articles it was noted that Sparkfun and Adafruit provide libraries for two of the I2C sensor boards being used (BME280 and TSL2561). An additional library is also required to support the third objective, setting the time from the Internet.

The additional libraries are installed from the Sketch -> Include Library -> Manage Libraries.. dialog. Open this dialog and install the following libraries:

  • Adafruit Unified Sensor (1.0.2)
  • Adafruit BME280 (1.0.3)
  • Sparkfun TSL2561 (1.1.0)
  • NtpClientLib by German Martin (1.3.0)

The version numbers are the ones available at the time of writing.

Software Walk Through

At this point the hardware should be in place and all of the necessary libraries installed. Let the coding begin.

This first thing we need is the includes for the libraries that are going to be used:

#include <ESP8266WiFi.h>
#include <ESP8266HTTPClient.h>
#include <Time.h>
#include <NtpClientLib.h>
#include <SparkFunTSL2561.h>
#include <Adafruit_Sensor.h>
#include <Adafruit_BME280.h>
#include <Wire.h>
#include <SPI.h>

The SPI library is not going to be used in this example but will be required in future articles.

Next up we need some definitions and global variables to support the sensors:

//  Definitions used in the code for pins etc.
#define VERSION           "0.04"
#define PIN_SLEEP_OR_RUN  7
#define PIN_ONBOARD_LED   1
#define PIN_ANEMOMETER    5
#define SLEEP_PERIOD      60

//  Light sensor (luminosity).
SFE_TSL2561 light;

//  Create a Temperature, humidity and pressure sensor.
Adafruit_BME280 bme;
float temperature;
float pressure;
float humidity;

//  TLS2561 related globals.
boolean gain;       //  Gain setting, 0 = X1, 1 = X16;
unsigned int ms;    //  Integration ("shutter") time in milliseconds
double lux;         //  Luminosity in lux.
boolean good;       //  True if neither sensor is saturated

//  Ultraviolet analog reading.
int ultraviolet;
#define UV_GRADIENT           0.12
#define REFERENCE_VOLTAGE     3.3
#define UV_OFFSET             1.025

//  Buffer for messages.
char buffer[256];
char number[20];

//  NTP class to provide system time.
ntpClient *ntp;

//  Wind Speed sensor, each pulse per second represents 1.492 miles per hour.
volatile int windSpeedPulseCount;
#define WINDSPEED_PER_PULSE     1.492

//  We are logging to Phant and we need somewhere to store the client and keys.
#define PHANT_DOMAIN        ""
#define PHANT_PAGE          "/input/---- Your stream ID goes here ----"
const int phantPort = 80;
#define PHANT_PRIVATE_KEY   "---- Your Private Key goes here ----"

Before we progress much further it should be acknowledged that some of the code for the BMS280 and the TSL2561 is modified from the Adafruit and Sparkfun example applications.

A key point to note from the above code is the definition of the windSpeedPulseCount variable. Note the use of the volatile keyword. This tells the compiler not to optimise the use of this variable.

Next up is some support code. Two methods are initially required, ftoa adds the limited ability to convert a floating point number to a char * for debugging. The second method outputs a debugging message. This has been abstracted to allow for possible network debug messages later in the project. At the moment the serial port will be used.

//  Output a diagnostic message if debugging is turned on.
void DebugMessage(String message)

//  Convert a float to a string for debugging.
char *ftoa(char *a, double f, int precision)
    long p[] = {0, 10, 100, 1000, 10000, 100000, 1000000, 10000000, 100000000};
    char *ret = a;
    long integer = (long) f;
    itoa(integer, a, 10);
    while (*a != '\0')
    if (precision != 0)
      *a++ = '.';
      long decimal = abs((long) ((f - integer) * p[precision]));
      itoa(decimal, a, 10);
    return ret;

TSL2561 – Luminosity Sensor

The next group of methods deal with the luminosity sensor, the TSL2561. these methods are slightly modified versions of the example code from Sparkfun. It is envisaged that future versions of these methods will deal with saturation and low light levels by dynamically changing the way the sensor works. For the proof of concept the basic code should suffice:

//  Convert the error from the TSL2561 into an error that a human can understand.
void PrintLuminosityError(byte error)
    switch (error)
        case 0:
            DebugMessage("TSL2561 Error: success");
        case 1:
            DebugMessage("TSL2561 Error: data too long for transmit buffer");
        case 2:
            DebugMessage("TSL2561 Error: received NACK on address (disconnected?)");
        case 3:
            DebugMessage("TSL2561 Error: received NACK on data");
        case 4:
            DebugMessage("TSL2561 Error: other error");
            DebugMessage("TSL2561 Error: unknown error");

//  Set up the luminsoity sensor.
void SetupLuminositySensor()

    // Get factory ID from sensor:
    // (Just for fun, you don't need to do this to operate the sensor)
    unsigned char id;
    if (light.getID(id))
        sprintf(buffer, "Retrieved TSL2561 device ID: 0x%x", id);
        byte error = light.getError();
    // The light sensor has a default integration time of 402ms,
    // and a default gain of low (1X).
    // If you would like to change either of these, you can
    // do so using the setTiming() command.
    // If gain = false (0), device is set to low gain (1X)
    // If gain = high (1), device is set to high gain (16X)  
    gain = 0;
    // If time = 0, integration will be 13.7ms
    // If time = 1, integration will be 101ms
    // If time = 2, integration will be 402ms
    // If time = 3, use manual start / stop to perform your own integration
    unsigned char time = 2;
    // setTiming() will set the third parameter (ms) to the
    // requested integration time in ms (this will be useful later):
    light.setTiming(gain, time, ms);
    // To start taking measurements, power up the sensor:
    DebugMessage((char *) "Powering up the luminosity sensor.");

//  Read the luminosity from the TSL2561 luminosity sensor.
void ReadLuminositySensor()
    unsigned int data0, data1;
    if (light.getData(data0, data1))
        sprintf(buffer, "TSL2561 data: 0x%04x, 0x%04x", data0, data1);
        //  To calculate lux, pass all your settings and readings to the getLux() function.
        //  The getLux() function will return 1 if the calculation was successful, or 0 if one or both of the sensors was
        //  saturated (too much light). If this happens, you can reduce the integration time and/or gain.
        //  For more information see the hookup guide at: 

        // Perform lux calculation.
        double localLux;
        good = light.getLux(gain ,ms, data0, data1, localLux);
        if (good)
            lux = localLux;
        byte error = light.getError();

//  Log the luminosity data to the debug stream.
void LogLuminosityData()
    sprintf(buffer, "Lux: %s", ftoa(number, lux, 2));

BME280 – Air Temperature, Pressure and Humidity Sensor

Dealing with this sensor is simpler than the luminosity sensor as can be seen from the code below:

//  Setup the Adafruit BME280 Temperature, pressure and humidity sensor.
void SetupTemperaturePressureSensor()
    if (!bme.begin())
        DebugMessage("Could not find a valid BME280 sensor, check wiring!");
        DebugMessage("BME280 sensor located on I2C bus.");

//  Log the data from the temperature, pressure and humidity sensor.
void LogTemperaturePressureData()
    sprintf(buffer, "Temperature: %s C", ftoa(number, temperature, 2));
    sprintf(buffer, "Humidity: %s %%", ftoa(number, humidity, 2));
    sprintf(buffer, "Pressure: %s hPa", ftoa(number, pressure / 100, 0));

//  Read the data from the Temperature, pressure and humidity sensor.
void ReadTemperaturePressureSensor()
    temperature = bme.readTemperature();
    pressure = bme.readPressure();
    humidity = bme.readHumidity();

This group of methods follows a similar line to the TSL2561 sensor methods, namely, setup, read and log methods.

ML8511 – Ultraviolet Light Sensor

This sensor is the simplest so far as it provides a simple analog value representing the intensity of the ultraviolet light falling on the sensor.

//  Read the ultraviolet light sensor.
void ReadUltravioletSensor()
    ultraviolet = analogRead(A0);

//  Log the reading from the ultraviolet light sensor.
void LogUltravioletData()
    char num[20];
    sprintf(buffer, "Ultraviolet light: %s", itoa(ultraviolet, num, 10));

Note that this sensor does allow for an enable line. This can be used to disable the sensor and put it into a low power mode if necessary. This will not be used in the proof of concept will be considered when the project moves to the point where power supplies / solar power is considered.

Wind Speed Sensor Interrupt Service Routine (ISR)

This method looks trivial and it is:

//  ISR to count the number of pulses from the anemometer (wind speed sensor).
void IncreaseWindSpeedPulseCount()

The non-trivial thing to remember about this code is that the method changes a global variable. The volatile keyword used in the variable definition is necessary to stop the compiler from optimising the global variable as this can have side effects.

Data Logging to Sparkfun’s Data Service

Assuming that we have a WiFi connection then we can log the data collected to the Sparkfun data service (this is based upon Phant).

//  Post the data to the Sparkfun web site.
void PostDataToPhant()
    String url = PHANT_PAGE "?private_key=" PHANT_PRIVATE_KEY "&airpressure=";
    url += ftoa(number, pressure / 100, 0);
    url += "&groundmoisture=0";
    url += "&groundtemperature=0";
    url += "&temperature=";
    url += ftoa(number, temperature, 2);
    url += "&humidity=";
    url += ftoa(number, humidity, 2);
    url += "&luminosity=";
    url += ftoa(number, lux, 2);
    url += "&rainfall=0";
    double uvStrength = (((double) ultraviolet) / MAXIMUM_ANALOG_VALUE) * REFERENCE_VOLTAGE;
    if (uvStrength < UV_OFFSET)
        uvStrength = 0;
        uvStrength = (uvStrength - UV_OFFSET) / UV_GRADIENT;
    url += "&ultravioletlight=";
    url += ftoa(number, uvStrength, 2);
    url += "&winddirection=0";
    double windSpeed = windSpeedPulseCount / WINDSPEED_DURATION;
    windSpeed *= WINDSPEED_PER_PULSE;
    url += "&windspeed=";
    url += ftoa(number, windSpeed, 2);
    //  Send the data to Phant (Sparkfun's data logging service).
    HTTPClient http;
    http.begin(PHANT_DOMAIN, phantPort, url);
    int httpCode = http.GET();
    sprintf(buffer, "Status code: %d", httpCode);
    String response = http.getString();
    sprintf(buffer, "Phant response code: %c", response[3]);
    if (response[3] != '1')
        //  Need to put some error handling here.

The stream has been set up to collect more data than is currently collected, for instance, ground temperature. Any parameter not measured at the moment is set to 0.

Some other things to consider following the proof of concept:

  1. Error handling for network issues
  2. Possible offline collection of data
  3. Using a local version of Phant

Something to bear in mind after the project moved from proof of concept.

Data Collection

The majority of the sensor have a read method to collect the data from the sensor. The only exception at the moment is the wind speed sensor. The data collection is performed inside the main method for collecting and logging the sensor readings:

//  Raad the sensors and publish the data.
void ReadAndPublishSensorData()
    digitalWrite(1, HIGH);
    DebugMessage("\r\nCurrent time: " + ntp->getTimeString());
    //  Read the current wind speed.
    DebugMessage("Reading wind speed.");
    windSpeedPulseCount = 0;
    attachInterrupt(PIN_ANEMOMETER, IncreaseWindSpeedPulseCount, RISING);
    delay(WINDSPEED_DURATION * 1000);
    digitalWrite(1, LOW);

Reading the wind speed is performed through the ISR described above. The algorithm is simple:

  1. Clear the count of the number of revolutions (pulses) from the sensor
  2. Attach an interrupt to the sensor (the interrupts increments the count every revolution of the sensor)
  3. Wait for a know number of seconds (in this case 5)
  4. Detach the interrupt to stop the count

By using this method we can provide an average over a number of seconds and the wind speed can be calculated as:

Wind Speed = (Revolution count / number of seconds) * 1.492

This is the calculation performed in the PostDataToPhant method.

Setup and Loop

The final things needed by an application developed in the Arduino environment are the setup and loop methods. So let’s start looking at the setup method.

//  Setup the application.
void setup()
    Serial.println("\r\n\r\n-----------------------------\r\nWeather Station Starting (version " VERSION ", built: " __TIME__ " on " __DATE__ ")");
    //  Connect to the WiFi.
    Serial.print("\r\nConnecting to default network");

At the start of setup we need to rely upon the Serial object being available and so there are no calls to the DebugMessage method as this may be modified later to use networking for debugging. The next step is to try and connect to the network:

    while (WiFi.status() != WL_CONNECTED) 
    Serial.print("WiFi connected, IP address: ");

At this point the application will be either looping indefinitely until the network becomes available or we will have an IP address output to the serial port. In a more complete application this will start the logging process and periodically try to connect to the network when the network is unavailable. This is only a proof of concept after all.

There is currently no Real Time Clock (RTC) attached to the system and so we need to check the network time at startup.

    //  Get the current date and time from a time server.
    DebugMessage("Setting time.");
    ntp = ntpClient::getInstance("", 0);
    ntp->setInterval(1, 1800);
    while (year(ntp->getTime()) == 1970)

This block of code loops until the network time is set correctly. At start up the year will be set to a default value of 1970, this is why the code loops until the year is something other than 1970.

Next up, setup the sensors and take our first reading:

    //  Set up the sensors and digital pins.
    //  Read the initial data set and publish the results.

Everything is setup, only thing left is to continue to collect and publish the data, enter loop:

//  Main program loop.
void loop()
    delay(SLEEP_PERIOD * 1000);

Example Output

Running the above code results in the following output in the serial monitor (note that some data (IP addresses has been modified):

Weather Station Starting (version 0.04, built: 06:46:39 on Apr  3 2016)

Connecting to default network...........
WiFi connected, IP address:
Setting time.
Retrieved TSL2561 device ID: 0x50
Powering up the luminosity sensor.
BME280 sensor located on I2C bus.

Current time: 06:48:45 03/04/2016
TSL2561 data: 0x0373, 0x004f
Lux: 399.65
Temperature: 18.89 C
Humidity: 52.86 %
Pressure: 1006 hPa
Ultraviolet light: 307
Reading wind speed.
Status code: 200
Phant response code: 1

The current public data stream for this service should be viewable here.

Checking this stream you can see the resulting data that was sent to Sparkfun for the above serial output:

Phant Data From Weather Station Proof of Concept

Phant Data From Weather Station Proof of Concept

One thing to remember when comparing the above data is that the Sparkfun data stream logs data using UTC and the application above was running during British Summer Time.


The code above is a proof of concept, it is not error proof nor does it take into account power usage or optimisation, it is merely meant to prove that the sensors can be read and the data output to a cloud service.

Next steps for the proof of concept:

  1. How to handle two sensors requiring an analog conversion
  2. The project is running out of digital pins
  3. Real time clocks
  4. Offline data logging

All of this before even looking at the location and powering the project in the wild (OK, my garden).

Back right after this break…

Wind Speed and Ultraviolet Light Sensors

April 2nd, 2016 • Electronics, ESP8266, Internet of Things1 Comment »

Last time we looked at the I2C Sensors that are part of the weather station. At the end of the article it was noted that these were working and able to log data to the Particle ecosystem.

Now it is time for the analog sensors:

  1. Wind speed
  2. Wind direction
  3. Rainfall
  4. Ultraviolet light

The Ultraviolet light and wind direction sensor will present us with a problem as they are both analog sensors and the Oak only has one Analog-to-Digital Convertor (ADC).

The rainfall gauge and the wind speed sensor both use a similar technology to generate a signal, namely a magnet that will trigger a reed switch. These two sensors also present an issue; while they are simple enough there are a finite number of pins on the Oak and they are being consumed at a fair rate.

ML8511 – Ultraviolet Light Sensor

The ML8511 measures ultraviolet light at a wavelength of 365nM. This is at the top end of the UVB band, the band which is harmful to living tissue. The sensor generates a voltage that is linear and proportional to the intensity of the UV light. The intensity of the light measured is on the scale 0 to 15 mW/cm2. The following chart is taken from the data sheet:

ML8511 UV Plot

ML8511 UV Plot

The chart above is generated when the supply voltage is 3.0V but the system under development will be using a 3.3V supply. Some investigation is required to determine the output voltage when there is no UV present. After running the sensor under constant temperature conditions with no UV light present the ADC was generating a reading in the 316 – 320 range. This gives out output voltage for the sensor in the range 1.02V – 1.03V.

If we assume that the output characteristics of the sensor remain linear at 3.3V and the gradient remains the same then we can generate a formula for calculating the intensity of the UV light based upon the sensor output.

All linear graphs can be represented by the following formula:

y = mx + c


  • y is the y coordinate (in our case the voltage output of the sensor)
  • x is the x coordinate (in our case the intensity of the UV light)
  • m is the gradient of the line
  • c is a constant offset of the y coordinate

If we set x to 0 then the equation becomes y = c This represents the output of the sensor when there is no UV light present. As we have seen from the above experiment, this is in the range 1.02V – 1.03V. So as an approximation we will use the value c = 1.025V.

The gradient of any line is represented by the following formula:

m = deltaY / deltaX

When given two points on a line (x1, y1) and (x2, y2), then the gradient becomes:

m = (y2 – y1) / (x2 – x1)

Luckily the data sheet gives us two points on the line at 25C, namely the output voltage in shade and the output voltage at 10mW/cm2. So assuming shade represents no ultraviolet light then the gradient of the line becomes:

m = (2.2 – 1.0) / (10 – 0)

So m is 0.12. Plugging m and c back into the original equation gives:

y = 0.12x + 1.025

Solving for x gives:

x = (y – 1.025) / 0.12

Or to put it in context:

UV Intensity = (Sensor output in volts – 1.025) / 0.12 mW / cm2

There are some assumptions in the above work and the data sheet shows that the output voltage can vary but we have a method for calculating the intensity of UV light; accurate enough for a home project anyway.

Wind Speed

Wind speed is measured by an anemometer. The anemometer in the Weather Sensor kit is a cup anemometer:

Cup Anemometer

Cup Anemometer

This has a magnet on the spindle connected to the cups. The magnet closes (and opens) a reed switch each time it passes the switch. So one pulse is generated per full revolution of the spindle. Each full revolution of the spindle (per second) represents a wind speed of 1.492 miles per hour (mph) or 2.4 km/h.

For the experienced, this sounds simple but we all know there could be a nasty shock in store for us, namely switch bounce. An easy way to find out, hook the output up to an oscilloscope.

The circuit is simple enough, connect one switch contact to 3.3V, one to a resistor and the other end of the resistor to ground. Connect the scope to the resistor / switch junction.

The scope was set up to have a trigger voltage of about 1.5V and to trigger on the falling edge of a signal. The cups on the anemometer were then position so that the reed switch was closed (a high output on the scope) and the scope setup in single shot mode (to capture and hold the trace when triggered). Spinning the anemometer gave the following output:

Anemometer Switch Bounce

Anemometer Switch Bounce

As you can see, the switch does bounce. Switch debouncing is a well known and documented problem, in fact I have written about it here so we will not go too deep into the problem in this article. The solution that will be used is a simple RC circuit:

MSP430 Launchpad Debounce Circuit

MSP430 Launchpad Debounce Circuit

The principle is that the RC circuit resists change and so filters out the glitch in the above trace. So we need a filter that is resistive enough to filter out the glitch but fast enough to respond in the minimum time between pulses.

The weather station is going to be located in the mainland UK, about 30 miles from the coast. In this location the wind speed is unlikely to rise above 60-80 mph unless in extreme conditions (tornadoes are known to occur in the UK). So assuming the maximum wind speed in 149.2 mph (this number is based upon the fact that 1.492 mph gives one revolution per second) then we have a maximum number of rotations of 100 per second. This gives a revolution time of 10 milliseconds.

So we have a 10 millisecond window for the pulse from the sensor. The pulse will be low for the majority of the time as the switch can only close when the magnet is above the switch, so for most of the 10ms we will have a low pulse. You can see this in the following trace:

Anemometer Duty Cycle

Anemometer Duty Cycle

The signal appears to be high for 30% of the time. An accurate measurement could be made using the scope but it does not appear to be necessary. In our case, 10 milliseconds per rotation, the signal would be low for approximately 70% of the time, i.e. 7 milliseconds.

Going back to the first trace should the switch bounce it is observed that the switch bounce lasts for approximately 50-60 microseconds. Several observations of both the rising and falling edges showed this to be reasonable consistent. This final piece of information helps to define the parameters for this problem:

  • The frequency of the pulses should be 100Hz maximum (i.e. 10 milliseconds between pulses)
  • Duty cycle is 30% (high for 3 milliseconds, low for 7 milliseconds)
  • Switch bounce can last 50-60 microseconds (assume 100 microseconds as a worst case)
  • Trigger voltage is 2.3V with a supply voltage of 3.3V

There are plenty of online calculators for this type of circuit, Layada has one here that covers this type of scenario. We know the supply voltage (3.3V) and the trigger voltage (2.3V) so the only thing to do is look at the components available and calculate the delay time. After a few tries a 10K resistor and a 100nF capacitor were found to give a delay of 1.1939 milliseconds. This covers our case with plenty of margin for error.

Putting together the circuit above where the anemometer is the switch and triggering on the rising edge gives the following output on the oscilloscope:

Anemometer Gradual Rise

Anemometer Gradual Rise

The rise time looks to be acceptable, smooth and slow enough to iron out the glitches but fast enough to allow a 10 millisecond duration with a 30% duty cycle.


At the start of the article it was noted that four analog sensors are present in the kit. This article has concentrated on just two, the UV and Wind Speed sensors. The Rainfall and Wind Direction sensors will be covered in future articles.

The two sensors here will require one digital pin and one analog pin. As there is only one analog pin on the Oak then some creative thinking will be required in order to connect the Wind Direction sensor.

Assuming a maximum wind velocity of 149.2 mph then we will have 100 rotations of the anemometer per second. This should be something that could be measured by the Oak using an interrupt on one of the digital pins

At this point the hardware for six measurements can be put together on breadboard and some prototype software put together. This is the subject of the next article, Proof of Concept Software

Reading I2C Sensor Data with the Oak

April 1st, 2016 • Electronics, ESP8266, Internet of Things, Software Development2 Comments »

The weather station project will be bringing together a number of sensors, light, ultraviolet light, air pressure, humidity, wind speed, wind direction and rain fall. This collection of sensors falls into three groups:

  • Electronic sensors on an I2C bus
  • Mechanical sensors using switches
  • Analog sensors

The current plan is for the weather station to use the Oak as the microcontroller running the show. The data from the sensors can then be uploaded to the cloud, destination to be determined, but let’s start with Sparkfun’s data service.

The I2C sensors will require the least amount of work to get up and running so let’s start with those. The two sensors operating on the I2C bus are:

Oak and  2C Sensors on Breadboard

Oak and 2C Sensors on Breadboard

One of the great things about working with these two sensors is the fact that there are prebuilt drivers and example code for both breakout boards available from Github. What could be simpler, well head over to the Sketch – Include Library – Board Manager… menu in the Arduino IDE and you can download the library and have the IDE install it for you.

TSL2561 – Luminosity Sensor

This sensor allows the radiance of the light to be calculated in a way that approximates the response of the human eye. It does this by combining the input from two photodiodes, one infra-red only and one visible light and infra-red light combined. The output from the two sensors can be used to luminous emittance in lux (lumens per square metre).

The following table gives an idea of the lux values for typical scenarios:

Lux Typical Environment
0.0001 Moonless, overcast night sky
0.002 Moonless clear night sky
0.27–1.0 Full moon on a clear night
80 Office building hallway
320–500 Office lighting
1000 Overcast day
10000–25000 Daylight

As you can see from the table above, the lux values for a “normal” human day can vary dramatically. The sensor copes with this by allowing the use of a variable time window and sensitivity when taking a reading. Effectively the sensor accumulates the readings over the time window (integration interval) into a single 16-bit number which can then be used to calculate the lux reading.

BME280 – Air Pressure, Temperature and Humidity Sensor

This sensor is produced by Bosch and is packaged in both I2C and SPI configuration on the same board. The accuracy of the sensor appears good, pressure and temperature both to 1% and humidity to 3%.


Both Sparkfun and Adafruit have provided libraries and example code for the boards. These were easy to add to the development environment.

One caveat, the BME280 requires the addition of the Adafruit sensor library as well as the BME280 library.

Once added it was a simple case of wiring up the sensor to 3.3V and the I2C bus and running the example code.

They both worked first time.

Some Code Modifications (for later)

The light sensor has been show to work in low light conditions but not to any degree of precision. A possible modification to the example code is to look at the sensitivity and integration window settings to see if the precision can be adjusted to make the sensor return better readings in low light.

Some of the values when calculated use the fractional part of a floating-point number, temperature and humidity spring to mind. This meant adding a method to convert a double into it’s string representation for debugging purposes. Trivially solved but an annoying omission from the implementation of sprintf.

Posting to Particle Dashboard

The Oak can also post to the Particle dashboard providing a second method of debugging your application. Statements such as Particle.publish(“Debug”, “Temperature data…”); will cause the string Temperature data… to be posted in a group/attribute Debug

So the readings from the office look like this:

Some Weather Data

Some Weather Data


Two simple to use sensors with good supported class libraries make these sensors quick and easy to hook up to the Oak. Merging the two examples was simple and sensor data is now appearing over the serial output from the Arduino.

Compiling the code gives some warnings about the I2C library being compiled for the ESP8266 whilst the target board is defined as an Oak. This can be ignored as the Oak is really a convenient wrapper around an ESP8266 module.

Next up, some analog sensors to measure ultraviolet light and wind and rain properties.

Setting Up the Oak – Flashing LED

March 27th, 2016 • Electronics, ESP8266, Internet of Things, Software DevelopmentComments Off on Setting Up the Oak – Flashing LED

The Oak microcontroller is new, Digistump only shipped version 1 firmware a few days ago (20th Match 2016). The hardware I have was shipped sometime in January. So the first thing to do is to upgrade the firmware and then try blinking an LED.

Upgrading the firmware

The Digistump Wiki contains a number of tutorials and troubleshooting guides. First stop the Connecting Your Oak for the First Time page. This shows how the Oak can be connected to your WiFi network and the firmware updated.

The initial over the air firmware update was problematic to say the least. Reading through the Digitsump forums it seems that I am not the only one having a problem with the first update. There are three methods for upgrading the firmware:

  • Over the Air using firmware from the internet
  • Over the air with a local server
  • Serial using pyserial or esptool

I started at the top of the list and slowly worked my way down. In the end the only method that worked for me was the serial update.

Flashing an LED

With the latest version of the firmware installed it is time to test the development environment. What could be simpler than flashing and LED. The on board LED is connected to pin 1 so lets try and use that.

Digistump offer two development environments:

At the time of writing there was a known issue with the Particle development environment which prevented an application being built and flashed successfully. This is an early release and so issues are expected.

This only leaves the Arduino environment, an IDE I really hate.

There are two methods for flashing an application to the Oak using the Arduino environment:

  • Over The Air (OTA)
  • Serial (requires Python)

Despite the earlier problems with the firmware update I decided to try the OTA method first. DigitStump have provided compehensive instructions in their WiKi on how to achieve this.

Following the example was easy and after only 15 minutes I have the LED on the Oak flashing at 1Hz. Just to prove it was not a fluke I then tried changing the frequency and reflashing the Oak.



My initial frustration with the firmware update was soon forgotten once I had an application successfully running on the Oak. I am hoping that the issues were caused by the fact I have an early release of the board with the original firmware installed.

Programming is easy enough and can be done over the air which is convenient.

Next up, talking to sensors.

Weather Station

March 25th, 2016 • Electronics, ESP8266, Software DevelopmentComments Off on Weather Station

The first part of this year has seen me working on some photography projects, now time to get the soldering iron out. I have three projects I’m hoping to complete this year and one long term project that may take me a while, so pay attention 007.

Weather Station

A few weeks ago I received a couple of Oaks. These boards are based around the ESP12 module, a Kickstarter project I backed at the end of 2015. The project aimed to make working with the ESP12 easier by providing integration with The latest firmware was release a few days ago so I think it is time to give this little device a go.

At the start of the year we came across a weather meter which allowed the measurement of rainfall, wind speed and wind direction. My wife mentioned that it might be nice to use one of these to record the weather in the garden.

And so the idea was born, weather meter, meet the Oak.

Mapping Out The Project

After a few hours the project started to become larger than just a couple of sensors, it currently looks something like the following:

Weather Station Mind Map

Weather Station Mind Map

I have coloured each block of concepts the same colour in order to give the project some structure. Let’s look at each block in turn and see how we can approach the problem.

Temperature, Pressure and Humidity

Starting with the green block and five o’clock on the diagram.

Temperature, Presure and Humidity Mind Map

Temperature, Presure and Humidity Mind Map

There are two temperatures we can measure here, air temperature and soil or ground level temperature. The air temperature will give an indication of how pleasant a day it is at the moment and the ground level sensor will give an indication of the progress of the seasons from the plants perspective. For this reason we are interested in both of these measurements.

Air pressure has long been used as a predictor for upcoming weather events.

Luckily we can get the air temperature, pressure and humidity sensor in one convenient package, the BME280. Ground level temperature will need a waterproof sensor. I think this would be best purchased as a unit rather than made so the DS18B20 with a 6 foot cable looks like it might be ideal.

Wind and Rain

Working clockwise, the next block are the wind and rain sensors:

Wind and Rain Sensor Mind Map

Wind and Rain Sensor Mind Map

All of these measurements come from the weather meter. The wind speed and rainfall sensors are simple switches that generate pulses whilst the wind direction is a resistor network.

Light Sensor

The next block shows the two light sensors:

Light Sensor Mind Map

Light Sensor Mind Map

The two light measurements are the overall light intensity (luminosity) and ultraviolet light intensity. Luckily there are a couple of sensors for these two measurements, the TSL2561 and the ML8511.

The ultraviolet light sensor is an analogue sensor and so we will have to consider the stability of the supply voltage when making the reading.

The luminosity sensor uses a measurement window and a sensitivity setting to take a reading. This means that for given settings the sensor may be overwhelmed and simply give a maximum reading. The work around for this is to make the measurement window and sensitivity dynamic. So long, sensitive windows at night and short less sensitive windows on bright sunny days.

Cases and Location

The next things we need to consider are cases and locations:

Case and Location Mind Map

Case and Location Mind Map

The microcontroller and power etc. will need to be located in a case of some form. This will need to be weather proof as water and electricity are not the best of friends. It would also be a good idea to keep any batteries in an environment with a reasonably stable temperature. A little research into how the professionals house weather Stations.

The sensors on the other hand need to be outdoors in a suitable location for the measurements being taken.

Power Supply

All of this equipment will need a stable power supply:

Power Supply Mind Map

Power Supply Mind Map

The initial work can be done using a bench power supply but when the project moves outdoors it will need to either be mains or battery powered. The long term aim is to use a solar cell and rechargeable battery, as they say on Kickstarter, a stretch goal.

Data Logging and Real Time Clock

Now we have collected all of this information we need to do store it somewhere:

Data Logging Mind Map

Data Logging Mind Map

The Oak is a WiFi enabled board so the most obvious place to put the data is the cloud. It might be an idea to also provide some local storage in case the WiFi network is unavailable.

The Real Time Clock (RTC) could have two uses, to wake the microcontroller and also provide a timestamp for the data items.


What started out as a simple wind and rain logging project has grown a little. The current specification is to measure and log the following readings:

  1. Rainfall
  2. Wind speed
  3. Wind direction
  4. Luminosity
  5. Ultraviolet light

The measurements should be collected and reliably logged either locally or preferably to the cloud.

There are several challenges with a number of interesting problems to overcome.

Let’s get this show on the road.

So I’m Sneaky

March 6th, 2016 • PhotographyComments Off on So I’m Sneaky

So it’s official, I’m sneaky. That’s what I have been told by my wife, Karen. Why you may ask, well not all of my projects are electronic or software related; I’m also a keen, average amateur photographer.

Flying Scotsman

Restoration on the Flying Scotsman was finally completed this year. This was a long term project after the locomotive was purchased for the nation by the National Railway Museum. The train re-entered service with a run from London to York in February 2016.

As part of the return, the National Railway Museum organised a couple of photo opportunities, one early in the morning and one in the evening when it was dark. In both cases the locomotive would be under light steam and the evening shoot would be atmospherically lit. Now I’m not a train enthusiat per se but as a photographer how could I resist.

Here are a couple of the photos from that evening:

Flying Scotsman in North Yard

Flying Scotsman in North Yard

And a low shot:

Flying Scotsman in North Yard Low Shot

Flying Scotsman in North Yard Low Shot

But that does not explain the sneaky…

Family Photographs

My wife’s family have always been very close but sadly her mother and father are no longer with us. We do have a collection of about 300 photographs from her mothers family collection. Unfortunetly they are in mixed condition, some of them rather poor. The photographs had a variety of problems from creases to marks and tears on the surface of the image. For example, the following is a snippit from one of the images:

Unprocessed Image With Creases

Unprocessed Image With Creases

Time for some restoration work. Modern image software is very impressive and not always expensive. The software I tried (and later purchased) last year is Mac specific, there is no PC version, and won Mac application of the year for 2015. The software is Affinity Photo. This software offers an impressive set of features and the video tutorials provide a new user with an insight on how to use the features.

So taking the image from above and 30 minutes with the software and I had the following:

Processed Images With Creases Removed

Processed Images With Creases Removed

Creases all gone, now on to improving the image a little.

Still no sign of sneaky…

Starting in January, I methodically took each of the 300+ images and started to restore the full collection. A few of the images were in too poor condition or out of focus and so they were discarded but the majority of the images could be improved a little. Six weeks later I had a comprehensive set of photographs going back to Victorian times.

The final stage of the processing was the creation of a family album starting with from Great Grandparents through to the current day. For this I used Boots Photo Service to create an 11×8 inch hardback album.

And that’s why my wife thinks I’m sneaky.


It has been a while since I was able to use the soldering iron and if I am to get through the projects planned for Christmas then I need to pick it up again. The time playing with Affinity Photo has certainly proved fruitful, I can well recommend this product.

The brief diversion into image processing did however bring a moment of pleasure into my wife’s life, a moment I’ll always treasure.

Projects for 2016

January 31st, 2016 • General2 Comments »

The Christmas break always gives time to plan the coming years projects. Like all hobbyists I’m sure that I’m going to be planning far more projects than I have time to complete. One thing that I am going to try and do is complete some of the projects from former years.

LED Cube

So let’s start with the LED cube. This project started a while ago and the electronics and software worked well but the LEDs have ended up sitting on a shelf for a few years. The aim was to house the cube in a suitable case but for some reason this was never really completed.

Time for a change.

The first step in the completing the project is to connect the LED cube to a base board. The board would provide the power to the LEDs, nothing more, just power to the LEDs. The design utilised a simple transistor circuit to turn the LEDs on or off. A pull-down resistor ensures that the default situation for the switch will be the off position:

Transistor LED Schematic

Transistor LED Schematic

Multiply this by 64 and you have the base board for the LED cube. The PCB layout looks like this:

LED Cube Base Board Layout

LED Cube Base Board Layout

The board design was sent for manufacture in early-mid December. This is by far the largest (although not the most complex) design I have had made. The board was 20cm x 20cm. The final boards arrived just before Christmas:

The top of the board has 64 pads laid out in a grid for the LEDs:


The bottom of the board has 64 copies of the power circuit above:

LED Base Board Power Circuit

LED Base Board Power Circuit

When I was applying the solder paste to this board I was wishing I could afford a solder mask. After building a couple of power circuits it occurred to me that a simple mask could be made using an overhead acetate (remember these from the 1980s/90s) and a dremel. It does not matter that the holes in the mask would be circular, only that they are in the correct position. A 1mm diameter drill bit in a dremel and a piece of acetate provide the right amount of solder paste to the board.

A tip for the future.

The next step was to mouth the LEDs. Easier said than done. The cube had already been assembled and so the legs of the LEDs had to be coaxed into position. Several hours later we have:

LEDs On Base Board

LEDs On Base Board

Soldering on some connectors allows a quick check of each LED in the cube. It works 🙂

Next step is to look at the control circuitry.

Other Plans

As well as completing the cube, there are a few other projects that are in the pipeline:

  • Christmas jumper for a friend – adding some lights, we all love LEDs
  • Get to grips with FPGAs
  • Docker and Vagrant – I know Vagrant can help me with ESP8266 development
  • ESP8266, ESP32?, Photon, Oak and RedBear Duo – There is going to be a lot of IoT in 2016
  • Complete the wireless infra-red remote control

2016 – time to complete some projects and start some new ones.

PiZero – Is it a Zero or Hero

December 6th, 2015 • Raspberry PiComments Off on PiZero – Is it a Zero or Hero

On 26th November the Raspberry Pi foundation announced the release of the PiZero, a $5 computer based upon the Broadcom 2835 processor. Looks like it could be interesting…

First Impressions

The board itself is about 60% the size of a credit card and has a reduced number of input output options compared to the other Raspberry Pis in the range but promises more processing power than the original Raspberry Pi. A quick look at the specifications reveals the following:

  • Broadcom BCM2835 processor (1GHz ARM11 core – 40% faster than the original Raspberry Pi)
  • 512MB of SDRAM
  • A micro-SD card slot
  • A mini-HDMI socket
  • Micro-USB sockets, one for data and one for power
  • An unpopulated 40-pin GPIO header (Identical to the Model A+/B+/2B)

The form factor and price combined are sure to make this ideal for the hobby/hacker market.


Some hobbyists are sure to have the additional connectors in their parts bucket, but not this one, every other cable you can think of but not the OTG or HDMI convertor! Over the years I have collected a fair number of cables and other spare parts but none are suitable for the PiZero. The micro-SD cards I have in abundance but the additional connectors needed to start work are another matter. In hindsight I should have ordered the additional cables when purchasing the PiZero from Pimoroni as these were on offer for £8 at the time of ordering.

Hindsight is a wonderful thing…

The PiZero arrived in short order, the accessories were another matter. It is not often that Amazon let me down but this was one of those times. A small pause while waiting for delivery.

Setting up the PiZero

Setting up the PiZero was as simple as the initial setup for the rest of the Raspberry Pi family. The easiest option is to use NOOBS. This just needs a formatted SD card, 16GB should do it.

Formatting the SD card and dropping NOOBS on the card took about 5 minutes, now time to finally power on the Raspberry Pi. But first a few connections, the HDMI was connected through a standard HDMI cable through a convertor to the the mini-HDMI. My USB hub was connected through an ingenious USB OTG Convertor. Keyboard, trackball and WiFi dongle were added to the hub. We’re now ready to go.

NOOBS starts to the desktop by default and so the mini-HDMI connection is essential. The board started reasonably quickly although users of the Raspberry Pi B+ etc will find the start up a little slow.

The first things NOOBS did was to reorganise the file store, this takes about 30 minutes.

The next step, booting to the desktop and the first thing to do, configure the system to boot to the command line and setup the network settings for the WiFi.

And then reboot.

One hour after starting the process I had a PiZero on the network connecting to the command line via SSH.


The PiZero created a fair amount of excitement at work but for what are probably the wrong reasons. The initial excitement about the $5 price tag soon starts to wane when you start buying the add-ons. Don’t get me wrong this board has a purpose, this is a fine low cost hackers board. My concern is the number of people thinking that this is a low cost home computer for the family. If you want a low cost family Raspberry Pi then buy the Raspberry B2/B+ as these will be far easier to connect to the kit most families are likely to have in the home.

So where does the PiZero fit in. Well I think this is really a good fit for the market the compute module market. The compute module was always going to be a pain to work with for the hobbyist and the full Raspberry Pi is too large for some applications (quadcopters etc). The PiZero fits nicely into this market. Getting the additional components (OTG cables etc) is a one off cost for the hobbyist embedding a PiZero into a project. This is where I think this board will fit nicely.

So, in conclusion, great for the hobbyist hacker but for families wanting to get kids into computing I’d go for the Raspberry Pi B2/B+ as it’s going to be far easier to get connected.

LED Board Has Arrived

November 1st, 2015 • ElectronicsComments Off on LED Board Has Arrived

A few weeks ago I started to work with KiCAD and put together a small LED board as a project to become familiar with the application. The boards were sent to manufacture and the the postman delivered the manufactured PCBs yesterday.

LED Board Front

LED Board Front

LED Board Front

LED Board Front

Time to add the components and test the boards.

Populated Board Right Angled Headers

Populated Board Right Angled Headers

And one with straight headers:

Populated Board Straight Headers

Populated Board Straight Headers

Everthing looks good, the boards work OK so time to move on to the big board for the cube.