AdaFruit IO basic desktop HTTP client

AdaFruit IO meets my basic criteria as it has support for HTTP/S clients (it also has an MQTT interface which I will look at in a future post) and the API is well documented.

My first Proof of Concept (PoC) was to build a desktop client which used the HttpWebRequest (for ease of porting to NetMF) classes to upload data.

The program uploaded one of three simulated values to AdaFruit.IO every 10 seconds.

I found the username, group, and feed keys to be case sensitive so pay close attention to the values displayed in the webby UI or copy n paste.

program.cs

 class Program
   {
      static void Main(string[] args)
      {
         string adaFruitIOApiBaseUrl = "https://IO.adafruit.com/api/v2/";
         string adaFruitIOUserName = "YourUserName"; // This is mixed case & case sensitive
         string adaFruitIOApiKey = "YourAPIKey";
         // The feed group and feed key are forced to lower case by UI
         const string feedGroup = "";
         //const string feedGroup = "devduinov2-dot-2";
         const string temperatureKey = "t";
         const double temperatureBase = 20.0;
         const double temperatureRange = 10.0;
         const string humidityKey = "h";
         const double humidityBase = 70.0;
         const double humidityRange = 20.0;
         const string batteryVoltageKey = "v";
         const double batteryVoltageBase = 3.00;
         const double batteryVoltageRange = -1.00;
         TimeSpan dataUpdateDelay = new TimeSpan(0, 0, 10);
         Random random = new Random();

         while (true)
         {
            double temperature = temperatureBase + random.NextDouble() * temperatureRange;
            Console.WriteLine("Temperature {0}°C", temperature.ToString("F1"));
            AdaFruitIoFeedUpdate(adaFruitIOApiBaseUrl, adaFruitIOUserName, adaFruitIOApiKey, feedGroup, temperatureKey, temperature.ToString("F1"));

            Thread.Sleep(dataUpdateDelay);

            double humidity = humidityBase + random.NextDouble() * humidityRange;
            Console.WriteLine("Humidity {0}%", humidity.ToString("F0"));
            AdaFruitIoFeedUpdate(adaFruitIOApiBaseUrl, adaFruitIOUserName, adaFruitIOApiKey, feedGroup, humidityKey, humidity.ToString("F0"));

            Thread.Sleep(dataUpdateDelay);

            double batteryVoltage = batteryVoltageBase + random.NextDouble() * batteryVoltageRange;
            Console.WriteLine("Battery voltage {0}V", batteryVoltage.ToString("F2"));
            AdaFruitIoFeedUpdate(adaFruitIOApiBaseUrl, adaFruitIOUserName, adaFruitIOApiKey, feedGroup, batteryVoltageKey, batteryVoltage.ToString("F2"));

            Thread.Sleep(dataUpdateDelay);
         }
      }

client.cs

      public void AdaFruitIoFeedUpdate(string apiBaseUrl, string userName, string apiKey, string group, string feedKey, string value, int httpRequestTimeoutmSec = 2500, int httpRequestReadWriteTimeoutmSec = 5000)
      {
         string feedUrl;

         if (group.Trim() == string.Empty)
         {
            feedUrl = apiBaseUrl + userName + @"/feeds/" + feedKey + @"/data";
         }
         else
         {
            feedUrl = apiBaseUrl + userName + @"/feeds/" + group.Trim() + "." + feedKey + @"/data";
         }

         Console.WriteLine(" Feed URL :{0}", feedUrl);

         try
         {
            HttpWebRequest request = (HttpWebRequest)WebRequest.Create(feedUrl);
            {
               string payload = @"{""value"": """ + value + @"""}";

               byte[] buffer = Encoding.UTF8.GetBytes(payload);

               DateTime httpRequestedStartedAtUtc = DateTime.UtcNow;

               request.Method = "POST";
               request.ContentLength = buffer.Length;
               request.ContentType = @"application/json";
               request.Headers.Add("X-AIO-Key", apiKey);
               request.KeepAlive = false;
               request.Timeout = httpRequestTimeoutmSec;
               request.ReadWriteTimeout = httpRequestReadWriteTimeoutmSec;

               using (Stream stream = request.GetRequestStream())
               {
                  stream.Write(buffer, 0, buffer.Length);
               }

               using (var response = (HttpWebResponse)request.GetResponse())
               {
                  Console.WriteLine(" Status: " + response.StatusCode + " : " + response.StatusDescription);
               }

               TimeSpan duration = DateTime.UtcNow - httpRequestedStartedAtUtc;
               Console.WriteLine(" Duration: " + duration.ToString());
            }
         }
         catch (Exception ex)
         {
            Console.WriteLine(ex.Message);
            throw;
         }
      }
   }

This approach seemed to work pretty reliably

DesktopHTTPRequest

nRF24 Windows 10 IoT Core Background Task

First step is to build a basic Windows 10 IoT Core background task which can receive and display messages sent from a variety of devices across an nRF24L01 wireless link.

If you create a new “Windows IoT Core” “Background Application” project then copy this code into StartupTasks.cs the namespace has to be changed in the C# file, project properties\library\Default namespace and “Package.appxmanifest”\declarations\Entry Point.

/*

Copyright ® 2017 December devMobile Software, All Rights Reserved

THIS CODE AND INFORMATION IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY
KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND/OR FITNESS FOR A PARTICULAR
PURPOSE.

http://www.devmobile.co.nz

*/
using System;
using System.Diagnostics;
using System.Text;
using Radios.RF24;
using Windows.ApplicationModel.Background;

namespace devmobile.IoTCore.nRF24BackgroundTask
{
    public sealed class StartupTask : IBackgroundTask
    {
      private const byte ChipEnablePin = 25;
      private const byte ChipSelectPin = 0;
      private const byte nRF24InterruptPin = 17;
      private const string BaseStationAddress = "Base1";
      private const byte nRF24Channel = 10;
      private RF24 Radio = new RF24();
      private BackgroundTaskDeferral deferral;

      public void Run(IBackgroundTaskInstance taskInstance)
        {
         Radio.OnDataReceived += Radio_OnDataReceived;
         Radio.OnTransmitFailed += Radio_OnTransmitFailed;
         Radio.OnTransmitSuccess += Radio_OnTransmitSuccess;

         Radio.Initialize(ChipEnablePin, ChipSelectPin, nRF24InterruptPin);
         Radio.Address = Encoding.UTF8.GetBytes(BaseStationAddress);
         Radio.Channel = nRF24Channel;
         Radio.PowerLevel = PowerLevel.High;
         Radio.DataRate = DataRate.DR250Kbps;
         Radio.IsEnabled = true;

         Debug.WriteLine("Address: " + Encoding.UTF8.GetString(Radio.Address));
         Debug.WriteLine("PA: " + Radio.PowerLevel);
         Debug.WriteLine("IsAutoAcknowledge: " + Radio.IsAutoAcknowledge);
         Debug.WriteLine("Channel: " + Radio.Channel);
         Debug.WriteLine("DataRate: " + Radio.DataRate);
         Debug.WriteLine("IsDynamicAcknowledge: " + Radio.IsDyanmicAcknowledge);
         Debug.WriteLine("IsDynamicPayload: " + Radio.IsDynamicPayload);
         Debug.WriteLine("IsEnabled: " + Radio.IsEnabled);
         Debug.WriteLine("Frequency: " + Radio.Frequency);
         Debug.WriteLine("IsInitialized: " + Radio.IsInitialized);
         Debug.WriteLine("IsPowered: " + Radio.IsPowered);

         deferral = taskInstance.GetDeferral();

         Debug.WriteLine("Run completed");
      }

      private void Radio_OnDataReceived(byte[] data)
      {
         // Display as Unicode
         string unicodeText = Encoding.UTF8.GetString(data);
         Debug.WriteLine("Unicode - Payload Length {0} Unicode Length {1} Unicode text {2}", data.Length, unicodeText.Length, unicodeText);

         // display as hex
         Debug.WriteLine("Hex - Length {0} Payload {1}", data.Length, BitConverter.ToString(data));
      }

      private void Radio_OnTransmitSuccess()
      {
         Debug.WriteLine("Transmit Succeeded!");
      }

      private void Radio_OnTransmitFailed()
      {
         Debug.WriteLine("Transmit Failed!");
      }
   }
}

This was displayed in the output window of Visual Studio

Address: Base1
PA: 15
IsAutoAcknowledge: True
Channel: 10
DataRate: DR250Kbps
IsDynamicAcknowledge: False
IsDynamicPayload: True
IsEnabled: True
Frequency: 2410
IsInitialized: True
IsPowered: True
Run completed

Interrupt Triggered: FallingEdge
Unicode – Payload Length 19 Unicode Length 19 Unicode text T  23.8,H  73,V 3.26
Hex – Length 19 Payload 54-20-32-33-2E-38-2C-48-20-20-37-33-2C-56-20-33-2E-32-36
Interrupt Triggered: RisingEdge

Note the odd formatting of the Temperature and humidity values which is due to the way dtostrf function in the Atmel AVR library works.

Also noticed the techfooninja nRF24 library has configurable output power level which I will try to retrofit onto the Gralin NetMF library.

Next, several simple Arduino, devDuino V2.2, Seeeduino V4.2 and Netduino 2/3 clients (plus possibly some others)

nRF24 Windows 10 IoT Core reboot

My first live deployment of the nRF24L01 Windows 10 IoT Core field gateway is now scheduled for mid Q1 2018 so time for a reboot. After digging out my Raspbery PI 2/3 devices and the nRF24L01+ shield (with modifications detailed here) I have a basic plan with some milestones.

My aim is to be able to wirelessly acquire data from several dozen Arduino, devduino, seeeduino, and Netduino devices, Then, using a field gateway on a Raspberry PI running Windows 10 IoT Core upload it to Microsoft IoT Central

First bit of code – Bleepy a simple background application to test the piezo beeper on the RPI NRF24 Shield

namespace devmobile.IoTCore.Bleepy
{
   public sealed class StartupTask : IBackgroundTask
   {
      private BackgroundTaskDeferral deferral;
      private const int ledPinNumber = 4;
      private GpioPin ledGpioPin;
      private ThreadPoolTimer timer;

      public void Run(IBackgroundTaskInstance taskInstance)
      {
         var gpioController = GpioController.GetDefault();
         if (gpioController == null)
         {
            Debug.WriteLine("GpioController.GetDefault failed");
            return;
         }

         ledGpioPin = gpioController.OpenPin(ledPinNumber);
         if (ledGpioPin == null)
         {
            Debug.WriteLine("gpioController.OpenPin failed");
            return;
         }

         ledGpioPin.SetDriveMode(GpioPinDriveMode.Output);

         this.timer = ThreadPoolTimer.CreatePeriodicTimer(Timer_Tick, TimeSpan.FromMilliseconds(500));

         deferral = taskInstance.GetDeferral();

         Debug.WriteLine("Rum completed");
      }

      private void Timer_Tick(ThreadPoolTimer timer)
      {
         GpioPinValue currentPinValue = ledGpioPin.Read();

         if (currentPinValue == GpioPinValue.High)
         {
            ledGpioPin.Write(GpioPinValue.Low);
         }
         else
         {
            ledGpioPin.Write(GpioPinValue.High);
         }
      }
   }
}

Note the blob of blu tack over the piezo beeper to mute noise
nRF24ShieldMuted

nRF24 Windows 10 IoT Core Hardware

Taking my own advice I decided to purchase a couple of Raspberry Pi to NRF24L01 shields from Ceech a vendor on Tindie.

The nRF24L01 libraries for my .Net Micro framework and WIndows 10 IoT Core devices use an interrupt driver approach rather than polling status registers to see what is going on.

Like most Raspberry PI shields intended to be used with a *nix based operating system the interrupt pin was not connected to a General Purpose Input/Output (GPIO) pin.

NRF24PiPlateModification

My first step was to add a jumper wire from the pin 8 on the nRF24L01 to GPIO pin 17 on Raspberry PI connector.

I then downloaded the techfooninja Radios.RF24 library for Windows IoT core and update the configuration to suit my modifcations. In the TestApp the modifications were limited to changing the interrupt pin from GPI 4 to GPO 17

private const byte IRQ_PIN = 4;

private const byte IRQ_PIN = 17;

I used a socket for the nRF24L01 device so I can trial different devices, for a production system I would solder the device to the shield to improve reliability.

RPiWithnRF24Plate

I then ran the my test application software in a stress test rig overnight to check for any reliability issues. The 5 x netduino devices were sending messages every 500mSec

RPIStressTester

nRF24L01 Raspberry PI Gateway Hardware

For those who came to my MS Ignite AU Intelligent Cloud booth session

Building Wireless Field Gateways

Connecting wireless sensor nodes to the cloud is not the mission it used to be, because the Azure team (and many OS projects) have developed tooling which can help hobbyist and professional developers build solutions. How could you build a home scale robust, reliable and secure solution with off the shelf kit without blowing the budget?

Sparkfun nRF24L01 module &Adafruit perma proto hat

NRF24L01 Raspberry PI DIY Gateway Hardware

BoM (all prices as at Feb 2016)

You will also need some short lengths of wire and a soldering iron.

For those who want an “off the shelf” solution (still requires a minor modification for interrupt support) I have used the Raspberry Pi to NRF24l01+ Shield USD9.90

2015-09-25t072754-447z-20150925_091942-855x570_q85_pad_rcrop

Instructions for modifications and software to follow.

Microsoft Sync Framework timezones

Over the last few months I have been working with the Microsoft Sync Framework and the time zone issues have been a problem.

New Zealand has a 12hr standard time or 13 hr daylight savings time offset from Coordinated Universal Time (UTC) and at a glance our customer data could look ok if treated as either local or UTC.

After some experimentation I found that it was due to Windows Communication Foundation(WCF) serialisation issues (The proposed solutions looks like it might have some limitations, especially across daylight savings time transitions).

For the initial synchronisation DateTime values in the database were unchanged, but for any later incremental synchronisations the DateTime values were adjusted to the timezone of the server (Our Azure Cloud Services are UTC timezone, though I don’t understand why Microsoft by default has them set to US locale with MM/DD/YY date formats)

In our scenario having all of the DateTime values in the cloud local looked like a reasonable option and this article provided some useful insights.

In the end I found that setting the DateSetDateTime  for every DateTime column in each DataTable in the synchronisation DataSet to unspecified in the ProcessChangeBatch (our code was based on the samples) method meant that no adjustment was applied to the incremental updates

public override void ProcessChangeBatch(ConflictResolutionPolicy resolutionPolicy, ChangeBatch sourceChanges, object changeDataRetriever, SyncCallbacks syncCallbacks, SyncSessionStatistics sessionStatistics)
{
try
{
DbSyncContext context = changeDataRetriever as DbSyncContext;

if (context != null)
{
foreach (DataTable table in context.DataSet.Tables)
{
foreach (DataColumn column in table.Columns)
{
// Switching from UnspecifiedLocal to Unspecified is allowed even after the DataSet has rows.
if ((column.DataType == typeof(DateTime)) && (column.DateTimeMode == DataSetDateTime.UnspecifiedLocal))
{
column.DateTimeMode = DataSetDateTime.Unspecified;
}
}
}
...

Hope this helps someone else