Hot Pi

In my last post, I explained how to set up the Pi with the drivers needed to allow it to read temperatures from DS120B sensors. In this post, I’ll show you the code and setup required to make it start taking readings automatically at bootup, and store the information in a Mysql database for future use. I’ll also add a couple of LEDs to the circuit, so that the system can provide some feedback and also to pave the way for controlling a relay or two.

Here’s the updated circuit schematic (click on it for a bigger version).

heating_2_bb

There are a few things to note about this slightly more complex schematic:

  1. I’ve added another temperature sensor. Its pins are simply connected directly to the pins of the first one (and any subsequent ones). The one-wire protocol and the fact that each sensor has a unique ID means that they can share a common bus.
  2. I’ve added two LEDs, just so that the system can indicate some state to me. One will be replaced with a relay in future.
  3. I’ve added an Adafruit level converter (the blue pcb). This is there to protect the Pi: its GPIOs run at 3.3V, and are not designed to carry much current. overloading them will cause damage to the Pi. The cheap level shifter allows the 3.3V Pi pins to be connected to 5V input or output from other devices, and protects the Pi from accidental over-voltage. It’s not strictly necessary just to drive LEDs, but the relays I want to use later need 5V.

I’m going to use Python as the main language for running the system. The reason for this is that I don’t know it very well, and there’s no better way to learn than to dive in and try and do something with it. Of course, this also means that my Python code may well be what is technically-termed ‘Not Very Good’. Use it at your peril. To make it work, you’ll have o install a couple of other things. One is MySql, the database into which the program will write the recorded temperature values (you’ll also need the associated Python library). The other is pigpiod which is, of course, a Pi GPIO library, though the name always looks like pig-pio to me. Pigpio (also available in its useful daemon form pigpiod) is a C library which exposes all the GPIO to C programs. Not useful for Python, I hear you say. True enough, but it also exposes the same functionality through a TCP socket interface, and it’s dead easy to use this from Python without installing any special python libraries at all. We’ll deal with that in my next post.

Installing MySql

In Linux, it down’t get much easier than this:

sudo apt-get update
sudo apt-get install mysql-server

At some point, the installer will ask you for a password for the root MySql user. In any serious situation, this should be a good, secure, password. I just used the same one as I’m using for everything else on this device. It’s only a toy, after all. Once MySql is installed, you can use the command line client to create a new user and to make the ;temperatures’ database and the ‘data’ table in which the readings will be stored. NB: when you are using the MySql command line client, don’t forget the semicolon at the end of a SQL command line. Multi-line commands (such as the CREATE TABLE command below) are fine in SQL, and they don’t get executed until the parser sees a semicolon. It catches me out all the time.

mysql --user=root --password=My5ecurePa$5word

mysql> create user 'pi'@'localhost' identified by 'raspberry';
mysql> grant all privileges on *.* to 'pi'@'localhost';
mysql> create database temperatures;
mysql> use temperatures;
mysql> CREATE TABLE 'data' ('id' int(11) NOT NULL AUTO_INCREMENT, 
'timestamp' datetime NOT NULL, 'sensor_id' int(11) NOT NULL, 
'temperature' float NOT NULL,  PRIMARY KEY ('id'));
mysql> quit;

When that’s done, you will have a database with an empty table in it, and a user ‘pi’ which you can use to connect to it from python. For that’ we’ll need the python-mysqldb module. This can be installed thus:

sudo apt-get install python-mysqldb

We can check that this works with a small python script:

#!/usr/bin/python
import MySQLdb as mdb
import sys

try:
    con = mdb.connect('localhost', 'pi', 'raspberry', 'temperatures');
    cur = con.cursor()
    cur.execute("SELECT VERSION()")
    ver = cur.fetchone()
    print ("Database version : %s " % ver)
except mdb.Error, e:
    print ("Error %d: %s" % (e.args[0],e.args[1]))
    sys.exit(1)
finally:    
    if con:    
        con.close()

Run this, and if all is well you will see something like

Database version 5.5.37-0+wheezy1

Or possibly an error. With unbridled optimism, I’m going to assume that you are not seeing an error. It’s now time to create the code which will actually read the temperature sensors and write the data to the database. Here it is:

#!/usr/bin/env python
import datetime, time, sys
import MySQLdb as mdb

def getTemp(chipid):
   tfile=open("/sys/bus/w1/devices/"+chipid+"/w1_slave")
   text=tfile.read()
   tfile.close()
   secondline=text.split("\n")[1]
   tempdata=secondline.split(" ")[9]
   temperature=float(tempdata[2:])
   temperature=temperature/1000
   return temperature

try:
   con = mdb.connect('localhost','pi','raspberry','temperatures')
   with con:
      cur = con.cursor()
      while(True):
          timestamp = datetime.datetime.fromtimestamp(time.time()).strftime('%Y-%m-%d %H:%M:%S')
          t1 = getTemp("28-000001a9b68a")
          t2 = getTemp("28-0000009bba2b")
          query = "INSERT INTO data(timestamp,sensor_id,temperature) VALUES ('%s',%s,%s)"%(timestamp,'1',str(t1))
          result = cur.execute(query)
          query = "INSERT INTO data(timestamp,sensor_id,temperature) VALUES ('%s',%s,%s)"%(timestamp,'2',str(t2))
          result = cur.execute(query)
          con.commit()
          print ("%s %s %s"%(timestamp, str(t1), str(t2))
          time.sleep(60) 
finally:
   print ("Goodbye")

s.close()

There are two main parts to this code. The function getTemp(chipid) reads the virtual file created by the 1-wire driver discussed in my last post for a specific temperature sensor, and extracts the temperature data from it. There’s nothing clever about it, just python string processing. You’ll need to modify the code further down to use the chip ids of your own sensors, of course. The main body of the program establishes a connection to the database, then enters an unending loop inside which it gets the temperature from each sensor and writes it to the database with a timestamp using a simple SQL INSERT query. Adding a time.sleep(10) means that this process happens every 10 seconds. Assuming you run this without errors, you’ll find that your database will gradually be populated with temperature readings. You can check this using the mysql command line client again:

mysql --user=pi --password=raspberry

mysql> use temperatures;
mysql> select * from data limit 10;

+----+---------------------+-----------+-------------+
| id | timestamp           | sensor_id | temperature |
+----+---------------------+-----------+-------------+
|  9 | 2014-05-17 17:17:17 |         1 |        20.3 |
| 10 | 2014-05-17 16:45:10 |         1 |      22.187 |
| 11 | 2014-05-17 16:45:10 |         2 |      23.875 |
| 12 | 2014-05-17 16:45:21 |         1 |       23.75 |
| 13 | 2014-05-17 16:45:21 |         2 |      23.875 |
| 14 | 2014-05-17 16:45:33 |         1 |       23.75 |
| 15 | 2014-05-17 16:45:33 |         2 |      23.875 |
| 16 | 2014-05-17 16:45:45 |         1 |      23.812 |
| 17 | 2014-05-17 16:45:45 |         2 |      23.875 |
| 18 | 2014-05-17 16:45:56 |         1 |      23.812 |
+----+---------------------+-----------+-------------+
10 rows in set (0.00 sec)

If you have got that far, and are seeing data in the database, that’s excellent. In my next post, I’ll show you how to install pigpio and use it to flash an LED to give confidence that the process is working without having to check the database all the time. The next step is then to create a webserver which will show the data without having to log on to the Pi. Keep watching this space.

Warming to the Pi and Python

I’ve had a Raspberry Pi since Christmas, but haven’t done much with it up till now.  That’s all changed since I moved house.  Getting to grips with a new central heating system, I find that (for reasons too dull to list here) I need to monitor and control it rather more flexibly than I can do with the standard timer.  I also don’t want to pay lots of money for a Nest controller, however beautiful it may expect itself to be considered.  My first thought was to use an Arduino – the problem I’m addressing does not need a lot of processing power, and I don’t want to spend a lot – but I came up against a problem which seems to me to be the Arduino’s biggest failing at the moment.  I want to be able to control the system over WiFi.  There are WiFi shields for Arduino, of course, but they start at £30, twice the price of the Arduino.  I could invest in a Yun which has WiFi on board, but that’s £70.  There simply seems to be no cheap way of getting WiFi onto an Arduino – a device which is crying out to be connected to the internet..  Rob suggested I could use a bluetooth adapter and control the system from a phone; that has attractions and is very cheap, but means I have to be quite close to use it (no controlling the heating system from work, for example).  This made me turn to the Pi.  At first sight it seems ludicrous to use a full Linux computer for this trivial task, but consider: the Pi costs £23 (or less, for a model A), and the tiny USB WiFi dongle was only £9.  That’s £32 for the whole thing.  Much cheaper than an Arduino and WiFi combination. 

The hardware requirements of the system I want to build are quite simple.  It must monitor two or more temperature sensors, and it must be able to control two or three relays capable of switching mains voltage.  Having a couple of controllable status LEDs would be good, too.  I don’t require any other user interface input or display in hardware, because I want to use a web interface.  The Pi is more than capable of handling all this, but it turns out that there is quite a bit of software installation and configuration to do.  All of it can be done over a ssh connection (I use PUTTY for this), which is fortunate because I don’t actually have a monitor with an HDMI input that I can use as a display for the Pi.

The temperature sensors I am using are DS120B types.  Rather than being dumb devices like thermistors (whose resistance varies with temperature) they have logic inside which allows them to communicate with a host over a three wire digital serial connection.  The manufacturer calls this ‘1-wire’, because the actual communication only needs one, but you also need power and ground, so really there are three).  This has several benefits: it means that many devices can share the same wire (as each has its own address), it reduces the likelihood of inaccurate readings caused by long wires, and it means that the host does not need an analogue to digital converter to read them.  It also means that the host must implement the 3-wire protocol in order to read the sensors.  Fortunately, cleverer people than I have already done the hard work, and there are Linux drivers for these and other three-wire devices.  The relevant ones are already present in the raspbian distro, and simply need enabling.  Simply?  Did I say simply?  This is Linux – nothing is simple until you know the magic incantations.  I need Linux to load the appropriate 1-wire driver modules when it boots up, so it’s necessary to edit the /etc/modules file, which contains the list of drivers to be loaded.  It’s necessary to have root privileges to edit this file:

sudo nano /etc/modules

blog_pi.1

The lines w1-gpio and w1-therm tell the system to load the one-wire driver for the pi’s GPIO port (GPIO4 is used for this) and also the drier for the DS18B20 chip.  You can also load them manually from the command line if you wish:

modprobe wire
modprobe w1_gpio
modprobe w1_therm

Once these modules are loaded, the system will probe in the background to see if there are any sensors it recognises on the wire.  It will then create a new folder for each one under /sys/bus/w1/devices, and pop the data it retrieves from each in a file in there.  Reading the temperatures is a simple as reading the values in these files.  For example, as I write I have two sensors.on my Pi.  If I look in the devices folder, this is what I get:

blog_pi.2

The long number ‘28-0000009bba2b’ is the unique id of the sensor (hard-coded into the sensor itself).  The 1-wire driver identifies these and creates links to each.  By browsing to the folder and looking at the ‘w1_slave’ file within, the temperature is revealed.  it’s the bit saying ‘t=21250’, which is 1000 times the temperature in degrees Celsius (21.25 degrees).  If you want the temperature in Fahrenheit, I suggest you go back to the eighteenth century.  If you are actually executing the commands, you may notice a short delay between the ‘cat’ and the output.  That’s because when you read the file, the system actually contacts the sensor and takes a reading.  This is perhaps slower than you might expect, but it’s fast enough for my purposes.  For completion, here’s a quick wiring diagram of how to connect up the senor to the pi. the only other component needed is a 4k7 resistor, to tie the communication wire to the 5V line.

blog_pi.3

The next step is to use python to read these temperatures and store them in a database, then get a web-based interface up to control the relays.  I’ll also need to make sure that everything runs automatically, with no user intervention.  Watch this space.  Unless you have better things to do, which I fervently hope is the case.