Getting GPS to work on a Raspberry PI

One of the tasks I want to use a Raspberry PI for is to take over the duties of an existing ITX based linux box running my weather station. Now in theory that should be pretty simple as the current setup uses pywws to connect to the station and as that’s written in python it should work.

Now the Raspberry PI has no onboard Real time clock – which means it needs to use an NTP server to get the time when it starts. Usually you would use the default settings and allow the PI to connect to thenet for it’s time. Now this is fine if you have a working net connection but what if you are not connected to the net? You might be in the field running the PI on batteries.

As the other projects I have lined up for it is to connect my Meade LX200GPS telescope to the local network or to work with my (in prototype) radio telescopes so having an accurate clock is going to be required.

Now the obvious solution here is to use GPS as a time source. GPS works by having a constellation of satellites in orbit and each one carries a highly accurate atomic clock & broadcast both their current position and the time. A GPS receiver then receives these signals and, as long as it has enough satellites and workout where you are by comparing the times from those clocks.

So this article shows how to use A GPS receiver with the Rasperry PI – although these instructions are not specific to the PI.

The hardware

For this experiment I’m using a USB GPS receiver from Maplin – product code A73KF. I bought this several months ago when they had it on special offer for £19.99 – it usually retails for £29.99.

Raspberry PI with the A73KF GPS receiver plugged in

Now it comes with a CD for Windows machines but we don’t need it – as the majority of GPS receivers I know of use serial & this is no exception. When plugged in it appears as a serial port.

Plug it in and run lsusb

pi@raspberrypi:~$ sudo lsusb
Bus 001 Device 003: ID 0424:ec00 Standard Microsystems Corp.
Bus 001 Device 004: ID 067b:2303 Prolific Technology, Inc. PL2303 Serial Port
Bus 001 Device 002: ID 0424:9512 Standard Microsystems Corp.
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

There the Prolific Technology entry is the GPS appearing as a serial port. If you look in /var/log/syslog you will also notice it will have created the port as /dev/ttyUSB0 as it’s the first serial port.

Using the PI as a GPS Receiver

Now the next step is to get the pi receiving data from the satellites. Now there is a suite of tools available for Linux called gpsd which we’ll install:

pi@raspberrypi:~$ sudo apt-get install gpsd gpsd-clients python-gps

Next we need to start the daemon:

pi@raspberrypi:~$ sudo gpsd /dev/ttyUSB0 -F /var/run/gpsd.sock

Ignore any messages from the console or in the log files, you may see it complaining about IPv6 but you can ignore that.

Viewing whats in the sky & your location

Now GPS doesn’t work indoors – as it needs a clear view of the sky so for this I’ve placed the PI on the window sill. Next I ssh into the pi and run cgps.

pi@raspberrypi:~$ cgps -s

The -s flag is there to tell the command not to write raw data to the screen as well as the processed data.

You should then get the following output:

┌───────────────────────────────────────────┐┌─────────────────────────────────┐
│    Time:       2012-06-18T15:05:10.0Z     ││PRN:   Elev:  Azim:  SNR:  Used: │
│    Latitude:    51.231848 N               ││  14    43    249    40      Y   │
│    Longitude:    0.514014 E               ││  25    75    283    37      Y   │
│    Altitude:   132.3 m                    ││   2    26    085    31      Y   │
│    Speed:      0.0 kph                    ││  12    56    070    18      Y   │
│    Heading:    0.0 deg (true)             ││   9    19    133    22      Y   │
│    Climb:      0.0 m/min                  ││  27    09    133    17      Y   │
│    Status:     3D FIX (1 secs)            ││   4    17    045    31      Y   │
│    GPS Type:                              ││  32    05    321    20      Y   │
│    Longitude Err:   +/- 8 m               ││  29    41    192    18      Y   │
│    Latitude Err:    +/- 9 m               ││  31    28    304    42      Y   │
│    Altitude Err:    +/- 27 m              ││                                 │
│    Course Err:      n/a                   ││                                 │
│    Speed Err:       +/- 68 kph            ││                                 │
│                                           ││                                 │
│                                           ││                                 │
│                                           ││                                 │
│                                           ││                                 │
│                                           ││                                 │
└───────────────────────────────────────────┘└─────────────────────────────────┘

Here you can see it’s receiving from 10 satellites and it has the time and your location. The 3D FIX section tells you it has enough data for a 3D fix on your location (i.e. altitude). The Err lines tell you the error in your position. If you leave it running you should see the Err values change every second or so.

Viewing GPS under X-Windows

Now above I showed how the GPS looks from an SSH connection but you can get a graphical display as well using the xgps client thats also been installed. Now if you have a monitor connected to the pi simply open a terminal and run xgps. However as I’ve not got a monitor against the window I’ve used ssh to connect to it from another machine. To get this to work you need to add -Y to the ssh command.

peter@somehost:~ $ ssh -Y pi@raspberrypi
pi@raspberrypi:~$ xgps

You should now get a window like the following open on your local machine – don’t worry if it takes a little while, it might take a second or two:

xgps running on a PI but being displayed on Mac OS-X

Setting the computer time using GPS

Now we have a working GPS we can now get the PI to use it for setting the time. To do this we need to configure ntp to use the GPS satellites as a time source. Now you should already have ntp installed but if not then you need to install it:

pi@raspberrypi:~$ sudo apt-get install ntp

Next you need to edit the file: /etc/ntp.conf and add a few lines to it defining the GPS. This can be either before or after the existing lines beginning with server:

# gps ntp
server 127.127.28.0 minpoll 4
fudge  127.127.28.0 time1 0.183 refid NMEA
server 127.127.28.1 minpoll 4 prefer
fudge  127.127.28.1 refid PPS

Now restart ntp:

pi@raspberrypi:~$ sudo service ntp restart

Now if you query the server you should after a while see it synchronize:

pi@raspberrypi:~$ ntpq -p
remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*ns1.luns.net.uk 33.117.170.50    2 u   54   64    7   65.454    2.666   5.800
+resntp-b-vip.lo 127.151.91.34    3 u   45   64   17   55.704   -5.169   8.482
+bart.nexellent. 194.242.34.149   2 u   17   64   17   76.585   -4.271  57.595
+v01.s01.be.it2g 193.190.230.65   2 u   20   64   37   86.464   -2.374 228.460
xSHM(0)          .NMEA.           0 l   11   16  377    0.000  144.714   3.026
SHM(1)          .PPS.            0 l    -   16    0    0.000    0.000   0.000

A couple of notes:

You might find that ntp doesn’t connect to the gps at first. It appears that it starts gpsd up without the link to the serial port. What I find I have to do is:

pi@raspberrypi:~$ sudo killall gpsd
pi@raspberrypi:~$ sudo gpsd /dev/ttyUSB0 -F /var/run/gpsd.sock
pi@raspberrypi:~$ sudo service ntp restart

Once I’ve done this then after it gets a fix then it starts working. Sometimes running cgps and waiting for it to get a fix also fixes this.

I don’t know why this happens but it’s an issue I’ve yet to solve.

Installing latest mercurial when Ubuntu or Linux Mint repos don’t have it

Ah I just hit an interesting problem with Mercurial. I had a repository which had been created with a recent version however I had to restore it from a backup onto another machine however I couldn’t commit to it as it’s format wasn’t supported:

$ hg st
abort: requirement ‘dotencode’ not supported!

The reason is that the dotencode format was only introduced into mercurial in version 1.7 so if you’re running an earlier version then you’ll get this error. Ok so upgrade – problem is that according to apt I had the latest version – 1.6.3 – erm nope thats not the latest.

Thankfully the fix is simple, all you need is to ass the correct repository first before installing the latest version

sudo add-apt-repository ppa:mercurial-ppa/releases
sudo apt-get update
sudo apt-get install mercurial

Once you’ve done that then you’ll find mercurial will work again as expected.