Actions

Notes on Soundmodem: Difference between revisions

From HacDC Wiki

No edit summary
No edit summary
Line 178: Line 178:


By using soundmodem we'll be setting up an AX.25 network interface that, insofar as we're concerned, acts just like an ethernet interface (just crazy slow, 9600bps at best 1200bps likely).  We can use ifconfig, route, and ip to configure it, IPtables to filter traffic, and /proc/net/ipv4/* to tweak it (and set up packet forwarding). IPv6 is not currently supported on AX.25 in the linux kernel (~v3.8.x).
By using soundmodem we'll be setting up an AX.25 network interface that, insofar as we're concerned, acts just like an ethernet interface (just crazy slow, 9600bps at best 1200bps likely).  We can use ifconfig, route, and ip to configure it, IPtables to filter traffic, and /proc/net/ipv4/* to tweak it (and set up packet forwarding). IPv6 is not currently supported on AX.25 in the linux kernel (~v3.8.x).
= Successful connection achieved on 2013-06-8 =
I was able to connect two of my laptops together and get them pinging each other and transferring files over FTP. Here are some notes on how I accomplished it. [[User:Sitwon|Sitwon]] ([[User talk:Sitwon|talk]]) 21:33, 18 June 2013 (PDT)
== T520/TH-D72 ==
The first system was the easiest. A ThinkPad T520 running Slackware64-current connected to a Kenwood TH-D72. The TH-D72 has a built-in TNC. To get it up and running you need to edit `/etc/ax25/axports` to define a port, such as:
<pre>ax0    CALLSIGN  9600    255    2      your comment here</pre>
Next, turn on the radio, set it to the desired frequency, turn on the TNC to packet12, and connect it via USB. It should register as /dev/ttyUSB0 or similar. Using minicom issue the following commands to the serial device.
<pre>kiss on
restart</pre>
Then exit minicom and bring up the interface with kissattach: `sudo /usr/sbin/kissattach /dev/ttyUSB0 ax0 44.x.y.z` supplying your own (valid) IP address at the end. It will bring up the interface. Now you're done with this computer.
== R60/UV-5R ==
The second system I used was a ThinkPad R60 running Slackware64-current connected to a Baofeng UV-5R. In this case, I used soundmodem and a pair of audio cables to connect it. The UV-5R was just connected directly to the sound card, no PTT circuit. The PTT is triggered by VOX, so in the menu of the UV-5R set VOX to 1. Run soundmodemconfig (as root) to setup the device with the following settings:
''insert settings here''
Set up your `/etc/ax25/axports` file the same as above. Then, start soundmodem by running `sudo soundmodem /etc/ax25/soundmodem.conf`. Finally, run kissattach, the same as above, but with a different IP address and pointing to the soundmodem device.
Now you're almost ready to go. The last thing to do is to adjust your audio levels. The way I did it was to start listen (as described in the section on debugging) on both computers, generate some traffic with ping, and adjust the audio levels with alsamixer until both computers could hear and respond to each other's traffic. (Make sure your firewall is off!)
When you've gotten the packet loss as low as you can, you can try starting up some connections. I tried FTP and it was slow but it worked. You can also try to use kissparms to fine-tune some of the settings, but I'm not entirely sure how best to do that fine-tuning.
Expect the connection to be very slow.


= Other resources =
= Other resources =

Revision as of 04:33, 19 June 2013

Soundmodem emulates a hardware TNC in software, vis a vis a daemon running in userspace.

v0.18 is the latest (as of June 2013), I recommend using that. It now only depends on two things, audiofile and gtk2.

Soundmodem will either turn your soundcard into a KISS radiomodem (serial modem) or an AX.25 network interface. We're going to be using it to create AX.25 network interfaces (/sm[0-9]*/ or /ax[0-9]*/). It seems to work best if we put it into KISS mode and use kissattach to bring up the device.

Installation

From Source: Ubuntu Karmic and Lucid

wget http://www.baycom.org/~tom/ham/soundmodem/soundmodem-0.18.tar.gz
sudo apt-get install libasound2-dev libxml2-dev libgtk2.0-dev libaudiofile-dev

tar xvzf soundmodem-0.18.tar.gz
cd soundmodem-0.18
./configure && sudo make install
sudo mkdir /etc/ax25  ## or else it won't be able to save your configurations

Configure it by running (as root) /usr/local/bin/soundmodemconfig. It will open a window on your desktop with nothing in it. Click File, New, Configuration. Give your configuration a name (I used 'sm0').

apt-get: Debian and Ubuntu

Debian
Gets you version 0.10 on Lenny, 0.15 on Squeeze (6.0), 0.16 on Wheezy (7.0), and 0.18 on Jessie (testing) and Sid.
Ubuntu
Gets you version 0.13 on Lucid, 0.15 on Maverick and Natty, 0.16 on Precise (12.04) and Quantal (12.10), and 0.18 on Raring (13.04).

As root do the following

apt-get install soundmodem
mkdir /etc/ax25
soundmodemconfig /etc/ax25/soundmodem.conf                           # see below for configs
soundmodem /etc/ax25/soundmodem.conf -R -M >/dev/null 2>/dev/null &  # this makes an awful noise when it works

On Fedora

???

Arch Linux

sudo packer -S soundmodem ax25-apps ax25-tools

On Slackware (current?)

wget http://www.baycom.org/~tom/ham/soundmodem/soundmodem-0.18.tar.gz

if [[ "$ARCH" == "x86_64" ]] ;then
  LIBDIRSUFFIX="64"
fi
mkdir soundmodem-0.18-pkg

tar xvzf soundmodem-0.18.tar.gz
cd soundmodem-0.18
./configure --prefix=/usr --libdir=/usr/lib${LIBDIRSUFFIX}
sudo make install DESTDIR=../soundmodem-0.18-pkg
cd ../soundmodem-0.18-pkg
makepkg -l y -c n /tmp/soundmodem-0.18-$ARCH-1_local.tgz
sudo mkdir /etc/ax25  ## or else it won't be able to save your configurations

Configuration

IO Tab

On the I/O tab, choose your sound driver (out of soundcard, file, simulation, and alsa, I picked alsa). You'll have to play around with the value of the ALSA Audio Driver field to get one that works. On Windbringer, I used plughw:0,0. Check half duplex for now. Leave the Capture Channel field set to 'mono'. Set the PTT Driver field to 'none' so Soundmodem will assume VOX.

Haxwithaxe, Dan, and Mirage are working on PTT circuits for our radios. It hasn't been an easy process because the Baofengs seem to kick out just enough EMI to make the circuits act weird as-is. We have a few partial solutions at this time.

Channel Access Tab

Don't change the settings on the Channel Access tab.

Channel 0

Right below the name of your configuration in the tree you will see a new entry, Channel 0. If you select it you'll be presented with three new tabs on the right.

Modulator Tab

The possible values for the Mode selector are Off, fsk, afsk, pam, psk, and newqpsk. I picked fsk because it grants the highest effective bit rate (9600 bps). The values for Frequency 0 (1200), Frequency 1 (2200), and Filter (df9ic/g3ruh) were left set to defaults. Leave Differential Encoding checked.

Demodulator Tab

On the Demodulator tab (which governs decoding audio signals into bits) the options are much the same because they match the modulation scheme by default. This is what we want.

Packet IO Tab

On the Packet IO tab, the network/non-serial side of soundmodem is configured. Pick an IP address, netmask, and broadcast addres for a regular network interface. There are two options in the Mode selector, KISS and MKISS. I went with the default (MKISS). The network settings we used are:

  • IP address: 44.x.y.z (pick something)
  • Netmask: 255.0.0.0
  • Broadcast: 44.255.255.255

We're using the AMPRnet IP block which was set aside for amateur radio use. We should probably use the private subnet of that, 44.128/16 because it's akin to the RFC-1918 subnets and is used for experimentation in this area.

Audio settings

From alsamixer:

  • Master: 100
  • Headphone: 100
  • Speaker: 55
  • PCM: 55
  • Mic: 55
  • Mic Boost: 0

If these settings are any more quiet, the radios' PTT circuits won't trip.

We also suggest setting the volume to between 60% and 100%.

Seems to me like these settings are radio/soundcard specific. YMMV. Sitwon (talk) 21:08, 18 June 2013 (PDT)

Theory

When running the soundmodem utility (soundmodem /etc/ax25/soundmodem.conf &), the daemon configures the sound card to send and receive bitstreams encoded as audio through the jacks in the soundcard. It allocates a virtual network interface (sm0) which has a callsign associated with it which serves as the MAC address (relevant for ham radio, not so much for other applications) as well as an IP address, network mask, and broadcast address. The idea is that whatever packets are sent through that network interface will be transparently converted into sound and transmitted using whatever device is connected to the headphone and microphone jacks of the computer. Conversely, any data received by the broadcast device (ham radio, laser, etc) is demodulated into bits, encapsulated into network frames, and made available to the OS.

In theory, it's just like a regular network connection, with all of the dynamics you'd expect from IPv4. This means that we can do things like IP multicasting over these links. We discovered, through analyzing the code for a utility called easypacket that the Linux kernel treats AX.25 interfaces as real network interfaces, so much so that it sets up some iptables rules to prevent arbitrary traffic from being routed over them because there isn't much bandwidth. We'll have to add that to our rules. We tried setting up IPv6 addressing on some AX.25 interfaces and weren't able to.

Diagnostics, Testing, and Configuration

As root:

Note: you can connect two computers mic->headset to help check things are working before they hit the radio. ping and netcat (with -u option) are decent options for quick checks on rough sanity (the config not you :P). Here's how we did it:

  • Run a UDP server on one machine: ncat -n -l -v -u 5000
    • -n == No DNS resolution
    • -l == Listen for incoming connections (server mode)
    • -v == Verbose
    • -u == UDP
    • 5000 == port number to listen on
  • Connect to it from the other machine: ncat -u 44.x.y.z 5000
  • Type something. If it worked you'll see whatever the client typed showing up in the terminal on the server.
  • Run a TCP server on one machine: ncat -n -l -v 8080
  • Do the same thing for TCP (i.e., no -u options)

How to use listen from the ax25-apps distribution:

  • listen -a -c -r -t
    • -a == Monitor outgoing frames as well as incoming ones
    • -c == Colorize output
    • -r == Display in a human-readable format
    • -t == Timestamp every frame displayed

Keep the volume low at first to avoid accidental overload of the other machine. 50% for every channel is a good place to start.

Here's where you debug your configuration. It's going to take some flipping back and forth between tabs to get right the first time.

soundmodemconfig now pops open windows with error output.

The Diagnostics menu only appears when you have a channel (Channel 0) highlighted.

The first option is Scope, for oscilloscope. To get a test signal, click the PTT button (make sure your speakers are turned down!)

The second option is Spectrum, for spectrum analyzer of the signal being send to the sound card. To get a test signal, click the PTT button (again, make sure your speakers are turned down!)

The third option, Modem, dumps the raw bitstreams that your soundcard-emulating-a-modem is sending and receiving. Seeing strings of bits (even when disconnected) means that you have probably configured soundmodem correctly.

I do not know what the AO40-P3D option does. (It's an amateur satellite protocol thingy. We don't need it?)

To save your config file, use the File, Quit option. It'll save it automagically to /etc/ax25/soundmodem.conf.

So, how is this supposed to work, anyway?

Assumptions

  • Assume that you have a computer running Linux (or Windows if you dare) and soundmodem (aka PC). That has a pair of audio i/o:
    • headphone (PC speaker)
    • microphone/line in (PC mic)
  • Assume that soundmodem is configured correctly.
  • Assume that you have a transceiver (aka radio) - a black box that:
    • That can receive and broadcast signals
    • Has a pair of audio i/o:
      • speaker/headphone (aka radio speaker)
      • microphone (aka radio mic)
  • Assume a method of PTT as described below is setup.

PTT

Further assume that there is a mechanism to switch the radio from receive to transmit. Known here as PTT circuit for the purposes of brevity. This may be:

  • the built in VOX circuit
  • an external VOX circuit
  • or a serial PTT circuit

Whenever a signal is sent to the PTT circuit, the black box stops receiving and transmits whatever signal it gets through its microphone jack. Alternatively two radios one transmitting only and one receiving only (possibly on different channels) such that there is no need for drivers or a switching mechanism.

Hook-up

  • Connect radio speaker to PC mic.
  • Then connect PC speaker to radio mic.
  • If applicable: Attach the PTT circuit as appropriate.


Whenever a packet (say an ICMP4 echo request) is transmitted to the IP address of the AX.25 interface on PC, it is converted by soundmodem into modulated sound (aka audio data) which goes out through PC speaker to the radio mic. The PTT circuit engages and the radio stops receiving and begins transmitting. The radio receives the audio data and transmits it. When the transmission is done the PTT circuit disengages and the radio stops transmitting and begins receiving. Conversely, whenever the radio receives a signal, it converts it into audio data, which then goes through radio speaker into PC mic. After the audio data is received by the PC soundmodem demodulates it into bits, packages them into IP packets, and drops it into the networking subsystem of the kernel.

By using soundmodem we'll be setting up an AX.25 network interface that, insofar as we're concerned, acts just like an ethernet interface (just crazy slow, 9600bps at best 1200bps likely). We can use ifconfig, route, and ip to configure it, IPtables to filter traffic, and /proc/net/ipv4/* to tweak it (and set up packet forwarding). IPv6 is not currently supported on AX.25 in the linux kernel (~v3.8.x).

Successful connection achieved on 2013-06-8

I was able to connect two of my laptops together and get them pinging each other and transferring files over FTP. Here are some notes on how I accomplished it. Sitwon (talk) 21:33, 18 June 2013 (PDT)

T520/TH-D72

The first system was the easiest. A ThinkPad T520 running Slackware64-current connected to a Kenwood TH-D72. The TH-D72 has a built-in TNC. To get it up and running you need to edit `/etc/ax25/axports` to define a port, such as:

ax0     CALLSIGN   9600    255     2       your comment here

Next, turn on the radio, set it to the desired frequency, turn on the TNC to packet12, and connect it via USB. It should register as /dev/ttyUSB0 or similar. Using minicom issue the following commands to the serial device.

kiss on
restart

Then exit minicom and bring up the interface with kissattach: `sudo /usr/sbin/kissattach /dev/ttyUSB0 ax0 44.x.y.z` supplying your own (valid) IP address at the end. It will bring up the interface. Now you're done with this computer.

R60/UV-5R

The second system I used was a ThinkPad R60 running Slackware64-current connected to a Baofeng UV-5R. In this case, I used soundmodem and a pair of audio cables to connect it. The UV-5R was just connected directly to the sound card, no PTT circuit. The PTT is triggered by VOX, so in the menu of the UV-5R set VOX to 1. Run soundmodemconfig (as root) to setup the device with the following settings:

insert settings here

Set up your `/etc/ax25/axports` file the same as above. Then, start soundmodem by running `sudo soundmodem /etc/ax25/soundmodem.conf`. Finally, run kissattach, the same as above, but with a different IP address and pointing to the soundmodem device.

Now you're almost ready to go. The last thing to do is to adjust your audio levels. The way I did it was to start listen (as described in the section on debugging) on both computers, generate some traffic with ping, and adjust the audio levels with alsamixer until both computers could hear and respond to each other's traffic. (Make sure your firewall is off!)

When you've gotten the packet loss as low as you can, you can try starting up some connections. I tried FTP and it was slow but it worked. You can also try to use kissparms to fine-tune some of the settings, but I'm not entirely sure how best to do that fine-tuning.

Expect the connection to be very slow.

Other resources