Alternet

Flight log results in Google Earth

I’ve been working on my private pilots license for the last few months.  The primary plane that I’ve been using is the Cessna SkyCatcher.  I’ve included a photo of an example of this plane below.  It’s a fun aircraft, other than the fact that two normal-sized adult males can’t take full fuel. Read more…

Share on Facebook

R820t tuner on a rtl-sdr compatible dongle, from eBay seller CosyCave

Relating to the rtl-sdr work that has been done, the E4000 tuner was the standard barer for a long time.  However, Elonics has discontinued this part, and it’s becoming difficult to find.  The popularity, and scarcity, of this part has encouraged sellers to offer products claiming to be built with the E4000 and are not.  Luckily, someone discovered the code for using the R820t tuner in the Linux V4Lin drivers.  They ported this code into the rtl-sdr source maintained by osmocom.

I just finished porting their code into Cocoa Radio.  Now, it’s possible to use my software with both the E4000 and the R820t.  On startup, Cocoa Radio will automatically detect which tuner you’re using and perform the appropriate actions.

It did take a little while to finish this work, and there are several more tuners out there.  If you are desperate for support of a specific tuner, you can donate a device for the cause and I’ll try to support it.  By the way, Softshell uses the same code for tuning as Cocoa Radio, if you recompile softshell, it should include this new code.

All the relevant code and binaries are, as usual, available at github.  Make absolutely sure that you also update the softshell repository!

Share on Facebook

Cocoa radio interface explained (click for full size)

Well, I’m back from vacation and I want to tell everyone about a new version of Cocoa Radio (my application for demodulating radio signals using the rtlsdr dongles on mac os x, written in Objective C).  This version seems to be running really well.  I’ve set the sample rate to 1024000 samples per second for the moment (though this value can be changed in the code), and at this rate everything seems really stable.  Please give it a try and create issues at the github issues page if you find any problems.  I should say that I’m a little tired of working on it, so unless there are major issues I’ll be working on other projects for a while.  I encourage others to take a look at the code if they’re interested in SDR.  It’s not as scary as it looks!

Also, the sliders are a little buggy (especially the bandwidth ones).  Move the a little bit once the app starts up and they’ll work correctly.

Finally, I don’t have any support for AM (amplitude modulation) yet.  It’s an easy modulation type, and I may add it soon.

Share on Facebook

I’ve been working on a stand-alone software defined radio (SDR) application for Mac OS on-and-off for a few months, and I think it’s good enough now to encourage people to test it and provide feedback.  Though it’s possible, of course, to do everything my app does in GNU Radio, it’s much more convenient to have a dedicated app.  Especially because this app uses a lot less CPU than GNU Radio.

This video shows the basic operation of the user interface:

And this one shows an ‘inside peek’ at what happens to the signal inside of the application. In normal usage, the waterfall display won’t do this, but it’s an interesting effect:

There are still many, many bugs, but it should be enough to play around with. I’ve been able to listen to broadcast FM radio for some time using the app. There is no squelch control, that’s on the list of things to add. Also, it’s possible to get audio buffer underruns. It’s likely caused by slight differences in the clock rate of the rtl-sdr dongle and the audio device that you’re using.

If you notice any bugs, or have specific issues that you would like addressed, please create an issue on the github page.  Also, if you are able to contribute, please let me know.  I’m obviously in need of some GUI assistance!

One word of advice, don’t try to change the modulation type using the drop-down menu, it doesn’t work!

Again, code is available at github, as is an application binary.

Share on Facebook

This tree randomly split in two and fell over!  No one was hurt, it could have been muuuch worse.

Share on Facebook

Fixing all the little issues with this process took me days, so I apologize if it’s a complicated and multi-step process.  These steps worked for me, hopefully they work for you too, but YMMV.

First of all, it is vital that you don’t have the macports boost 1.50 installed.  There is a problem with that version where the x86_64 version of the library isn’t compiled.  This is mentioned in an earlier post, and the mac ports trac entry is here.

The easiest way to begin is to install boost 1.49 then run “sudo port install gnuradio-core” and let macports install all of the pre-requisite packages (you may need to perform the fix I mentioned in that earlier post to fix netpbm) .  When it finally gets to gnuradio-core, it will fail.  Now, what you need to do is:

$sudo port edit gnuradio-core

Follow the instructions, again in that earlier post (update 2).

$sudo port clean gnuradio-core
$sudo port -n install gnuradio-core
---> Computing dependencies for gnuradio-core
---> Fetching archive for gnuradio-core
---> Attempting to fetch gnuradio-core-3.3.0_0+python26.darwin_12.x86_64.tbz2 from http://packages.macports.org/gnuradio-core
---> Fetching distfiles for gnuradio-core
---> Verifying checksum(s) for gnuradio-core
---> Extracting gnuradio-core
---> Applying patches to gnuradio-core
---> Configuring gnuradio-core

THIS IS IMPORTANT!: Cancel (control-c) when it says “Configuring gnuradio-core.”  At this point, we need to hand-edit the configure script in the gnuradio source directory.  The reason for this is because some of the assembler code in gnuradio uses 32-bit only opcodes.  When compiling for 64-bit machines they generate errors.  It’s necessary for them to be compiled differently.  Luckily, when Lion was released, a fix was devised and added to macports.  The same exact fix (in principle) should work for Mountain Lion.  But, in the configure script, the change looks for Lion and doesn’t detect Mountain Lion.  We just need to change the test to detect Mountain Lion.  The difference is only the version of darwin used.  This information is in this trac.

$cd /opt/local/var/macports/build/
$ls
<cd to the long directory that ends with science_gnuradio-core>
$cd gnuradio-core/work/gnuradio-3.3.0
$sudo vi configure

Once editing the configure script, search for “darwin*10*” or “darwin*11*”.  This is easy if you hit the forward slash and type “darwin\*1″:

/darwin\*1

The region of interest should look like this (the numbers are the line numbers):

20154     *darwin*11*) 
20155 # The cast to long int works around a bug in the HP C Compiler 
20156 # version HP92453-01 B.11.11.23709.GP, which incorrectly rejects 
20157 # declarations like `int a3[[(sizeof (unsigned char)) >= 0]];'. 
20158 # This bug is HP SR number 8606223364.

Change the *darwin*11* (or *darwin*10*) to: *darwin*12*

Close the vi session by hitting <esc> then : then type wq and enter.

Now, run “sudo port -n install gnuradio-core”.  Make sure that you DO NOT clean the package.  This will destroy our edited configure script.

When that finishes (hopefully it does!) you should be all set!  You’ll probably want to install gnuradio-companion as well as gnuradio-audio-osx.

If you have any problems or questions, let me know in the comments.

Share on Facebook

It has been a little while since I released the very early code for softshell (not that the code has advanced much), and I’ve received a few requests for a bit more information about how it’s intended to be used.

I admit that I hacked it together very quickly so that I could make some basic use of the rtlsdr dongles on my mac.  To be very clear, Softshell does no actual SDR itself.  You can really look at it more like a driver for the rtlsdr.  Softshell opens a connection to the rtl device, allows you to tune its internal oscillator, and puts the data on the network.

To start, install the rtl device in your USB port, then open Softshell.

Freshly opened

If you see a similar window, it means that Softshell has found your device (the ezcap in this case).  It is, perhaps, a good time to mention that I’ve only ported the tuner code for the Elonics E4000 tuner.  Click the “Open” button to have the program open the connection to the device.  If it detects that you have the E4000 the “Tuner type” field will be filled in with “Elonics E4000.”

Once this is done, changes to the sample rate and center frequency will take effect with the “Update” button is clicked.  The Center frequency is provided in Hz.

Now, that’s all fine and good, but you’re just tuning the device.  To actually get the data out of it, you need to setup the network settings.  Choose a port number for Softshell to listen to, I use “12345,” and click the “Running” checkbox.

Finally, in GnuRadio, you need to use a “TCP Source” block setup as a client with the same port number you used before.

Setting up GnuRadio to work with Softshell (click for full size)

Once that’s done you should be up & running.  Note that, natively, the rtl device actually outputs unsigned bytes and that Softshell converts these to floats centered around zero.  Some GnuRadio examples include the blocks that perform this conversion.  If you come across this, just remove those blocks.

Good luck!  Please feel free to comment with any questions or issues!

Share on Facebook

I’ve been struggling with this for a few days, so hopefully this post saves some people from headaches.

Some packages in macports fail to compile using CLANG.  The reasons are varied, but inline assembler is a common reason.  The trick to getting them to work is using a different compiler.  LLVM-GCC seems to be a good choice.

The macports guide for choosing compilers isn’t as clear as I would like.  After some googling, I learned that you can edit the port file using macports.

sudo port edit netpbm

This command opens the port file for that package in vi (or maybe your default editor, if different) and lets you make changes.  The compiler is set by the TCL variable ${configure.compiler} which is set thusly:

configure.compiler llvm-gcc-4.2

Once this changes is made, clean and re-attempt the port:

sudo port clean netpbm

sudo port install netpbm

And it should work.  It did for me.  At least, you can use these series of steps to try different compilers out to see which one works.

Update:

To compile the gnuradio-gruel library, it may be necessary to downgrade to boost version 1.49 if you have version 1.50 installed.  For some reason, the x86-64 version of the libraries aren’t build with the new version of boost.  Here is the trac entry.

Update 2:

While compiling the gnuradio-core package, a failure occurs.  The log entry is, unfortunately, not helpful:

:info:build make[6]: Leaving directory `/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/filter'
:info:build make[5]: *** [all] Error 2

However, if you navigate to that folder and run make (as sudo, because the files are owned by macports) you get a more helpful result:

libtool: compile: /usr/bin/clang -DHAVE_CONFIG_H -I. -I../../../.. -I/opt/local/include -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/runtime -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/general -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/general -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/gengen -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/gengen -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/filter -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/filter -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/missing -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/reed-solomon -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/viterbi -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/io -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/g72x -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/swig -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/hier -I/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_science_gnuradio-core/gnuradio-core/work/gnuradio-3.3.0/gnuradio-core/src/lib/swig -I/opt/local/include -I/opt/local/include -I/opt/local/include -I/opt/local/include -I/opt/local/include -pipe -O2 -arch x86_64 -MT float_dotprod_sse.lo -MD -MP -MF .deps/float_dotprod_sse.Tpo -c float_dotprod_sse.S -fno-common -DPIC -o .libs/float_dotprod_sse.o
float_dotprod_sse.S:57:2: error: unknown directive
 .version "01.01"
 ^
make[1]: *** [float_dotprod_sse.lo] Error 1
make: *** [all] Error 2

It’s about here that you realize that you’re dealing with assembler.  Ick.  Apparently, the .version directive isn’t supported.  I tried commenting them out, but when gnuradio is finally built, I get SEGFAULTs from the filter methods.  I don’t think it’s a coincidence that we’re in the filter directory and this is the dotproduct source file.

Here, again, I tried the compiler ‘trick’.  No dice.  This time, however, the error changed:

float_dotprod_sse.S:63:suffix or operands invalid for `push'
float_dotprod_sse.S:163:suffix or operands invalid for `pop'

A quick google later, and it turns out that when you compile assembly code designed for a 32 bit machine on 64 bit, the assembler doesn’t know what to do with some of the instructions.  The solution is to add -m32 to the compile command line.

Unfortunately, all this code is compiled with -arch x86_64 (the opposite of the -m32 flag!).  If I removed -arch x86_64 and added -m32 it worked.  Now what to do?  Unfortunately, I’m stuck for now… ideas?

Share on Facebook

“Well, there’s your problem!”

This is waaay out of left field for this blog, but I’ve spent the last week or so working on this project, so I figure I should describe it.

In the list few weeks, my engine in my ride-on power has been burning tons of oil.  It would smoke a bit all the time, but after a while, it would bog down and unleash a massive cloud of blue smoke.  I read some forums and things online to try to get an idea of what was happening.  Many people said that it was caused by the crank case breather.  I figured this would be an easy thing to try.  When I went to the small engine store (Willamette Saw) they said that it was almost guaranteed to be the head gasket.  Bummer.  I came back the next day with the engine model number, and got a new gasket.

You can see the tons of oil swirling around in the air filter housing.  What I think was happening is that the head gasket leak was pressuring the crank case, which was causing the blower to expel lots of air and oil into the housing.  When it overflowed, it would pour into the intake causing the huge smoke cloud. Read more…

Share on Facebook

Keynote time!

I’ve never been less impressed by fo-hawks than I am right now.

The Stig!?

Share on Facebook