Working on SDR & GNURadio

[Terry Fox]

The latest from the Charleston chapter:

The Charleston hamfest was yesterday.  Temp was approx. 65deg.  An SDR demo
was set up, which was based on my Brainerd demo transceiver that I had at
Manassas.  Using that, I split the audio IF output, and fed one copy to
Rocky on a Windows laptop, and the other to a second laptop running GNURadio
under Linux.

There was some interest in SDR, a couple people asked where to go for more
info.  I gave them the AMRAD web site, but told them that we are in the
process of creating info there, so also do a Google search.

Since John Schwacke is using GNURadio to test his A/D board, Mel dragged me
into trying to use GNURadio instead of DttSP.  It is kind of overwhelming at
first, but the instructions on prerequisites and installation are mostly
correct.  It took me a frustrating week to make any progress with it.  But,
the second week (last week), was less so.  One nice thing is that there is a
GUI provided (grc) that allows you to actually build a radio using visual
building blocks for the necessary modules.  For example, I had an AM
demodulator working with a SoftRock (or Brainerd board) in one night.  I
found a general flow diagram for a Weaver method SSB demod, and got that
basically running in a day.  Then it took a few more days to fine tune the
design, and learn more about how each module works.

The documentation for GNURadio is virtually nonexistent, and I am still
trying to learn where to go to understand the individual modules, and how
they all interact.  The grc GUI mentioned above does not have all the proper
modules necessary for a complete, useful radio design, so adding to it's
list of modules is also something to figure out.  The DSP part of GNURadio
is implemented in C/C++, but the GUI stuff is in Python, with a little XML
thrown in to assure total confusion.

I know GNURadio is used for the USRP project, and Maitland has that
boardset.  I don't know if you have messed with it very much Mait.  My
impression is that it's better for VHF and UHF than HF (12-bit A/D with
little other "RF" hardware).

So, those with Linux can start playing around with GNURadio
(www.gnuradio.org).  Go there, download the files (either tarball or svn
checkout), then follow carefully the installation instructions.  The only
problem I have is that it cannot find certain library files, I had to add an
export variable to my /etc/profile file.  Plus, one of the necessary
components just updated from version 38 to version 39.  No breakage occured
however.

I have ordered the active components to build two more of John's A/D boards,
he has only two more bare boards left (one each for Mel and Terry).  I
wasn't sure until yesterday of the passive components sizes, theya re being
ordered tomorrow.  I'm creating a parts list for the board, which I will put
on the AMRAD web site.  Once the design is further proven with these two
boards, John is willing to have more boards made.  He uses the stencil and
skillet method of soldering, which looks better than my hand-soldering.  I
should have a board running well before my trip to DC in June, and hope to
demo in DC then.

I also have the HPSDR four-board-set (Janus/Mercurey/Penelope/Atlas) for a
complete direct-sampling HF radio.  An RF PA kit is on order, and I need to
make a TR switch and a few other goodies.  So, that project is also slowly
underway.

Plus, I still hope to work more on the DttSP GUI that I was writing, mostly
to get more comfortable with QT GUI design.  The DttSP gurus (Frank and Bob)
are changing the interfacing to DttSP from named pipes to UDP, which should
actually make it easier to interface to, especially remotely.  I am
controlling the DttSP, and reading meter data.  Next, I need to get the
spectrum data from DttSP and display it, then I should be over the major
hurdles.  Generating the graphic display is not the issue, getting the data
from DttSP in an acceptable manner is...  I just need to get over my
internal mental stoppage on that front.

The HPSDR Verilog course is interesting.  I haven't actually run any
simulations yet, but I have loaded many of the tools - Altera, Xilinx, and
others.  Maybe in my spare time!!  I know Frank is watching it.  I have
videtaped the first three lectures, but they are also available directly on
the web.  John writes in Verilog, which is also what the HPSDR and Phil
Covington's FPGA code are in.  So, Verilog it is.

I have added a couple things to the AMRAD SDR web site, but plan to do much
more.  My next step will be to include some of this information (along with
links).  No, I don't plan a Wiki, I'm a builder, not a documenter
(unfortunately).

So, in summary, I am slowly moving in the direction of direct-sampling SDR
designs instead of QSDs/QSEs.  I still plan to play with, and support the
QSD/QSE, but I have proven to myself that there are inherent limitations.
Mostly related to keeping the IQ phase and gain balanced across a spectrum
swath.  Yes, you can "tune" the radio (if your radio is not xtal controlled)
to reduce these effects, but the direct-sampling design just seems
"cleaner".  Unless you live in Europe directly under an HF broadcaster, that
is.

Bottom-line is let's get going on SDR!!!  There are plenty of things to do,
with fun and/or frustration to be had!!!!

More later.
Terry WB4JFI