FW: SDR Software and hardware

Terry Fox tfox at knology.net
Fri Oct 10 16:16:31 CDT 2008

-----Original Message-----
From: Terry Fox [mailto:tfox at knology.net]
Sent: Friday, October 10, 2008 3:48 PM
To: taco at amrad.org
Cc: Terry Fox
Subject: SDR Software and hardware

Hey guys.  Are we doin' anything as a group?  The election season is almost
over, it's time to be thinking about what to talk about at physical Tacos
and this forum besides politics, ancient radios, and the latest humorous
finding on the Internet.  How about new radios?

I see that the HPSDR group is about to build the Mercury boards for sale.
This is a 0-65MHz direct-sampling SDR board that plugs into the HPSDR Atlas
bus, and uses Ozy for USB communications to the host computer.  It is meant
to be the receiving companion to the Penelope direct-sampling exciter board.
The cost for Mercury will be about $330 or so, through TAPR.  Sort of like
Phil Covington's QS-1R.  The QS-1R looks very good, btw, but the cost is a
little steep for me right now.

My personal opinion is that while the QSD/QSE-based radios (SoftRocks,
Brainerd, Flex products) are good, the direct-sampling devices will
represent a true leap in technology.  They virtually eliminate the image and
LO radiation (leakage) of most or all QSD/QSE designs.  Even the recent Flex
radios, which I believe are still QSD/QSE.  Note that even the latest
versions of PowerSDR does not do receive I/Q balance at multiple points on
every band.  The same with transmit.  Can you assure that your transmitted
RF meets the FCC requiements all across the band with only a single IQ
sample/balance point?  Both the QSD/QSE hardware and the sound card response
contribute to this imbalance issue.

With 16-bit 130MHz (or higher) RF A/D converters, the hardware can deliver
at least 14-bits of accurate samples.  Then, processing gain due to digital
down-sampling can provide more bits.  Yes, there are occasions where the
dynamic range is still not quite good enough, but those conditions are
getting few and far between.  the IQ balancing is not an issue, because this
process is done via firmware inside the FPGA on digital numbers.  No
real-world analog components to be out of balance at different frequencies.

Also, Mel and I are playing with SDR software under Linux (and maybe
Windows).  We are learning to use the QT GUI software, along with a couple
of IDEs.  QT is supposed to work on both Linux and Windows, so the same
basic code would be usable on both platforms.  Reality is that while QT will
build the same GUI on both platforms, some of the IO will need to be
different, due to the different approaches Windows (portaudio) and Linux
(jack) takes.  We are using the no-cost, open-source version of QT, which is
full-blown but does not include company support.  See www.Trolltech.com

Eclipse (www.eclipse.org) is an IDE that can have QT integrated into it, and
is also available for both Linux and Windows.  However, we have found some
inconsistencies with the most recent combination of QT/Eclipse integration

While waiting for these inconsistencies to be sorted out, I have started
playing with QT on Linux, with a couple of other IDEs:  Monkey Studio and
EDYUK.  Both of these are available at no cost from www.qt-apps.org, under
the development section.  Both requiring compilation - via QT ironically.

It is important to note that when using the IDEs mentioned above, they all
use the same project source code tree, and do NOT require additional special
files.  So, a project can actually be built with any of the three, then
another used to work on it, depending on the programmer's whim.  All three
also have full source code.  I have not tried Monkey Studio or EDYUK under
Windows yet, so I cannot guarantee they are cross-platform.  But, QT and
Eclipse behave identically on Linux or Windows.

Someone here in Charleston is also designing a smaller/less expensive
direct-sampling RF A/D board that plugs into the Digilent Nexys-2 FPGA demo
board.  He has it undersampling 2 meter activity and displaying the RF
spectrum in a window, and is working on demodulating the signals now.  This
is a neat board, not cutting-edge regarding performance (12-bit A/D I
think), but a great way to learn FPGA SDR software.

Is anybody else in AMRAD-land interested in working on SDR software?  Care
to join us?


More information about the Tacos mailing list