TAPR High Performance Software Defined Radio

WB4JFI wb4jfi at knology.net
Wed Jan 7 00:58:44 CST 2009

Please stand by, a long message follows:

I received my Mercury board just after XMAS.  Now, I'm getting ready to fire
it (and Penelope exciter) up shortly.

I've ended up with two HPSDR setups: one Atlas/Janus(sound card
replacement)/Ozy, and one Atlas/Ozy/Mercury/Penelope.  The first one is used
for SoftRock and other QSD/QSE experimentation.  The second boardset will be
built up as a stand-alone rig.

I firmly believe that while QSD/QSE rigs are low-cost and fun, but as the
cost of A/D converters that can do RF continues to drop and performance
increases, we will eventually switch to direct digital up (DUC) and down
(DDC) conversion SDR radios.

The 16-bit, 135Ms/s A/D on Mercury is the same one that Phil Covington uses
on the QS-1R.  With the processing gain that down-sampling and other DSP
techniques provide, we are already very close to the complete dynamic range
necessary for very good HF radio that is almost all digital.  Both the QS-1R
and the Mercury are proving to be excellent performers at first blush.  If I
remember, Joseph Mitola III recommends about a 130dB dynamic range for HF,
which translates to about 22 bits.  Throw out a bit or two (as the A/D LSb
is usually not reliable), and a 24-bit system is about the minimum.

If you look at it from the perspective that you need 130dB dynamic range,
and you can get about 90dB from the 16-bit A/D, you need another 40dB of
dynamic range without playing with front-end Variable Gain Amplifiers, or
switching pads and/or preamps in and out.  That's a tall order for the DDC,
but is close to feasible.

There's an oft-quoted formula for processing gain:
10log(Nyquist_BW/output_BW) (my words).  In a note regarding the SDR-14
(14-bit A/D DDC receiver), which samples at 66.666MHz (Nyquist=33.333MHz)
and output bandwidth of 100kHz, the example is 10log(33.333MHz/0.1MHz) =
25.2dB of processing gain.  Narrower bandwidths provide even more gain.  So,
90dB of dynamic range from the A/D itself, plus 25dB of processing gain gets
us to 115dB overall gain.  Not too bad for a receiver covering a 100kHz
swath of a band.  And yes, as both the QS-1R and the Mercury are proving,
you can design RF front ends to approach or achieve the 16-bit noise limit.

However, the DDC/DUC radios cannot compare to the cost-effectiveness of a
SoftRock.  For limited funds (or for great fun), get a SoftRock from Tony
Parks.  I LOVE SoftRocks.  For a higher-end, build-it-yourself, HF rig that
will be the wave of the future, get either the HPSDR boardset, or Phil's
QS-1R and the upcoming preamp board and matching transmit exciter (QS-1T?).

Someone here in Charleston has also built-up a direct-digital conversion
receiver using a 80Ms/s, 12-bit A/D, preceeded by a VGA, and followed by an
ASIC that can control the VGA plus do some digital down-sampling.  A
Digilent Nexys II FPGA demo board does the rest of the down-sampling and
interfacing to the host PC.  He has it receiving both 40M (directly) and 2M
(via subsampling).  The software he is using is Linrad I believe.  THe cost
should be less than the HPSDR or QS-1R.  While the A/D is only 12-bits, a
combination of band-pass filters before the board, and the control of the
VGA, should be able to reduce problems with overload.  More on this as it

Penelope boards are not available right now, however.  There is some
movement to push for either another run from TAPR, or another source.

THe HPSDR group has already released software for the new boards, a variant
of PowerSDR.  It includes source and the FPGA code as well, if you are
interested in playing with it, in an SVN.  Improvements are being made as
moe boards are fired up and USED ON THE AIR FOR REAL QSOs.

There are also other DDC and DUC boards available.  The SDR-14, SDR-IQ from
RF Space, Perseus from Europe, and a recent QEX  article on building a DUC
somewhat similar to Penelope are just a few.

I am slowly building up the SDR and DSP pages on the AMRAD web site
(www.amrad.org).  I plan to add sections on specific hardware and software,
along with some stuff from my experiments.  Please check it out
periodically, and feel free to offer suggestions.  But, I can't let this get
in the way of the true fun, experimenting!!

Join the SDR revolution!!  It's great fun, an interesting challenge, and you
will learn a lot.


> -----Original Message-----
> From: tacos-bounces+wb4jfi=amrad.org at amrad.org
> [mailto:tacos-bounces+wb4jfi=amrad.org at amrad.org]On Behalf Of
> fgentges at mindspring.com
> Sent: Monday, October 20, 2008 1:35 PM
> To: Tacos
> Subject: TAPR High Performance Software Defined Radio
> The TAPR group has been working on a high performance SDR.  Now they
> have asked for a show of hands of people that would be interested in
> buying their Mercury receiver module.  You would need to pay by 15
> November and receive the board around mid-January 2009.
> This receiver shows real promise as the front end has a 16 bit A/D and a
> sample rate to support up to 65 MHz.  That is over twice the high
> frequency limit of the SDR-IQ and two more bits of A/D resolution.  By
> upping the resolution the dynamic range is improved.  This could be an
> important advance.
> To run this board, two other boards are needed, a backplane named ATLAS
> and the USB to computer interface named OZY.
> So, to end up with a working receiver you will need the following:
> TAPR Membership, $20
> ATLAS, Kit, $25
> OZY, $137
> MERCURY, $329
> Total, $511
> The ATLAS board is only available as a kit and is quite reasonable to
> assemble yourself.  It is mainly a backplane board and the individual
> board connectors.
> The OZY and MERCURY boards can be bought bare but for our purposes here
> we are listing the assembled and tested option.
> There are some other boards to fill out the backplane into a full
> transceiver and you can see and learn more at
> www.tapr.org at
> http://hpsdr.org/ and ordering details at
> http://www.tapr.org/kits_merc.html
> And yes, by paying extra, you can buy the boards without joining TAPR.
> Note that this is just the hardware.  Some software is available but a
> lot of work is expected on improved and expanded versions.  I am not
> sure a Mercury package is ready yet.
> Terry, WB4JFI has some of these and can provide some hands-on experiences.
> Frank K0BRA
> _______________________________________________
> Tacos mailing list
> Tacos at amrad.org
> http://www.amrad.org/mailman/listinfo/tacos

More information about the Tacos mailing list