[Elecraft] Software Defined Ham Radio

[Lyle Johnson]

> >I've been running one of these for nearly a year now...
> [snip]
> >It works very nicely indeed -- but remember it is a DDS
> >system and, as such, strong stations can give spurious
> >readings...
>
> Can you clarify what you mean by the above statement?
> What sort of spurious readings are you talking about?
> What does the unit being DDS based have to do with it?
> Your statement implies that Elecraft's new KX1, which
> is also DDS based, would suffer the same problem.

These questions seem to have been unanswered, so...

DDS spurs are a well-known problem.

The SDR-1000 is a very basic, image rejection mixer, direct conversion
radio.  It gives a good account of itself considering its straightforward
design, and has attracted a following of early adopters, many of whom are
contributing to the software which defines it.

I confess to being an early adopter, as is Dave, who posted the comment
about DDS spurs.

A series of articles appeared in QEX over the last year and a half which
pretty well describes the design. The articles can be downloaded from Flex
Radio's website.

In order to get a direct conversion receiver to provide image rejection, you
need to provide a quadrature local oscillator.  This is like the old phasing
SSB rigs - OK, antique SSB phasing rigs - where the undesired sideband could
be rejected with careful adjustment of the amplitude and phase of the
oscillator and the signal being mixed with the oscillator.

In the case of the SDR-1000, an Analog Devices AD9854 DDS chip is used for
the local oscillator.  This chip has the advantage of providing a quadrature
signal (two signals whose relative phase is 90 degrees).

But, the design has a couple of issues.

In order to cover the range from DC to 65 MHz, the reconstruction filter
following the DDS output necessarily cuts off above 65 MHz.  This means that
all sorts of DDS artifacts can appear in the DDS output when you are tuning
much lower in frequency, like on 40 or 20 meters.

One way to minimize the large spurious outputs that result from this
approach is to carefully pick the DDS output frequencies.  Some are worse
than others, and some are very good.  In the case of the SDR-1000, as long
as the DDS output frequency is such that the output digital-to-analog
converters have something to do at every sample time, the spurs will be
minimized.  With a 200 MHz clock and a 14-bit DAC, this amounts to a tuning
step size of about 3 kHz.  The fine tuning between these steps is handled by
the DSP code running in the PC attached to the radio.

There are still some spurs, but they are fewer and mostly of little
consequence when an antenna is attached and band noise and real signals of
low to moderate strength are present.  There are various steps that can be
taken to minimize this problem, in addition to the one mentioned above.

Many modern radios use a DDS as part of a multi-loop phase locked loop
design for the receiver oscillator(s).  Look in any recent copy of the ARRL
Handbook for an example of a DDS/PLL oscillator system from an Icom
transceiver.

These designs can be very complex.  They offer the advantage of precision
frequency generation and rapid tuning if implemented carefully.

The SDR-100 also has very wide input filters, so DDS spurs have plenty of
opportunity to mix with strong, out-of-band (or out of IF passband) signals.

And, because it attaches to a required PC, there is plenty of opportunity
for all sorts of interaction. Add to this the fact that the radio is
unshielded -- it is just a stack of 3 PC boards with no case -- and you can
well imagine that there will be additional opportunities to create phantom
signals.

The K2 avoids the whole issue of DDS spurs by using a phase locked loop
whose reference oscillator is stepped by a tuning voltage from a DAC, rather
than using a DDS.

In the case of the KX1, where the DDS is directly sued as the receiver local
oscillator and the transmit carrier oscillator, the DDS spurs are tamed by a
couple of design tricks.  The tuning range is limited, which makes the DDS
output filter capable of better "smoothing" of the DDS output.  The
microcontroller is programmed to avoid certain frequencies.  The KX1 uses
narrow, double-tuned input filters to attenuate out-of-band signals.

Finally, it is designed for low power, low current operation, so it is not
attached to a computer.  It generally is in a fairly low overall noise
environment.

For further information about DDS design considerations, you might want to
go to the Analog Devices website and download a tutorial file in PDF format:

http://www.analog.com/UploadedFiles/Tutorials/104002517DDSTutor.pdf

The file is about one megabyte in size.

73,

Lyle KK7P