QUESTION: Spectrally pure DDS Output?
 

I have the following question regarding DDS chips:

The amqrp website and others detail techniques for using Analog
Devices', DDS chip as an accurate digitally controller oscillator. One
of the issues that needs to be bourne in mind according to the articles
is the spectral purity of this chip's output. There seem to be various
schemes using PLL and filtering to kill the harmonics and spectral
"grass" that these things seem to generate. Given the onward march of
technology, is there any chip similar to the Analog Devices DDS, that
would give be a spectrally pure, low noise sinusoid with any further
filtering or PLL? I'd like something good enough to be a high quality
VFO for an HF receiver (I and Q available)?? Is such a thing
available, preferably offering a presoldered evaluation board (I don't
want to go blind soldering it!!) for 30 or 40 bucks kind of price
range??

Thanks,

Tim

Wellll... I know a company that sell a DDS-based LO VME module with
very low spurious, but the price isn't "30 or 40 bucks." More like 500
times that. They are pretty tight-lipped about just how they do it.

Cheers,
Tom

A DDS chip on its own may cost more than 30...40 dollars, unless the
manufacturer is willing to send you a sample. And what is wrong with AD 995X
series DDS? They are of a very low "spectral grass". You can also get an
idea for how much AD sells their DDS evaluation boards by browsing their
product catalog.

wrote in message
oups.com...
I have the following question regarding DDS chips:

The amqrp website and others detail techniques for using Analog
Devices', DDS chip as an accurate digitally controller oscillator. One
of the issues that needs to be bourne in mind according to the articles
is the spectral purity of this chip's output. There seem to be various
schemes using PLL and filtering to kill the harmonics and spectral
"grass" that these things seem to generate. Given the onward march of
technology, is there any chip similar to the Analog Devices DDS, that
would give be a spectrally pure, low noise sinusoid with any further
filtering or PLL? I'd like something good enough to be a high quality
VFO for an HF receiver (I and Q available)?? Is such a thing
available, preferably offering a presoldered evaluation board (I don't
want to go blind soldering it!!) for 30 or 40 bucks kind of price
range??

Thanks,

Tim

I think you'll have to say what is acceptable to you and define what you
mean by "spectral grass". In terms of phase noise, you'll do better with a
normally designed DDS than you will with a normally defined PLL. For data on
what a DDS can do rephase noise, look at various models of synthesizers on
the PTS web site.

Now of course there are spurs to contend with in any DDS design, but
discrete spurs are not what I would call "spectral grass". If spurs are
indeed your issue, you can go to a 14 bit DDS (AD9951 or others) and get at
least spurs at least 80 db down, or add filtering to the DDS output.


Chuck

On Fri, 15 Jul 2005 04:52:57 GMT, wrote:

I think you'll have to say what is acceptable to you and define what you
mean by "spectral grass". In terms of phase noise, you'll do better with a
normally designed DDS than you will with a normally defined PLL. For data on
what a DDS can do rephase noise, look at various models of synthesizers on
the PTS web site.

Now of course there are spurs to contend with in any DDS design, but
discrete spurs are not what I would call "spectral grass". If spurs are
indeed your issue, you can go to a 14 bit DDS (AD9951 or others) and get at
least spurs at least 80 db down, or add filtering to the DDS output.

The total spur power seems to be more or less constant in the
traditional DDS. However the distribution of the spur power varies
constantly with the frequency (i.e. phase accumulator increment).

At some settings there will be a huge number of low level spurs more
or less evenly distributed along the frequency band. I assume that the
OP is calling this spectral grass.

However, at some specific phase accumulator increments, the spur power
is concentrated on just a few discrete frequencies, adding up to
easily visible spur peaks, but between these peaks the spectrum is
clean (only thermal noise).

If the discrete spurs cause problems, it would be a good idea to check
if the frequency management could be modified in such a way that the
spot frequencies that cause large discrete spurs are skipped.

OTOH, if a very clean signal is required in a narrow band, it might be
a good idea to concentrate the total spur power into a few discrete
spurs outside the band and thus have a clean band of interest.

Paul OH3LWR