I have a question about DDS performance. I look forward to the day when DDS is not a bad word with regard to receiver LO performance. There has been much discussion in all the email lists that involve DDS vfo's in their operation. At present, the DDS spectral issue is limiting our ability to produce an ultra high performance receiver.
I have thought about this 'spur' problem quite a bit, and wonder if it's possible to filter out the spurs in the present generation of DDS chips.
I present a possible method for doing so in the following paragraphs and invite comments and discussion.
------------------------------------------------------
If one uses a DDS in a conventional "typical" QSD with software defined radio type software, the bins in the receiver might be 10 Hz wide. By this, I mean the receivers DSP software looks at 10 Hz wide chunks of spectrum and quantifies the energy in each 10 Hz wide chunk of spectrum.
If the DDS chip has a small output corresponding to a particular 10 Hz bin, the software cannot distinguish between the desired signal having energy in this chunk of spectrum or whether it's the DDS (LO) that has a spur that falls within that 10 Hz chunk of spectrum....so, the software has to treat the DDS spur as though it was a weak, but desired signal. Hence, the DDS with a spur actually generates 'interference'. According to one QSD/SDR icons comments on another mailing list, it is unlikely that any 10 Hz wide bin escapes the spur problem because they are small but prevalent! So the bins are to large to avoid DDS spurs, the spurs are very widespread (well distributed in the output spectrum of the DDS). I know FlexRadio uses math to look for 'sweet spots' and jogs the LO and the sound card IF frequency to keep the DDS tuned to these spur free zones........
BUT.........
What would happen if the bins were (say, for instance) .1 Hz (or even .01 Hz) wide? That's 100 (or 1000) times less spectrum per each bin, and 100 (or 1000) times less chance of having a spur fall within a particular (randomly selected) bin. It's much narrower than we need to receive a cw or ssb signal, except in some specialized weak signal work where we use argo to look for very very weak (but coherent) constant frequency carriers.
With .1 Hz wide FFT bins, many of the bins would be spur free.....
BUT......
How can our FFT software tell whether the bin that has energy above the baseline level is the result of a weak desired signal or is the result of a DDS spur??
How about (temporarily) shorting the antenna terminals once the user has stopped tuning, and scanning for bins with energy in them? Those bins with energy in them are probably due to the DDS LO generated spurs. Once the bins with energy in them are identified, they can be removed electronically by having the FFT software toss out the energy in that particular bin before recombining the energy in the entire 3 KHz spectrum in order to listen to our desired ssb signal.
This can only work if the spurs are coherent or remain on the same frequency (once the user stops tuning). Since the DDS output is digital, presumably these spurs are not random in nature and occur at the same amplitude and frequency all the time.
I am software challenged myself. Maybe some of the more DDS savvy hardware people can comment on the coherence, amplitude stability and how often spurs occur when you zoom in on a 3 Khz slice of spectrum with much finer resolution than the typical 10 Hz wide FFT bin. Obviously this method would involve a heavier DSP processing load, but the spurs could be removed to allow for detection of weaker signals that are presently masked by the DDS spurs.
Would this scheme allow us to use a 'dirty' DDS output for our LO's without concern for spurs at all (trading processor load for improved weak signal receive capability)?
Regards,
Art
--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.344 / Virus Database: 267.11.5/110 - Release Date: 9/22/2005
