开云体育

I believe the gunk you see near the center frequency with or w/o hardware attached is an artifact of the signal processing architecture, and noise near DC on the audio input (ie 60 hz hum,. etc)

How the software tuning works is by having a software mixer fed with the local oscillator to dowconvert the signal of interest. In the case of the SoftRock, you have the hardware QSD downconverting and quadrature sampling a 48 khz swath of RF centered on 7.056 Mhz. To listen to a signal at 7.060 MHz, the software mixer is set to run with a local oscillator at 4 khz to mix down the 7.056 Mhz centered signal. As you tune closer to the center frequency, the freq of the software local oscillator gets lower and lower. Very near the center frequency, the local osc will be in the 100's of hz, Mixed with 60 hz (and harmonics) hum this will tend to product responses at 60Hz (and harmonics) +/- SoftwareLO etc -- thereby getting you the gunk around the center frequency.

I will have to admit that my understanding of this phenomenon is not as crisp as I would like it to be. As Tony said in a prior post, there's some hope that some of this can be eliminated with DSP magic. One approach would be to record and characterize the noise characteristics with no input and then try and remove them with DSP magic when fed with a real signal. One could do this in either the time domain or frequency domain, although I'm not sure how one would maintain sync with the noise signal in the time domain. My DSP skills are not strong enough (yet) to really know how feasible such an approach would be.

Another (potentially naive) approach might be to try a different tuning approach. If one is interested in a sig at 7.060, I'd think one could take an FFT of the input signal, zero the FFT bins outside the passband of interest, shift the spectrum over by 4 khz and then IFFT the result to recover the signal of interest. I think this should work, but not having tried it really can't say for sure. Also not too clear in my mind if this would be better than the software mixing approach currently used as you get closer to DC.

One thing to be aware of -- in a more complete SDR with a frequency agile downconverter, you typically don't tune in software down near DC. For example, on the SDR 1000, the tuning is always such that you're approximately 11khz above DC for software tuning. The reason for this being approximate is that the DDS tuning is limited to tuning to freqs where the DDS generates a minimum of spurs.

I will be the first to admit I'm somewhat of a neophyte with DSP. SDR is an area rich with opportunities for learning and experimenting, and we've now got hardware and software accessible to the amateur community to experiment with. Hoping folks get in there and do some experimenting and learning.

Cheers,

Bill (kd5tfd)