At 03:11 PM 8/4/2008, you wrote: >You can always have multiple copies of PowerSDR on your computer. Dedicate a >copy per band with the optimal image rejection settings for that band. > >Band changing will not be as easy but at least you'll have the settings you >desire. > >Just a thought until something better comes along. > >Dave >wo2x
Shouldn't need that.. The data structures in PowerSDR are already "per band", and assuming a single calibration pair for each band is sufficient (which it should be), then it's more a matter of loading the structure. But, one might want to take a step back and see why you would want this. I did some analysis about 2 years ago on the calibration performance achievable with the SDR1K, and a single point cal does pretty well (phase within 1 degree for frequencies below 30 MHz) http://home.earthlink.net/~w6rmk/sdr1000/index.htm The actual writeup is at http://home.earthlink.net/~w6rmk/sdr1000/ddsfilter.pdf While I accept that it would be nice to have a better calibration at every possible carrier frequency, one has to ask about what size grit you want to polish that cannon ball with.. Maybe taking Dave's idea and running two copies... one setup for low bands (say, calibrated at 5-6 MHz) and another for the high bands (calibrated at 14-20 MHz) A bigger issue, and one not addressed by the current PowerSDR, is that you might want variable calibrations across the audio passband (especially if you are using a 192 ksps sample rate, and want to move around anywhere in the 150 kHz band with equal performance). I suspect that (particularly for the SDR1K and some random audio card) the variations between left and right (as a function of audio frequency) are bigger than those between I/Q from the DDS and filters as a function of HF center frequency. In the SDR1K, I'd be looking at the matching of C20,C5,C28, and C27 for Tx and C23,C35,C26, and C30 for Rx. The OpAmps and InAmps are probably pretty well matched as are the channels of the FST3253 MUX chips. Since R10/R11 and C23, etc., form 4 little lowpass filters, all of which need to be relatively well matched, that's of some concern. Analysis of the cutoff frequency is a bit tricky, but it's clear that if R changes or C changes, the cutoff will change, and so will the phase shift at other frequencies (particularly those near cutoff) _______________________________________________ FlexRadio Systems Mailing List FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archives: http://www.mail-archive.com/flexradio%40flex-radio.biz/ Knowledge Base: http://kb.flex-radio.com/ Homepage: http://www.flex-radio.com/