At 06:30 AM 2/5/2007, Bill Tippett wrote:
>Mark Amos wrote: > >The Flex-Radio, on the other hand, was pretty much unuseable when the >transmitter & amp were running. It saw every dit from that transmitter out to >dozens of KHz's away - spurs popping up and drowning out even S9+ stations. > > BDR seems to be the weak point for SDR rigs. For *some* SDR's... It all hangs on gain control in the RF chain and the spurious free dynamic range of the analog to digital converters. The SDR1000 has very little selectivity ahead of the A/D, other than the bandwidth of the A/D circuits, so it's digitizing a huge swath (tens of kHz), in which there may be fairly strong signals. You typically want the gain of the system set so that the A/D never clips: Although, depending on the kinds of interfering signals you're seeing, and the type of modulation you're using, some clipping might be ok. A typical approach might be to set the gain so that the biggest single signal is 6dB below full scale (that lets you have a couple strong signals that just happen to align and still not clip). Assuming you can get 20 bits of effective performance out of the converter, you'll have 120 dB dynamic range in the converter. You lose 6 off the top for the strong signal headroom, and you probably want your desired signal at 10 dB over the quantization noise.. so the blocking dynamic range might be about 100 dB. The only way to get more is to have more bits. Either you run higher resolution converters (or gang up multiple converters interleaving either in time or voltage) or do something like digitizing multiple overlapping subbands, each with narrower bandwidth, and then combining them in post processing, or, use multiple antennas to form multiple beams, each which gets its own converter, and hoping that your interfering signal is coming from a different direction. There's also a variety of fancy adaptive cancellation strategies (not all that much different, conceptually, from the sigma delta architecture used in high performance A/Ds like the venerable AD9042) where you measure the strong signal and subtract a modeled version from the analog signal before digitizing. (this was used decades ago to remove the 60 Hz artifact from EEG recordings, for instance). For cosite interference (the field day type problem) people have been using analog adaptive interference cancellers for years, since you have access to a high quality replica of the interfering signal. Your receiving system (receiver and antenna) essentially has a extremely sharp pattern null right on the interference source. >Mark's results are consistent with ARRL's measurements. >Is there any hope of the Janus project improving BDR >performance or will it take 32 bits to do it? janus is a 24 bit converter with (I think) 110dB dynamic range. Sort of like drag racing.. how much instantaneous dynamic range can you afford to buy? Pretty quickly, you get out of the regime where you can leverage off cheap consumer data converters with convenient hardware and software interfaces. _______________________________________________ FlexRadio mailing list FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archive Link: http://www.mail-archive.com/flexradio%40flex-radio.biz/ FlexRadio Homepage: http://www.flex-radio.com/ FlexRadio Knowledge Base: http://kb.flex-radio.com/