Keep in mind SDR is still mostly a military thing -- it has a ways to go before it's cost-effective and low enough power for everyday commercial applications.
A product brief on National's website says 4.1 watts @ 1.9V in a 292 pin thermally enhanced BGA package. You're clearly not going to put that in a cell phone, but that's not the target market.
Scott, is this a continuous time sigma delta architecture? Can you tell us if it's CMOS, SiGE, or what?
What is the power consumption and cost of the 12-bit 3.6 Gsps ADC ? I suspect it might still be too high to prohibit its use for commercial applications.
What do you folks think about the future trends? As far as my understanding goes, it might be a long time till we have low-cost low-power Gsps ADCs.
Good thing nonetheless is that there is growing awareness among academicians as well about the associated signal processing challenges posed by this ADC bottleneck. There have recently been efforts towards exploring design with extremely low-precision ADCs (1-4 buts), and also massively parallel time-interleaved ADCs that can provide higher resolution.
Very interesting to read about this article and we will look forward to read part 2. Speed in which ADC samples is very important and the same define how close we are towards true SDR. Do we have FPGA fabric to handle 3.5 GSPS samples and create I and Q channel and perform other operations in it. It is clear that present generation DSPís may not able to do, but if these ADCís fit in PCI or PXI bus of INTEL processors then we might able crank some amount of computations. Still challenge is there the way we do signal processing in Digital Part. May be we need to look at Neural Network based algorithms and innovation to work with high speed data at one side and very low speed decisions at other side of network. I am keen on looking for part 2 of Scott and hope he does touches the above part. (email@example.com)
Interesting article Scott, and I look forward to Part 2.
It's true that digital radio designers are more interested in specs that are not typically quoted for ADCs. It would eliminate a lot of the guesswork if comms-oriented ADCs quoted perfomance data on parameters like IMD and NPR.