I'm not sure why one would want to use a FPGA on a RF chip , given the existence of RF MCU's like the Si1000 series. You can program pretty much every RF function, and the MCU already has capability built into it to do complex packet assembly, and resend on error, and buffer handling.
So $4 buys you a SDR already off the shelf, without having to spent $1000's on getting the FPGA right.
(you can access the important parameters like frequenecy, modulation mode, IF filters etc through the serial port)
@GSMD: There is another problem with wanting to do multi band radios with a single chip, and that relates to the necessity of having inductors and filters to limit harmonics , not a problem if your SDR is receive only , but transmitters and transceivers need fairly complex analogue design to meet various regulatory requirements. If you stick to one band , say 2400MHz, then you can use the LCL filter on the datasheet.
Silabs do make Devkits, and modules for the Si1000 range (and others) . Search your favorite distributor for Si1000DK , pretty good value at < $100. (The Si1000 uses 8051 opcodes so no steep learning curve there) . (The Si1000 does 250-960MHz , 20dBm Rx -121dBm Rx , which is as good as you would get from an analog radio).
Yes, GSMD. As one of the contributors noted, Nuand (http://nuand.com/) amongst others has combined an FPRF and an FPGA to provide a low cost board capable of SDR. If you want just the FPRF on a dev kit, then the company supports an open source initiative (http://myriadrf.org/) where users can order a board and find more information.
Hi Truekop. Spot on. This blog was originally posted last year, before the ADI AD9361 device was released. But hot news this week is that Lime has a new part too called LMS7002M. I am planning a blog to look at both the new devices in the near future.
My meaning was that - viewed from 30,000 feet - the programmability of the part gives a flexiblity in the analog RF domain that is analogous (in my mind) to the flexibity that FPGAs bring to logic design.
I did not mean to imply that it was some sort of new FPGA architecture. In fact the combination of the two product types is very powerful and they often are used together. The FPGA can perform the baseband functions while the FPRF provides programmability of the RF characteristics. The alternative ways to implement RF designs tend to be a bunch of stand-alone functions such as discrete data convertors and filters, which are normally fixed for a particular specification. In comparison, the FPRF is a single chip that is frequency agile, with all the major parameters programmable. And an FPGA can be used to program the RF on-the-fly. So in something like a small cell, the pair can establish which RF channels to use so they avoid interference with the frequencies being used by the macro basestation.
The mechanics of programming are different from FPGAs. There are no look-up tables and programming takes the form of writing a configuration to a memory map via SPI. Only limited routing options are required, as the signal flow might be RF to LNA -> mixer -> filter -> ADC. Users can route analog signals off-chip, for example, if they want to use an ADC with a higher resolution, but they don't need much else.