I have always built my own hardware when working with FPGAs, and for quick test projects, I once made
this board (some leftover space on one customer project panel
) . The board has Altera Cyclone II EP2C8 with configuration memory and 32 IO-pins routed to two pin headers. And most important, every IO-pin has ground pin next to it. With high-edge rate pins, it can really be a PITA if return current path is not near the actual signal.
At my work, there is quite general consensus about that Altera Quartus is somewhat more straightforward to use than Xilinx ISE. That might be one thing to consider.
I think that FPGAs are at end pretty simple compared to MCU's, after all it is just a logic chip. With MCU's, you'll need to figure out what chip designer has thought when (s)he created the chip, but with FPGA, you'll create your own "empire" from scratch, thus you actually know what there is. Hardware (logic) bugfixes are also easy.
I was very surprised how simple it was to make an
experimental stereo three-way active filter (phase linear FIR) using FPGA internal multipliers. That particular filter requires something like 135 million sustained multiply-accumulate (24x32 bit) operations per second. It would have taken much longer if I have used a DSP processor. The purpose of the board was originally to test the Freescale's DSP56371, but I ended up coding the filter to the FPGA
with VHDL.
Regards,
Janne