EDN (March 06, 2014)
If your board design experience with FPGAs is limited or nil, the prospect of putting one in your next project can be daunting – especially if it's a 1,000-pin monster. Continue reading to get a feel for the selection and design process, and for lots of gotchas you'll need to avoid.
Pick a vendor, any vendor
Your first issue is of course vendor and device selection. Often, the vendor decision comes down to whichever you've had the best experience with previously – not applicable if you're an FPGA virgin. Or perhaps the decision has already been made by the person who will be designing the internal logic (which may be you), based perhaps on available vendor or 3rd-party IP and its cost. Local FAE support should be considered too.
The vendor's software tools can also factor into your decision. Download and play with them – no hardware required. You can still take a design all the way to the simulation stage. This is also a way to determine how big an FPGA you'll need, assuming your internal logic design is mostly done.
For a deep dive into FPGA waters, head to the various vendors' sites. Just make sure you have a day free if you hope to grok the vast amount of information presented (and not always as clearly as might be wished for). Altera and Xilinx are frontrunners, both in market share and leading-edge technology. Their parts use internal configuration RAM, so an external ROM with the configuration data is required to "boot" the parts (though both companies also have small, non-volatile, CPLD-like parts). Other vendors to consider are Microsemi/Actel, Lattice, and Cypress. Some features you'll find in their parts include very low quiescent power, ROM-based configuration for "instant-on" startup, and analog peripherals.
Click here to read more ...