FPGA's vs. ASIC's

Deciding between ASICs and FPGAs requires designers to answer tough questions concerning costs, tool availability and effectiveness, as well as how best to present the information to management to guarantee support throughout the design process.

The first step is to make a block diagram of what you want to integrate. Sometimes it helps to get some help from an experienced field applications engineer. Remember that time is money. Your next move is to come up with some idea of production volume. Next, make a list of design objectives in order of importance. These could include cost (including nonrecurring engineering charges), die size, time-to-market, tools, performance and intellectual property requirements. You should also take into account your own design skills, what you have time to do and what you should farm out. Remember that it must make sense financially or you are doomed from the start.

Time-to-market is often at the top of the list. Some large ASICs can take a year or more to design. A good way to shorten development time is to make prototypes using FPGAs and then switch to an ASIC. But the most common mistake that designers make when they decide to build an ASIC is that they never formally pitch their idea to management. Then, after working on it for a week, the project is shot down for time-to-market or cost reasons. Designers should never overlook the important step of making their case to their managers.

Before starting on an ASIC, ask yourself or your management team if it is wise to spend $250,000 or more on NRE charges. If the answer is yes and you get the green light, then go. If the answer is no, then you'll need to gather more information before taking the ASIC route. Understand that most bean counters do not see any value in handing someone $250,000 for a one-time charge. They prefer to add cost to the production.

Say your project has a NRE of $300,000, a volume of 5,000, and it replaces circuitry that costs $80. The final ASIC cost is $40. You do some math and determine the break-even point is three years. If you amortize the same design over five years, this could save your company $400,000 even after NRE has been absorbed.

Another option is to do a "rapid ASIC" using preformed ASIC blocks, which saves time and lowers NRE costs. It could also make sense to convert an FPGA to ASIC directly, which lowers NRE a small amount from the rapid type.

Now let's say your company will not fund an ASIC effort. That means it's time to consider FPGAs. First, be aware that while the tools are free on the Web for the smaller FPGAs, you'll have to pay for a license file for the ones with high gate counts. The good news is that there are no NRE charges.

Modern FPGAs are packed with features that were not previously available. Today's FPGAs usually come with phase-locked loops), low-voltage differential signal, clock data recovery, more internal routing, high speed (most tools measure timing in picoseconds), hardware multipliers for DSPs, memory, programmable I/O, IP cores and microprocessor cores. You can integrate all your digital functions into one part and really have a system on a chip. When you look at all these features, it can be tough to argue for an ASIC.

Moreover, FPGA can be reprogrammed in a snap while an ASIC can take $50,000 and six weeks to make the same changes. FPGA costs start from a couple of dollars to several hundred or more depending on the features listed above.

So before you get moving, make sure to enlist some help, get the managers to support you, come up with a meaningful cost estimate, choose the right weapon — be it ASIC or FPGA — and then move into production.

Jeff Kriegbaum (jkriegbaum@arrow.com) is a field applications engineer for Arrow Electronics Components (Seattle, Wash.).