FPGAs have progressed well beyond the original core fabric composed of logic and flops, surrounded by general-purposed I/O. The fabric has been upgraded with embedded memory, DSP blocks, AI processors, all connected by a network on chip. In addition to an upgraded fabric, FPGAs now feature multi-core processors.
The I/O ring has also been upgraded with a range of hard IP blocks and high-speed SerDes to support several interfaces such as Gigabit Ethernet, PCIe, DRR memory, etc. Today’s modern FPGA has become a programmable SoC with complex clocking requirements.
FPGAs Represent a Complex Clocking Environment
As the functionality of FPGAs has increased, the complexity of clocking needs has grown. As a result, FPGA vendors have added several built-in PLLs and clock management functions. This increase in functionality is mirrored by an increase in needed clock sources:
- Multiple reference clocks for the embedded PLLs
- Per I/O bank reference clocks
- Clocks for user logic
- Various supporting clock sources for features such as real-time clocks, configuration controllers, etc.
For SRAM-based FPGAs, additional external logic may be required to control the configuration. Often a small CPU plus flash memory is used, requiring its own clock source.
These FPGAs and the supporting configuration logic do not exist on a board in isolation. For example, higher-end CPUs can be found in conjunction to the FPGA, with the FPGA acting as a hardware accelerator to the CPU. Often other devices such as transceivers, DRAM, and other ASSPs can be found on the board. All these devices have their own clocking requirements, collectively creating a complex clocking environment. Designers need a supplier who can deliver a range of clocking solutions from oscillators to clock management devices.
FPGAs are Ubiquitous
FPGAs are a niche product used everywhere, from set top boxes to GPS-guided munitions, from the ocean floor to space — anywhere a custom solution is needed. But a custom ASIC would be cost prohibitive (from an NRE perspective), or not be able to meet time-to-market needs. Consequently, FPGA designers need a supplier whose product line supports a range of environments from the benign conditions of an office, to harsh environments with wide temperature ranges, high vibration and pressure.