Convergence Of Microcontrollers And DSPs

[aruljothi]

The two divergent paths of Microcontroller and DSP may cross occasionally with each performing both tasks. Barriers to entry depend on peripherals, filling the data pipe, and quality of development tools. Designers of today's emerging systems that require both real-time control as well as analog signal processing are looking for ways to reduce cost and speed development time in a market that becoming increasingly competitive.

Both technical and personal considerations rule. For a microcontroller engineer to consider a DSP or for a DSP engineer to use a microcontroller three strict criteria must be met: price/performance, peripheral set, and development tool quality. The absolute convergence of these two architectures are being met by MSPs, which are a joining of the most advanced features of both, as well as the fastest clock rates in the embedded market.

The simplicity of merging the microcontroller and DSP architectures into a single instruction stream seemed the natural step. By implementing parallel execution units, high clock speeds, and glueless interfaces for common memories, deterministic real-time signal processing performance for DSP can exist simultaneously with the multilevel interrupt hierarchy and rapid context switching required for the controller needs of the system.

MSPs also have wide data paths, on-chip cache, and 100MHz+ clock speeds. Development tools for MSPs are more complex than for microcontrollers, but the instruction sets and architectures lend themselves to easier implementations of C compilers compared to a DSP..