I was surprised this spring when I was reviewing Texas Instruments’ latest version of its TMS320F2803x Piccolo Digital Signal Controller and discovered that the press release was referring to the devices as microcontrollers (MCUs). Huh, did I miss something? A quick inspection of the C2000 processor family’s product home page assured me that I had not somehow mis-categorized the its architecture in my mind, nor had TI snuck in a non-DSP variant of the controller while I wasn’t looking.

So, why then did TI quietly re-brand its exceptionally capable DSCs as "mere" microcontrollers? The marketing people I talked to at TI did not have any answers. In fact, they could not see what all the fuss was about. But it the whole incident still strikes me odd – sort of like going down to the hardware store and finding out that you now had to ask for a “screwdriver” in order to get one of those things with a handle that you drive nails into boards with.  

Certainly the line between DSPs and microcontrollers is blurring, with offerings from companies like Microchip STMicro, and TI serving as both signal and control processors in many of the applications they power. Likewise, many higher-powered DSPs now pack a set of RISC-like instructions that allows them to support protocol processing tasks and serve as their own host controllers. But has it blurred enough to completely ignore the unique value DSPs can bring to many applications?

Perhaps the answer lies in the fact that, until a few years ago, DSPs enjoyed a somewhat well-deserved reputation of being tough to work with. Between their peculiar architectures, odd instruction sets, and the finicky real-time applications they are mostly used for, DSP software jockeys were a breed apart. This has changed a bit in the past few years as programming tools have evolved, allowing most of the development to take place using high-level languages and libraries of tightly-written low-level code to handle the repetitive, compute-intensive or time critical tasks. Hardware development has also gotten an order of magnitude easier, thanks in good part to the many application-specific reference designs that have become so fashionable for most common applications.

So, did TI chose to re-brand its Piccolo/C2000 processors as MCUs because they felt that DSPs still make the average embedded developer nervous, or, because they felt that their rival  MicroChip had a lock on the term “DSC”? Or was TI’s decision based on some other esoteric piece of marketing calculus us mere geeks would not understand?