In many smaller, low-power applications, “they” provide the microcontroller unit (MCU) with a power rail, and the MCU sips that power. (Of course, the “they” may be “you” when wearing a different hat in the project cycle.} And while an efficient supply matched to the application is key, it normally provides a fixed voltage. Power saving is done by the MCU shutting down its internal functional blocks when not needed, and going into sleep mode, usually under the direction of the application software.
Larger processors in complex systems do communicate with their supplies, throttling the supply back to save power while the processor goes into a slower clock mode or even to sleep. But MCU's don't have that talk-back, closed-loop option, since their supplies are usually a basic switcher or even an LDO.
That's why I was surprised when I saw a new LDO from Texas Instruments, designed to work with (but not limited to) their MPS430 series of ultralow-power MCUs. This LDO can switch between two output voltages, set via a single control line. The MCU, in turn, uses a simple digital I/O line to tell the LDO which of the two voltage values it wants, depending on its operating mode. The result is that the MCU can direct the supply, and itself, into a lower-power more.
Sounds like a win/win arrangement, and it many ways, it is. But it does mean that the system developer of the MCU-based project must be aware, early in the software design, of the use of this type of dual-mode LDO, and program that control line as part of the algorithms and code. It also means that the MCU is tied, for better and worse, to this particular LDO (clever, those TI folks!) in order to gain the power savings. The days of “just give me a clean, stiff supply rail so I can run my MCU” may fade away, in order to achieve additional system-level power savings.
-x x x-