A project I just read gave their consumption data in µW. The main component was a microcontroller, which has stated in its datasheet what different current consumptions it has in different modes (nRF51422).
Since P = U*I, should I have any reason to believe that the datasheets current numbers are actually larger when the voltage goes below its nominal voltage? The micro can run from 1.8V-3.3V I think.
consumption data in µW. The main component was a microcontroller [...] nRF51422
This NRF µC has a step-down DCDC included, which draws less current at higher voltages - resulting in almost constant power consumtion for the controller.
Note that this DCDC can be switched off in the user firmware: Then it will work like a LDO, which means the current consumtion will not depend on the voltage (in the working range).
There are a couple of "main" mechanisms that cause power to be comsumed in CMOS circuit: -
The mechanisms are conduction due to both P and N devices being on during transitions of input waveforms AND the charging/discharging of output/input capacitances. Charging the capacitor from Vdd puts energy into the cap that is basically converted to heat when the N channel device switches on.
The higher the supply voltage is, the more energy is wasted in the transitions of the input waveform because the conduction current is proportional to voltage and the charge energy is proportional to voltage squared.
Faster clocks mean more energy per second (power) wasted.
A third mechanism is static power loss and this is usually much smaller than dynamic losses mentioned above. It is due to gate leakage currents: -
Microcontrollers do not work with constant power, so they should take less current with lower voltages. Think about them as a bunch of charging and discharging capacitors (gates). The lower charge voltage is, thevlower energy is stored. This is, by the way, one of the reasons to go to very low voltages in faster circuits: pumping higher voltaged with faster clock would drain much more power.
