I'm currently doing some power analysis for an electronics project and am attempting to make my design more efficient. My design uses a 3V coin cell as it's primary power source. Out of good practice (and because my design operates at 3.3V nominal), I've decided to add a voltage regulator. I've selected this Microchip MCP1252 (http://ww1.microchip.com/downloads/en/DeviceDoc/20001752C.pdf).
This chip is able to be configured as either a 5V regulator or 3.3V regulator. Looking at Figure 2-5, it shows the power efficiency when the regulator is configured to output 3.3V. Looking at my application using a 3V coin cell supply, the regulator is only 55% efficient. This is not that good... However, looking at Figure 2-4, the regulator appears to be much more efficient when configured to be a 5V regulator.
So my question - why is this regulator much less efficient when configured for 3.3V output? Does it have something to do with I^2 R losses? Furthermore, why does the efficiency trend from Figure 2-5 look the way it does? It seems to be much more efficient as soon as you change your input supply to 4V. Why is this?