Originally posted by biz-engineer If the DR increase as the pixel size decreases, then point & shots would have a higher DR then FF.
Phone cameras have a small DR range because the sensors are tiny, and they're collecting much less total light for the same exposure. There is no (or very little) causal link to pixel size there.
DR vs. MP is a false choice, largely, when we're comparing apple-apples and staying in the same sensor tech. There can be some correlation based on product life cycle, but we shouldn't take a manufacturers efforts at any point in time as a definitive statement about physics - Sony wants to sell a lot of lower-MP DR monsters, then provide the higher-MP replacements - with the same amount of DR or better - as 'upgrade' bait to those same folks, or to the folks who care more about resolution and want a reason (higher DR) to upgrade from their D800, etc.
And there have been pages upon pages
and pages (read link for example) written about this, with everyone in the industry who bothers to read and contribute to dpreview chiming in. High image DR is not tied directly to 'larger pixels'.
".
..With respect to why a high MP camera might have better DR than a lower MP one, it has to do with the question of the conversion gain, above. The conversion gain controls the correspondence between apparent photon noise and electronic noise. Imagine ten photoelectron corresponds to 1 millivolt at the pixel output. That means that one millivolt of electronic noise will look like one photon's worth of noise. If we now double the conversion gain so that 5 photoelectrons gives 1 millivolt, then that same millivolt of noise looks like 5 photons worth of noise rather than 10, we have effectively halved the read noise. Now, above I explained that the conversion gain needs to be designed proportional to the pixel capacity. So, imagine that we halve the pixel size, which means it gathers half the number of photoelectrons at the same exposure, then we can double the conversion gain resulting in half the read noise per pixel. What happens when we look at the noise of the two sub-pixels together? Noise adds 'in quadrature' or 'root-mean-square'. So, the combined noise of the two pixels is not double the noise in each, but just 1.4 (square root of 2) times. The signal is the same, so viewed at the same scale as the original sensor, we've increased the DR by 1.4 times." ~ Bob Newman