I believe the A9 & RX10IV use a stacked sensor...quite an improvement over the BSI and front side sensors.
The stacked sensor offers
no IQ advantage. My RX10iii and RX10iv have the same IQ despite the fact that the 10iv has a stacked sensor and the 10iii does not. It's purpose is for faster read out allowing high frame rate video, reducing the "jello effect" for video and faster burst speeds for stills. BSI is an improvement over front side sensors in that a little more light is able to reach the photosites. BSI offers very little advantage for larger sensors.
BSI offers a lot of advantage to larger sensors, too. Compare the designs:
- basic CCD: 100% fill rate, flooding columns with charge if overexposed
- drain protected CCD: 70% fill rate, no flooding, but exposure is constant, requires a mechanical shutter. Would be fine, but once you precharge the sensor for exposure, noise accumulates and until you read it out, it will continue.
- dual area CCD: <35% fill rate, no flooding, and once exposure is made, you can transfer the accumulated electrons to an area where they wait until they're read out. Sensor doesn't require a mechanical shutter
- in comes front-side illuminated CMOS: between 10 and 60% fill rate (depending on electronics integrated into the photosite, but typically around 50% for large sensors). With improving processs lithography, electronics shrink, but at the same time, you're introducing more and more circuits to photosites, so it stays at around 50-70%
- in comes back-side illuminated CMOS: 100% fill rate. No more microlenses, no more very bad vignetting with short pupil distances (in wide angle lenses).
BSI offers incredible advantages for all designs. Stacked sensors don't, really.
It's best seen with the difference between RX100 Mk1 and Mk2 (the Mk2 is a VERY underrated camera). I specifically opted for Mk2 to get BSI and IMHO, the more versatile lens (compared to Mk3) and the hot shoe was a nice bonus.
