Here you can see a full-size SD card with discrete flash and controllers. Maybe that part number is a start?
Seeing as microSD cards have really taken ground in the past few years, it's probable (to say the least) that the controller that provides glue logic and wear leveling for microSDs is on-die (that is, integrated on silicon) with the FGMOS data area. Vendors are likely to "roll their own" here; while the SD specs are available, the specifics of those implementations are not and probably vary. The controller is designed so that the SD data interface is the only means to interact with it; I've seen potential security vulnerabilities talked about that are created from potentially subverting that controller's firmware. Those sorts of implications make it much more unlikely that a manufacturer will disclose internal behaviors of their controllers.
This bodes well for you, though, because it means you wouldn't need the information on a
specific implementation; if you designed a small microcontroller to handle wear leveling and flash access via either the SD or SPI protocols, you will have effectively designed the proper SD controller. This is a sort of "
duck test"-style argument - if it implements flash, the SD or SD/SPI protocols, and performs the proper data manipulations, then it is an SD controller.
As an expansion on Harald's notes toward CompactFlash above, original Type-I and Type-II CF cards use a large subset of the original (P)ATA/IDE data protocol, which makes them
for the most part identical from an electrical and software interface perspective to non-SATA hard disks (in fact, many adapters exist to allow fitting of a CF card as a small hard disk). Current capacities are up to 512GB for CF cards, and the ATA specs are freely available (
ATA-6 here).