SD Card

The SD card was introduced in 1999 by SanDisk, Toshiba, and Panasonic as a successor to the older MultiMediaCard (MMC) format. The “SD” stands for Secure Digital, originally a reference to the digital rights management features built into the spec for the music industry. Those DRM features were rarely used in practice, but the form factor caught on hard for cameras, then phones, then everything portable that needed removable storage. Today the SD Association manages the specification, and SD cards are the most common removable storage medium in consumer electronics.¹

The four core components inside

Every SD card contains the same handful of parts, regardless of capacity or physical size:

• NAND flash memory. Where the data actually lives. Non-volatile (keeps data without power), organized into pages and blocks. Modern cards use TLC (3 bits per cell) or QLC (4 bits per cell) NAND for high capacity at the cost of write endurance.
• Controller chip. The brain. Handles the SD protocol, manages the FTL (flash translation layer) that maps logical addresses to physical NAND cells, performs error correction, and (on better cards) implements wear leveling. When the controller fails, the card disappears or shows wrong capacity.
• External contacts. The metal pads on the back of the card that mate with the host’s card reader. UHS-I cards have one row of 9 contacts. UHS-II and UHS-III have a second row of contacts for the high-speed differential bus.²
• Card body. Plastic housing that protects the silicon and exposes the contacts. A microSD card body is mostly empty space; the actual silicon die is much smaller than the casing.

Monolithic construction is the SD card default

This is the single most important fact about SD cards that consumer-facing pages skip. Almost every modern SD card is built as a monolithic device, meaning the controller and NAND flash share a single piece of silicon, sealed in epoxy with no removable components, no exposed test points, and no published pinout. MicroSD cards are monolithic by physical necessity: the form factor is too small for separate components. Most modern full-size SD cards are also monolithic, with the actual silicon occupying only a fraction of the card body and the rest being empty plastic to fit the standard SD form factor.³

Older full-size SD cards (mostly pre-2015 designs) used PCB-based construction with separate NAND and controller chips. These can still be found in industrial and embedded applications, and they’re significantly easier to recover when they fail because the NAND chip can be desoldered and read directly. New cards in 2026 are almost universally monolithic across all major brands (SanDisk, Samsung, Kingston, Lexar, Sony, ProGrade), regardless of price tier.

The SD protocol and host communication

SD cards communicate with the host device through one of three protocols, depending on the card and host generation:

• SD Mode (SPI / 1-bit / 4-bit). The original SD protocol from 1999. Used by entry-level UHS-I cards and devices like older cameras and microcontrollers. Maximum bus speed 50 MB/s.
• UHS-I Bus. Introduced in 2009. Single row of contacts, half-duplex. Theoretical maximum 104 MB/s. Most consumer SD cards in 2026 still use UHS-I as the dominant interface.
• UHS-II Bus. Introduced in 2011. Adds a second row of contacts for differential signaling. Theoretical maximum 312 MB/s. Used in professional photography and 4K-capable cameras.⁴
• UHS-III Bus. Announced in 2017. Third row of contacts. Theoretical maximum 624 MB/s. Almost no commercial deployment; superseded by SD Express before achieving wide adoption.
• SD Express. Introduced 2018. Uses PCIe and NVMe over the SD card connector instead of the legacy SD bus. Theoretical maximum 4 GB/s on dual-lane PCIe 4.0. Adoption has been slow until Nintendo announced in 2025 that the Switch 2 would require microSD Express cards.⁵

Cards are backward-compatible with older buses but only run at the slowest link’s speed. A UHS-II card in a UHS-I camera will work but at UHS-I speeds. An SD Express card in a non-Express host will run at UHS-I speeds (most current cameras) or fail to mount entirely on very old hosts.

Wear leveling: present but limited on most SD cards

NAND cells wear out after a finite number of program/erase cycles: roughly 1,000 to 3,000 cycles for consumer TLC NAND and as few as 300 cycles for budget QLC. SD card controllers implement wear leveling in firmware, but the implementation is often simpler than what SSDs use. Most SD cards use dynamic wear leveling (only spreading writes across blocks that contain dynamic data) rather than the more thorough static wear leveling that SSDs use. The result: heavily-written areas (FAT tables, file directories, frequently-overwritten log files) can wear out years before the rest of the card, making the whole card fail. This is why a 32 GB camera SD card used as a working disk can fail in a year while one used only for vacation photos lasts a decade.

The SD specification has been extended several times since 1999 to support larger capacities and higher speeds. The card type tells you the maximum capacity and default file system; the speed class tells you the minimum sustained write speed the card guarantees. Both matter when choosing a card and when troubleshooting why an old camera won’t read a new card.

SD capacity types: SDSC, SDHC, SDXC, SDUC

Type	Year	Capacity range	Default file system	Status in 2026
SD (SDSC)	1999	128 MB – 2 GB	FAT12 / FAT16	Effectively obsolete
SDHC	2006	4 GB – 32 GB	FAT32	Common for budget cards
SDXC	2009	64 GB – 2 TB	exFAT	Dominant consumer tier
SDUC	2018	2 TB – 128 TB	exFAT	First 4 TB card released 2025

Backward compatibility runs in one direction only. A device that supports SDXC will read all four types (SD, SDHC, SDXC), but a device that supports only SDHC cannot read SDXC or SDUC cards even though the slot looks identical. This is the source of most “my camera won’t read my new SD card” complaints: the camera is too old for the card’s capacity tier, not the slot’s physical compatibility.⁶

Speed classes: C, U, V, E, and Application Performance Class

The SD Association defined four speed class systems for sequential write speed and one for random access. They’ve coexisted on cards because manufacturers like to print as many symbols as possible:

• Speed Class (C2, C4, C6, C10). The original system. Number = MB/s minimum sustained write. C10 means 10 MB/s. Largely superseded but still printed.
• UHS Speed Class (U1, U3). Introduced 2009 with UHS-I. U1 = 10 MB/s minimum, U3 = 30 MB/s minimum. Same speeds as Speed Class but newer marking.
• Video Speed Class (V6, V10, V30, V60, V90). Introduced 2016. Number = MB/s minimum, like the others. V30 supports 4K, V60 supports 8K, V90 supports raw 8K. The most useful current rating because it reflects real video capture requirements.⁷
• SD Express Speed Class (E150, E300, E450, E600). Introduced for SD Express. Number = MB/s sustained, applicable only to PCIe-NVMe-capable cards.
• Application Performance Class (A1, A2). Different metric. Measures random IOPS for cards used to host applications (Android external storage, dashcam logging). A1 = 1,500 read / 500 write IOPS, A2 = 4,000 read / 2,000 write IOPS. Mostly relevant for cards in Android phones or smart cameras.

SD cards fail more often than any other consumer storage device except USB flash drives, and the failure modes are largely the same as USB drive failures with a few SD-specific additions. The recovery path depends almost entirely on whether the controller is alive and whether the card is monolithic (almost always yes) or discrete-component (rare in 2026).⁸

• Controller failure. The most common SD card failure. ESD damage, accumulated wear, or sudden disconnection during a write corrupts the controller’s firmware or kills the silicon outright. The card disappears from File Explorer / Disk Utility, shows wrong capacity (0 MB, 8 MB, 32 MB), or refuses to mount. On monolithic cards (almost all of them), recovery requires bond-pad probing to bypass the dead controller.
• Damaged or worn external contacts. The metal pads on the back of the card oxidize, scratch, or break from repeated insertion. The card becomes intermittent or undetected. Cleaning with isopropyl alcohol sometimes helps; if the contacts are physically damaged, recovery requires bypassing them entirely.
• Cracked or fractured PCB. The silicon die is small, but the card body is rigid. Bending, dropping, or stepping on a card can fracture the internal traces between the die and the contacts. Card becomes undetected; lab recovery requires opening the casing and bonding directly to the die.
• NAND wear-out. Cells reach their program/erase cycle limit. Symptoms develop gradually: slow writes, unreadable files, then full failure. Modern cards drop into read-only mode as a final safeguard. Copy data off as soon as you notice slow writes.
• Camera ejection during write (RAW partition state). Pulling an SD card while the camera is still writing, or a battery dying mid-write, frequently corrupts the file system tables. The card shows up but appears empty or prompts for formatting. Almost always recoverable with software because the data is still on the NAND chips, just unreachable through the corrupted file system. Critical: do not click “Format” when the OS prompts you.
• Counterfeit / capacity-fake cards. Fraudulent cards programmed to report a higher capacity than they actually have. Writes beyond the real capacity wrap around and overwrite earlier data. The data on the first portion may be recoverable; data written beyond the real capacity is destroyed.⁹
• Firmware corruption (factory mode lock). Controller firmware tables become unreadable, often after sudden power loss during a write. The card enters a factory test mode and shows up with no usable capacity or with a vendor-specific model string. Mass production tools can sometimes reset the card but destroy all data.
• Logical corruption. Healthy hardware, damaged file system or partition table. The card enumerates and shows correct capacity but files are missing or reported as zero bytes. Software like R-Studio, Disk Drill, or EaseUS Data Recovery Wizard can usually recover, especially because most SD cards don’t run TRIM aggressively.

Warning signs your SD card is failing

SD cards rarely give explicit warnings; the first sign of failure is often complete failure. But a few patterns indicate trouble:

• Slow file writes that used to be fast, especially during continuous video recording or burst photography. Cells are wearing or the controller is retrying multiple times to get clean data.
• Files appearing then disappearing between camera and computer. Often FAT/exFAT corruption from ejection during write.
• “Card error” or “memory card needs to be reformatted” messages in cameras. Either logical corruption (recoverable) or controller issues (lab-only).
• Write protection indicating “card locked” when the physical switch is unlocked. Failing controller or PCB issue. Do not try to force writes.
• Wrong capacity reported (0 MB, 8 MB, 32 MB on a card that should be larger). Firmware corruption.
• Card not recognized in some readers but recognized in others. Worn external contacts or marginal controller. Copy data off through whichever reader still works.
• Camera freezes when writing to the card or takes seconds to save each photo. Bad-block accumulation past the controller’s spare-block reserve.

SD cards come in three physical sizes, all carrying the same protocol and similar internals. The physical differences are mostly about fitting into different host devices.

Form Factor	Dimensions	Typical Devices	Common Capacity	Status in 2026
Full-size SD	32 × 24 × 2.1 mm	DSLRs, mirrorless cameras, camcorders, dashcams	32 GB – 1 TB	Dominant in cameras
miniSD	21.5 × 20 × 1.4 mm	Older feature phones (mid-2000s)	1 GB – 16 GB	Effectively obsolete
microSD	15 × 11 × 1 mm	Smartphones, drones, GoPros, Switch, dashcams	32 GB – 2 TB	Dominant overall

The full-size SD card and microSD card use the same electrical specification and the same protocols. A microSD-to-SD adapter is just a passive plastic sleeve with passthrough contacts; the card inside the adapter does all the work. This is why a microSD card in an SD adapter performs identically to that microSD card in a microSD reader. The SD adapter is mechanically a frame and electrically a wire.¹⁰

MiniSD existed briefly in the mid-2000s for mobile phones before microSD won. It’s effectively obsolete; you’ll only see one if you’re looking at a 2007-era flip phone. Both miniSD and microSD use SD-compatible electrical signals, just in smaller physical packaging.

Beyond the three main sizes, SD cards have several specialized variants worth knowing about. Eye-Fi cards (discontinued 2016) added Wi-Fi to a full-size SD form factor for wireless photo transfer; FlashAir cards from Toshiba did the same thing. Tough cards (SanDisk Extreme Pro, ProGrade Cobalt) add waterproof, shockproof, and X-ray-resistant casings for professional photographers in difficult environments. Industrial-grade SD cards use SLC NAND with much higher endurance ratings (50,000+ P/E cycles vs 1,000-3,000 for consumer TLC) and are used in dashcams, surveillance systems, and embedded devices.

SD cards occupy a specific niche: the smallest practical removable storage in consumer electronics, with the worst recovery profile of any common storage medium. Both halves matter.

SD card vs other portable media at a glance

Property	SD Card	USB Flash Drive	External SSD	CompactFlash / CFexpress
Physical size	15 × 11 mm (microSD)	Pen-shaped, varies	Pocket-sized	~36 × 43 mm
Capacity ceiling	2 TB consumer / 128 TB SDUC theoretical	2 TB consumer	16 TB consumer	4 TB CFexpress
Speed (typical)	50–300 MB/s (UHS-I/II)	50–1,000 MB/s	500–2,000 MB/s	800–4,000 MB/s (CFexpress)
Cost per TB (2026)	$80–$200	$50–$200	$80–$150	$200–$500
Reliability	Low	Low	High	Medium-high
Construction	Almost always monolithic	Mixed (monolithic or discrete)	Discrete (PCB-based)	Discrete (PCB-based)
Recovery cost	$400–$1,800 (monolithic)	$300–$1,800	$500–$2,000 (similar to internal SSD)	$500–$2,000
Best for	Cameras, drones, phones	Transit / occasional transfer	Working storage, backups	Professional photo / video

SD card advantages and drawbacks