Hard Disk Drive (HDD)

Every traditional hard drive is built around a few core components, sealed inside a precision-engineered housing because a single particle of dust can scratch the disk surface and destroy data. Understanding what each part does is the foundation for understanding how HDDs fail.¹

Platters and spindle

The platters are the actual disks data is stored on. They are rigid circular discs made from aluminum, ceramic, or glass and coated with a thin layer of magnetic material. Most desktop drives use multiple platters stacked on a single shaft. Both sides of each platter store data, so a drive with three platters has six recording surfaces.²

The spindle motor spins the platters at a fixed speed: 5,400 RPM for laptop and energy-saving drives, 7,200 RPM for typical desktop drives, and up to 15,000 RPM for enterprise drives. The platters never stop spinning while the drive is powered on, which is what makes any physical shock to a running HDD potentially catastrophic.

Read/write heads and the actuator

Each platter surface gets its own dedicated read/write head, mounted at the end of an actuator arm. The heads are tiny electromagnetic devices that float on a cushion of air created by the spinning platters. In a modern drive, the gap between head and platter is roughly 5 nanometers, smaller than a strand of DNA.³

The actuator moves all arms together as a single unit, swept by a voice coil motor (VCM): a copper coil between two strong neodymium magnets. By varying the current through the coil, the drive’s controller can position the heads with nanometer precision over any track on the platters. Older drives used stepper motors; modern drives have used voice coils for decades because stepper motors couldn’t keep up with shrinking track widths.

Tracks, sectors, and the service area

Data is organized in concentric circles called tracks, each subdivided into sectors. Traditionally each sector held 512 bytes; modern Advanced Format drives use 4K (4,096-byte) physical sectors to improve density and error correction. Multiple sectors are grouped into clusters by the file system above the hardware layer.

Every HDD also has a hidden service area: special tracks reserved for the drive’s own firmware, defect lists, and S.M.A.R.T. logs. The service area is invisible to the operating system but critical for recovery: if it’s corrupted, the drive may refuse to spin up at all even if the platters and heads are perfect.

PCB and firmware

The printed circuit board (PCB) on the bottom of the drive holds the controller chip, RAM cache, motor controller, and SATA/SAS interface. The PCB runs firmware stored partly in a flash chip on the board and partly in the platter service area. PCB failures from power surges are common, and not always recoverable by simply swapping in another PCB, since modern drives tie the firmware to a unique ROM chip on each board.

HDDs fail in predictable ways, and the failure mode determines whether recovery is a software job, a lab job, or impossible. Backblaze’s 2024 fleet of more than 300,000 drives reported an annualized failure rate of 1.36%, about 1 in 73 drives per year on average.⁴ Here are the failures that show up most often in actual recovery cases.

• Bad sectors and gradual surface degradation. Individual sectors fail and the drive remaps them to spare sectors in the service area. This is normal and expected, until the drive runs out of spares and uncorrectable read errors start appearing. Recoverable with imaging software like ddrescue or HDD Raw Copy Tool. See bad sectors.
• Head crash. The read/write head touches the platter, usually from physical shock, sudden power loss while parking, or wear on the slider. Causes the classic “click of death” sound as the actuator repeatedly slams the head against an end-stop. Lab recovery only.
• Spindle motor seizure. The bearings supporting the spindle wear out or the motor coil shorts. The drive won’t spin up, often making a faint hum or no sound at all. Lab recovery via platter swap to a donor drive: exceptionally difficult work.
• PCB failure. A power surge or capacitor failure kills the controller board. The drive shows no signs of life: no spin-up, no detection in BIOS. Recoverable by transferring the unique ROM chip from the original PCB to a matching donor PCB.
• Service-area / firmware corruption. The drive’s internal firmware tracks become unreadable. The drive may detect, sometimes spin up, but reports zero capacity or wrong model information. Requires specialized hardware tools (PC-3000) to repair the service area.
• Stuck heads. After a hard shutdown, the heads sometimes fail to park properly and stick to the platter surface. Dangerous to recover: the heads must be unstuck without scratching the platters, which is cleanroom work.
• Logical corruption. The drive is mechanically fine but the partition table, file system, or boot sector is damaged. Recoverable by software like R-Studio, Disk Drill, or DMDE.

Three physical form factors dominate the modern HDD market, plus a steadily climbing capacity ceiling driven by enterprise demand.

Form Factor	Typical Capacity	Common Uses	Spin Speed
3.5-inch desktop	1 TB – 24 TB	Desktop PCs, NAS, data center	5,400 / 7,200 RPM
2.5-inch laptop	500 GB – 5 TB	Laptops, external portable drives	5,400 RPM (mostly)
2.5-inch enterprise	900 GB – 2.4 TB	Servers, SANs	10,000 / 15,000 RPM
External / USB	Mostly 3.5″ or 2.5″ inside an enclosure	Backup, portable storage	Varies by drive inside

As of 2026, the largest shipping HDDs are 24 TB drives from Seagate (model ST24000NM002H) and 22 TB drives from Western Digital, both targeting data-center customers.⁵ Manufacturers continue to push capacity higher each year using HAMR (heat-assisted magnetic recording) and SMR (shingled magnetic recording), though SMR drives are notoriously slower to recover from because of how they overlap data tracks.