Tech Behind ThingsHow the ordinary machinery actually works

Devices

Why A Memory Card Wears Out From Being Written To

Flash storage degrades a little with every write because storing a bit means forcing charge through an insulating layer that gradually stops insulating.

A classroom setting featuring laptops and desks, capturing a modern educational environment.
Photograph by Adam Sondel via Pexels
Editorial note. Independent reporting and analysis. Nothing here is sponsored or paid for. How we work.

A memory card has no moving parts, yet it has a limited life measured in writes rather than years. The wear happens at the level of individual cells.

A bit is a trapped charge

Each flash cell holds electrons in an isolated region surrounded by insulation. The presence or absence of that charge is what the card reads back as a one or a zero.

Because the region is insulated, the charge stays put with no power applied. That is why the card remembers its contents when disconnected.

Getting charge in and out is the difficult part. Electrons must be forced across the insulating barrier using a relatively high voltage, since there is no ordinary path.

Every crossing damages the barrier

Each pass leaves the insulating layer marginally less perfect. Defects accumulate, and eventually charge begins leaking through of its own accord.

A worn cell loses its contents over time rather than failing outright. Data written months ago becomes unreadable while newly written data is fine.

Cells also become slower and harder to program as they age, because the controller must apply more effort to achieve the same stored charge.

Erasing works in large blocks

Flash can be written in small units but only erased in much larger ones. Changing a few bytes therefore means rewriting an entire block elsewhere.

The result is that a small logical change can cause a much larger physical write. This amplification is the main reason wear accumulates faster than the file sizes suggest.

Controllers reduce it by collecting changes and writing them together, which is why a card can appear busy for a while after copying appears to have finished.

Wear levelling spreads the damage

If the same block held the same file forever, a handful of cells would wear out while most stayed pristine. The controller therefore moves data around continuously.

Addresses the device sees are mapped to physical locations that change without notice. A file that never appears to move may have been relocated many times.

Cards also hold spare blocks in reserve. As blocks fail, they are retired and replaced from the pool, so capacity stays constant while the margin quietly shrinks.

Density traded endurance for capacity

Storing more bits per cell means distinguishing several charge levels instead of two. The gaps between levels are narrower, so smaller leaks cause errors sooner.

This is the main reason a modern high-capacity card can tolerate fewer writes per cell than an older, smaller one. Capacity per chip rose and endurance fell.

For a card holding photographs, that trade is invisible. For one recording continuously in a dashboard camera, it is the specification that matters most.

Questions readers ask

Why does brightness jump when I unlock the phone?

The sensor is often only sampled while the screen is on, so the first reading after unlocking replaces a stale value from earlier.

Does automatic brightness save battery?

Usually yes, because most people set a fixed level high enough for the worst case and then leave it there in dim rooms.

Devicesdisplayssensorsperceptionpower
Grigor Petrov
Hardware writer, Tech Behind Things

Grigor writes about silicon, thermals and the physical limits designers keep bumping into.

Also by Grigor Petrov