When a USB flash drive is no longer recognised, shows a nonsensical size or asks to be formatted, you must neither format it nor write anything to it: in most cases the files are still present in the memory chip, and the fault lies with the connector, the controller or a corrupted allocation table, not with the data itself. Eject the drive properly, set it aside, and have it analysed before any action that could overwrite what is still there.

What follows explains what is really inside a USB flash drive, what each symptom means, and why recovery, when the controller is dead, is laboratory work in its own right.

What is really inside a USB flash drive

To understand the faults, you need to know what a drive is made of. The principle applies equally, moreover, to a USB flash drive and to a memory card.

On a small circuit board there are four elements that work together. A USB or USB-C connector, soldered onto the board, through which everything comes in and goes out. One or more memory chips, called NAND, where your data is actually stored: there may be a single one, sometimes up to four. A small clock generator, which sets the tempo of all the exchanges. And, at the centre of it all, a controller.

The controller is the conductor: it is what talks to the computer on one side, translating the USB commands, and communicates with the memory chips on the other. A key point: the connector is never linked directly to the memory. Everything passes through the controller. This architecture explains much of the range of faults, since each link can fail independently of the others.

There remains a distinction that changes everything for recovery: the age of the drive. On older-generation drives, all the components are visible and soldered onto the board; you can make out the controller, the memory chip, the capacitors, the resistors. On recent drives, everything is set into a single block called a monolith: the components are laid bare and then embedded in a sealed resin. There is then no longer any visual access to the components, which significantly complicates the intervention.

Your symptoms, and what they mean

A drive that is not well shows it in several ways, and each sign guides the diagnosis.

Symptom	Probable cause
Wobbling connector, drive bent in the port	Torn-off connector or a solder joint that has come loose
Not detected at all	Dead controller, short circuit, connector
Detected but wrong size (0 bytes, a few MB)	The controller can no longer read the information in the NAND
Recognised but files unreadable or corrupted	Logical corruption, unstable memory chip
"Please format the disk"	Corrupted allocation table (see below)

The torn-off connector is one of the most common faults: a drive left plugged in that you knock by accident, and the solder joint gives way. Short circuits can be spotted on certain components of a standard drive. A memory chip problem is recognised by a precise sign: the controller is still talking to the computer, but does not report the correct size, because it cannot read the information in the NAND. Some chips are simply unstable, or carry a temperature sensor which, when faulty, puts the chip out of service. Finally, there are the purely software problems: deleted files, corrupted files, as on any storage medium.

A useful marker: if the drive reports its exact size, that is already a good sign. It means the controller can read the memory and communicate with the computer. The problem is then more often logical than physical.

"Please format the disk": the trap you must never accept

It is the message that causes panic, and this is precisely where you must not give in.

The most frequent scenario: you remove the drive without having ejected it, right in the middle of a write operation. The operation is interrupted along the way and the allocation table, the index that says where the files are, ends up corrupted. Windows no longer understands the structure and then, very helpfully, offers to format the disk to make it usable again.

Above all, do not accept. Your files are still physically present in the memory chip; only the index is damaged. Formatting would mean starting again from a blank structure and seriously complicating recovery. Close the window, remove the drive, and do not use it again until the analysis.

What a laboratory actually does

This is where the recovery of a USB flash drive proves far more complex than one imagines. Several situations, several techniques.

When the chip is unstable or its temperature sensor is faulty, we read the chips separately and play on the temperature: by cooling the chip very strongly, or on the contrary heating it a great deal, one can bias the internal sensor and make it believe it is working at the right temperature. The chip then becomes readable again, just long enough to extract its content.

When the controller is dead, on a standard drive, we unsolder the memory chip to read it in a dedicated reader. This operation is called chip-off. But reading the NAND is not enough, and this is the heart of the difficulty. The controller does not store the raw data: it applies several successive algorithms to it, in particular to spread out wear and reduce the load on the memory during writes. Reading the chip therefore does not give you your files the way reading a hard drive does. The data must then be reconstructed by applying, in the right order, algorithms of inversion, verification and reordering. And since reading is often accompanied by errors, the chip must be re-read several times and progressive error corrections (ECC) applied until a coherent content is obtained.

When it is a monolith, the chip cannot be removed, since everything is embedded in a single block. We then remove protective layers to reach the test access points, those used at the factory during development. By connecting around fifteen of these points on the surface, with specialised equipment, we manage to read the content of the memory. What follows is identical to the standard controller case: the necessary algorithms are applied to reconstruct and restore the data.

In all these cases, reading a memory chip is a long, exacting and uncommon task. This is also why no consumer software can do it for you: it simply has no access to the memory when the controller no longer responds.

The right moves

Faced with a drive that is failing, a few simple reflexes preserve your chances.

✅ To do	❌ To avoid
Eject properly before removing the drive	Format when Windows suggests it
Stop using the drive at the slightest sign	Write new files to it
Set it aside and have it analysed	Force or straighten a bent connector
Keep a copy of the files elsewhere	Run chkdsk or a repair program

The most important move, precisely the one that avoids the "please format" message, is to eject the drive properly instead of pulling it out abruptly. On Windows, use "Safely Remove Hardware" via the taskbar icon. On Mac, click the Eject button next to the volume in the Finder, or drag the volume to the eject icon. You thereby give the system time to finish ongoing writes before cutting the connection.

A USB flash drive is not a backup

It bears repeating, because it is the cause of many disasters: a USB flash drive is not a storage system. Its function is to move content from a point A to a point B when you do not have a convenient connection to hand, or to serve as a temporary stopgap. You prepare a presentation and copy it onto a drive in case your computer breaks down or has no battery left on site: that is the right use. As a primary storage location, however, never. We see this case come back endlessly, particularly with students whose only copy of a piece of work lives on a drive.

Quality also comes into play. Not all drives are equal, and the promotional drives handed out free of charge are often of mediocre quality: handy for a one-off transfer, to be avoided for entrusting anything important to them.

Frequently asked questions

Why is my USB flash drive no longer recognised?

The most frequent causes are a torn-off connector or a solder joint that has come loose, a faulty controller, a short circuit, or a corrupted allocation table. In the majority of cases, the data remains present in the memory chip: it is access to that data that has failed, not the data itself.

Should you format a USB flash drive that asks to be formatted?

No, never, if it contains important data. The message "please format the disk" comes from a corrupted allocation table, often after removal without ejection. The files are still there; formatting would only complicate recovery. Close the window and stop using the drive.

Can data be recovered from a broken or dead USB flash drive?

Often, yes. If the connector is broken, it can be repaired. If the controller is dead, a laboratory unsolders the memory chip (chip-off) or, on a monolith model, works through the factory test points to read the memory directly, then reconstructs the data by algorithms.

What is chip-off?

Chip-off is a technique that consists of unsoldering the NAND memory chip of a USB flash drive or a card to read it in a specialised reader, without going through the faulty controller. The content read must then be reconstructed by several algorithms, because the controller does not store the data in a direct way.

My USB flash drive shows a wrong size (0 bytes or a few MB), why?

Because the controller can no longer correctly read the information in the memory chip. It still communicates with the computer, but cannot report the true capacity or the content. This is a hardware problem that calls for analysis in a laboratory.

Can a USB flash drive be used as a backup?

No. A USB flash drive is for transporting files or providing a temporary stopgap, not for storing them durably. It is a fragile medium that can fail without warning. Always keep your important data on at least one other medium, ideally several.