RFID Encoding: Writing, Verifying and Locking the EPC in Production

DI Isabelle Urschitz
25. September 2026

RFID encoding is the step where a unique identity gets written into a tag’s memory, most often the EPC (Electronic Product Code). Done properly it is not a single action but three. First the data is written, then it is read back to confirm the write actually worked, and finally the memory is locked so nobody can overwrite it. Skipping the read back turns encoding into guesswork, because a write can fail silently and still look fine. On a modern production line all three steps happen in one pass at high speed, so every tag leaves the factory carrying the right identity, verified and secured.

A blank RFID tag is not much use. It can talk to a reader, but it has nothing meaningful to say until someone writes an identity into it. That writing step is encoding, and it is where a generic inlay becomes a specific product: this bottle, this garment, this pallet. Get it right and the tag carries a trustworthy identity for the rest of its life. Get it wrong, or fail to check it, and you have shipped a tag that points at nothing, or worse, at the wrong thing.

This article explains what RFID encoding is, what actually gets written, why verification and locking matter as much as the write itself, and how the whole thing runs at production speed.

RFID encoding is the process of writing data into the memory of an RFID tag. In the vast majority of cases that data is the EPC, a unique identifier that tells any reader exactly which item the tag belongs to. For RAIN RFID this follows the EPC Gen2 protocol, standardised as ISO/IEC 18000-63.

The word encoding sometimes gets used loosely to mean the whole personalisation step, but at its core it is simple. A reader sends a write command, the tag stores the value in the right part of its memory, and from then on it reports that value whenever it is read. The complexity is not in the writing itself, it is in making sure the writing was correct and cannot be undone.

A RAIN tag does not have one big block of memory. It has several banks, and understanding them makes encoding much clearer.

The EPC bank is the one encoding usually targets. It holds the identifier that gets serialised per item, often as a GS1 structure such as an SGTIN, so every single tag carries a unique value. Then there is the TID, a serial number programmed into the chip at the factory that is read only and cannot be changed, which is useful for authentication because it is fixed. There is also a user memory area on some chips for extra application data, and a reserved bank that holds the access and kill passwords used to protect and, if ever needed, retire the tag.

So EPC encoding is the everyday work, writing a unique identity into the EPC bank. The TID is already there and untouchable, and the other banks come into play when an application needs extra data or security.

Reliable encoding is really three actions, not one, and treating it as a single write is where a lot of quality problems begin.

The write is the obvious part. The system sends the EPC value to the tag and the tag stores it. For a standard 96 bit EPC this takes on the order of twenty milliseconds, though it varies with the chip.

This is the step people skip and regret. After writing, the system reads the value straight back off the tag and compares it to what it meant to write. A write can fail without any obvious sign, especially at speed or on a marginal tag, and the only way to know it succeeded is to check. A write that is never verified is not encoding, it is hoping. Read back is what turns a hopeful write into a guaranteed one.

Once the right value is confirmed, locking protects it. A lock command secures the memory so it cannot be overwritten, either by accident later in the process or deliberately by someone downstream. Locking can be reversible with a password or made permanent, depending on the application. For retail it mostly guards against accidental corruption. For pharmaceutical goods, luxury items and anything where authenticity matters, it is a genuine security control.

It is worth dwelling on the read back, because it is the difference between an encoding process you can trust and one you cannot.

Imagine a line writing hundreds of thousands of unique EPCs an hour. If even a small fraction of writes fail silently and nobody checks, that is a steady stream of tags going out with missing or wrong identities. In a serialised system, where every EPC is meant to be unique and tied to a specific item, that is not a cosmetic flaw. It breaks the link between the physical product and everything that depends on the tag downstream, from inventory systems to a customs check to a digital product passport.

Verification closes that gap. By reading back every tag and comparing it against the intended value, the system catches a bad write the instant it happens and can mark the tag for rejection before it goes any further. That is why encoding and verification belong together in the same step, not as an afterthought.

Encoding is often assumed to be the slow part of production, and on older setups it could be. On modern equipment it is fast and, more importantly, predictable. The times below are per tag for a single test point, and they scale in a linear, foreseeable way, which lets integrators estimate real throughput in advance.

Times depend on chip type and configuration. The useful property is that they scale predictably, so achievable units per hour can be worked out before committing.

The takeaway is that writing and locking a full EPC lands in the region of twenty five milliseconds, and spread across multiple lanes that adds up to very high throughput without holding the line back.

There are two broad places encoding gets done, and the right choice depends on volume.

For lower volumes, a standalone RFID encoding machine or an RFID label printer with a built in encoder does the job. You feed it tags, it writes and often prints them, and it suits short runs and on demand work.

For a lot of products, encoding is also the moment security gets applied. Locking the EPC stops it being altered. Access and kill passwords in the reserved bank control who can change or retire the tag. And for the most sensitive applications, cryptographic features add authentication so a tag, and therefore a product, can be proven genuine rather than merely read.

Building this in during encoding is far more efficient than bolting it on later. When testing, encoding, locking and security all happen in the same production pass, a secured tag is simply the normal output of the line, not an extra downstream operation with its own handling and cost.

  • RFID encoding writes a unique identity, usually the EPC, into a tag’s memory so it points to a specific product.
  • Done properly it is three steps: write the data, read it back to verify, then lock it so it cannot be changed.
  • The read back is the part that makes encoding trustworthy, because writes can fail silently and only verification catches them.
  • Writing a 96 bit EPC takes about twenty milliseconds and writing plus locking about twenty five, and these scale predictably across lanes.
  • On modern lines, testing, encoding, locking and security all run in a single pass, so every tag leaves the factory verified and secured without slowing production.

Related Articles

Contact

Any questions about this topic?

Content