Traceable Quality®
Talk to us
Menu

Insight

Pick the antenna, then the reader

Gain points the same watt in fewer directions, and past about 6 dBi the law turns the reader down. Polarisation decides whether a turned tag reads at all. What that means for the box you buy.

A tag reads on the bench and not on the floor.

We get that call more than any other. Somebody tested a reader at their desk, held a label up to it, watched it read from across the room, and bought six. Then the labels went onto cartons, the cartons went onto a pallet, and a third of them stopped answering.

Nearly every time, the reader is fine. It’s the antenna, or the way the tag is turned against it.

A reader is a radio with no speaker. Gain points the same watt in fewer directions. A linear antenna reads a lined-up tag and misses a crossed one; a circular antenna reads at any angle and pays 3 dB for it.

A radio with no speaker

A UHF reader is a radio. It sends power out, listens for the faint reply a tag makes by reflecting some of that power back, and decodes it. The radio inside a handheld and the one inside a dock-door portal speak the same Gen2 protocol, and most of them are built on one of a handful of reader chips. What the reader can hear is decided mostly by the antenna on the end of the cable.

A dipole hears all round, like a bare bulb. It has 2.15 dBi of gain. A patch is the same bulb with a reflector behind it. Nothing goes out the back, so more goes out the front: 6 dBi of gain, and a beam about 85 degrees wide. Patches run from about 70 to 100 degrees, wider at low gain and narrower at high.

Gain doesn’t make more power. It points the same watt in fewer directions.

So a bigger number isn’t a better antenna. It’s a narrower one. That’s what you want on a dock door, where there’s one thing to light up and a forklift lane beside it you’d rather leave dark. It’s the wrong thing on a shelf, where the tags at the edge of a narrow beam never wake.

Then the law turns it down

There’s a ceiling on what a reader may radiate. In India it’s 2 W ERP for RFID readers between 865 and 868 MHz, the same as Europe. That’s about 3.3 W measured against an isotropic antenna, or 35 dBm.

Our usual setup puts 1 W into 1 dB of cable and a 6 dBi patch. That comes out at 35 dBm. It’s already at the ceiling.

Swap in a 9 dBi panel and you’d be radiating nearly twice what’s allowed, so the reader has to drop to about half a watt. The beam does get narrower. On the sheets I checked, Laird’s 8.5 dBic panel and FEIG’s 9 dBic one, it’s 65 to 70 degrees, against 100 for Laird’s 5.5 dBic patch. But the tag wakes at about the same distance it did with the patch, 22 or 23 m in free space with our parts.

What did the extra 3 dB buy? The reply. On the way back the panel’s gain counts in full, because nothing caps how well a reader may listen. That matters when the reader is the weak side of the link, which with a cheap reader chip it often is. With a good one it rarely is. Either way it’s a narrower beam and a better ear. It isn’t more range on the way out.

Polarisation

A tag’s antenna is a line. Look at a wet inlay and you’ll see it: two meandering arms either side of the chip.

A linear reader antenna sends a line too. Line them up and nothing is lost: a 6 dBi linear patch on our usual parts wakes a UCODE 9 tag 31 m away in free space. Turn the tag 45 degrees and you lose 3 dB each way. At 60 degrees it’s 6 dB each way, and half the range. At 90 degrees the two lines are crossed, and what gets through is whatever the antennas leak across. Our sums assume they keep the two apart by 25 dB, which leaves a couple of metres.

On a bench you hold the label straight without thinking. On a pallet the cartons are turned every which way.

A circular antenna sends a corkscrew instead. Whatever angle the tag is at, it picks up half the power. That’s 3 dB lost against every tag, every time, and 22 m instead of 31. Nearly every fixed reader we install has circular antennas on it. Nine metres is cheap. A pallet that goes out with four cartons uncounted isn’t.

Linear still has its place. On a conveyor where every carton arrives the same way up, with the label in the same spot, a linear antenna lined up with the tags gets those 3 dB back.

Which box

Once the antenna is settled, the reader is mostly a question of how many antennas and what they’re bolted to.

A handheld has one antenna built in. You point it. It’s the cheapest way to start and the right tool for cycle counts and finding one item.

A fixed reader has two, four or eight ports, and you choose the antennas. One reader can cover a dock door with two panels, or a long run of shelving with eight. A multiplexer gets you more ports.

Portals and tunnels are fixed readers with the antennas already mounted: a frame for pallets, a shielded box for cartons on a conveyor. Overhead arrays steer a beam across a whole floor from the ceiling and tell you roughly where a tag is, not just that it’s there. A printer-encoder has a small antenna inside, aimed at one label, so it can write the EPC as it prints and not the label next to it.

Seven boxes. Much the same radio inside each one. What changes is the antenna and where it’s pointed.

The numbers, and where they come from

The ranges are free-space sums from the same link budget as the range calculator and Wake the tag: 866 MHz, 1 W into 1 dB of cable, an NXP UCODE 9 tag (−24 dBm) on a half-wave dipole, and an Impinj E710 reader in dense-reader mode. A real room is usually 30-40% shorter.

The 2 W ERP limit is Table IV of G.S.R. 853(E), the rules India published in December 2021 for 865-868 MHz. They replaced the 2005 RFID rules, which allowed 1 W into the antenna and 4 W ERP in 865-867 MHz. Readers type-approved under the old rules keep them for the rest of their life. ETSI EN 302 208 gives Europe the same 2 W. In the US, FCC Part 15.247 allows 1 W into an antenna of up to 6 dBi and asks you to turn down by every decibel above that, which is 4 W EIRP.

The beamwidths are from makers’ sheets: Laird’s RFID antenna catalogue (5.5 dBic, 100 degrees; 8.5 dBic, 70 degrees), FEIG’s ID ANT.U290/290 (9 dBic, 65 by 65 degrees) and Chainway’s Ant-RC12 (12 dBi, 40 by 38 degrees). Sheets for the same gain differ by 10 to 15 degrees. A 6 dBi patch falls between Laird’s two, so the film and the explorer draw it at 85.

The polarisation loss is the textbook one: the cosine squared of the angle between two linear antennas, and a flat 3 dB between circular and linear.

Try it

The antenna explorer draws each antenna’s beam to scale on a plan view. Pick one, drag the tag out until it stops reading, then turn it. Switch to linear and turn the tag to 90 degrees.

That’s the call we get. Now you’ve seen it before it happens on your floor.

Next step

Tell us what you are trying to trace.

Describe the flow: what arrives, what happens to it and what leaves. We will tell you what it takes to capture it.