[email protected] Product Catalog → India RFID Store

RFID Tag Sensitivity in dBm: How to Read a Tag Datasheet

RFID tag sensitivity in dBm is the smallest radio power a passive UHF tag’s chip needs before it will answer a reader (read sensitivity) or accept new data (write sensitivity). It is negative because it is a tiny fraction of a milliwatt, so more negative means more sensitive. It is not a read range: it is a typical value from a reference measurement, and your distance depends on the tag antenna, what the tag is fixed to and the reader. This guide reads a datasheet line by line, for passive UHF (EPC Gen2, ISO/IEC 18000-63, RAIN RFID) only.

What the dBm figure on a tag datasheet actually measures

dBm is a power level in decibels relative to one milliwatt, as defined in ITU-R Recommendation V.574. 0 dBm is 1 mW and every 10 dB is a factor of ten, so −24 dBm is about 4 µW, −22 dBm about 6.3 µW, and +33 dBm, a reader-side figure, about 2 W.

The tag figure is the chip’s sensitivity threshold: the minimum RF power that turns the chip on. Nikitin and Rao’s analysis of passive UHF performance limits (IEEE APS, 2006) calls it the most important tag limitation, because the lower the threshold, the farther away the tag can be detected.

Chip sensitivity or tag sensitivity: find out which number you are looking at

One chip can be described by three numbers. NXP’s UCODE 8 datasheet (Rev. 3.6, December 2021) headlines −23 dBm read and −18 dBm write. Its characteristics table gives −22.9 dBm typical for read: the power needed to process a QUERY command, measured on the chip with a 50 Ω source. A footnote restates it as about −23 dBm tag sensitivity on a 2.15 dBi antenna. Write is −17.8 dBm typical, stated as tag sensitivity on a 2.15 dBi antenna.

An inlay datasheet may give the chip’s figures rather than a measurement of the finished inlay: a May 2025 M730/M750 inlay datasheet lists −24.0 dBm read and −21.0 dBm write under “IC specifications”, and its read figure matches Impinj’s brief.

Chip and sourceReadWriteBasis stated
NXP UCODE 9 (Rev. 3.5, Feb 2025)−24 dBm−22 dBmTypical; tag sensitivity on a 2.15 dBi antenna
NXP UCODE 8 (Rev. 3.6, Dec 2021)−22.9 dBm typ.−17.8 dBm typ.Read at chip level, 50 Ω source; write on a 2.15 dBi antenna
Impinj M730, M750, M770, M775 (M700 brief)−24 dBmNot in product briefNone stated (product brief)
Impinj M780, M781 (same brief)−23.5 dBmNot in product briefNone stated (product brief)
Impinj M830, M850 (M800 spec table)−25.5 dBmNot in product briefNone stated; brief claims up to 2 dB more with Gen2X Performance

NXP marks these as typical values, not guaranteed minimums, and Impinj’s briefs say performance is “based on Impinj’s modeling and test data”. The Impinj write cells are empty only because the briefs omit write; ask for Impinj’s own datasheet write figure, and the basis it is stated on, before comparing it with NXP’s. The extra M800 margin needs a reader that supports Gen2X (see our Gen2X and Gen2v3 guide). Choosing a family is covered in our UHF chip comparison and the store’s Impinj M700 vs NXP UCODE 9 guide. Here the point is narrower: compare dBm figures only when they share a basis.

Read sensitivity versus write sensitivity

On both NXP chips the write figure is less negative, because writing needs more power than reading, so one tag has a read range and a shorter write range. In free space:

  • UCODE 9: a 2 dB gap puts write-limited distance at about 79% of read-limited.
  • UCODE 8: on a like-for-like tag basis (about −23 dBm read and −17.8 dBm write, both referenced to a 2.15 dBi antenna), a gap of about 5.2 dB puts it at about 55%.

That matters wherever you encode at a distance: in-line commissioning, a print-and-encode station at the edge of its field, or rewriting a returnable asset’s EPC. A tag that reads at a given distance may not accept a write there; for voided labels, see our encoding-failure troubleshooting guide.

Nikitin and Rao’s forward-link condition shows what a dBm figure buys: power reaching the chip is reader e.i.r.p. × path loss × tag antenna gain × polarisation match × impedance-match coefficient (at most 1), and the tag answers only if that reaches the chip threshold. Free-space path loss is (λ/4πd)², so range scales with the square root of the margin:

  • Every 6 dB of sensitivity or link margin roughly doubles free-space range.
  • 2 dB is worth about 26% more; 1.5 dB (−24 against −25.5 dBm) about 19%.
  • The ratio is 10^(dB/20).

In clutter, path loss goes as d^−n with n between 1 and 4, so this is a free-space rule; where n exceeds 2, each decibel buys less distance.

India’s licence-exemption rules, G.S.R. 853(E) of 10 December 2021 (Table IV), permit 2 W e.r.p. in four channels of up to 200 kHz, which is about 35.15 dBm (3.28 W) e.i.r.p.; the two scales differ by about 2.15 dB. US rules (47 CFR 15.247) allow about 4 W e.i.r.p. for frequency-hopping readers in 902–928 MHz. A free-space range calculated at 4 W is about 10% longer than at India’s limit, so ask which power an imported range plot assumed.

An idealised calculation, for illustration only:

  1. Reader at India’s limit: about 35.15 dBm e.i.r.p.
  2. Tag antenna gain 2.15 dBi, perfect match and alignment.
  3. Chip threshold −24 dBm, so allowable path loss is 35.15 + 2.15 + 24 ≈ 61.3 dB.
  4. At about 866 MHz, that is roughly 32 m of free space.
  5. Add the 3 dB of a circularly polarised reader antenna reading a linear tag: about 23 m.

That is an arithmetic ceiling for an imaginary tag in empty space, forward link only, not any product’s range; the next sections explain why installations fall short.

Why free-space range is a ceiling: orientation, material and frequency

Orientation. Range is greatest along the tag antenna’s maximum gain, and the 3 dB circular-to-linear polarisation loss costs roughly 29% of free-space range. The M730/M750 inlay datasheet specifies which edge should lead and advises testing under application conditions; our antenna gain and polarisation guide covers the reader side.

Material. Fixing a tag to an object, or near others, changes its antenna characteristics and degrades gain and impedance match. Liquids are the extreme case. In a 2024 study at 915 MHz (Claucherty et al., Foods), one tag’s theoretical read range was 16.34 m on an empty PET bottle, from a measured threshold of −20.78 dBm. On a bottle of deionised water its threshold rose to +8.37 dBm, about 29 dB worse. Our fresh-food tagging guide covers the rest of that study. Metal needs a dedicated construction, such as our on-metal tags. The M730/M750 inlay datasheet plots theoretical range only for free air, cardboard, PVC and fabric, from laboratory conditions.

Frequency. Chip impedance changes across the UHF span: UCODE 8 lists 15−j265 Ω at 866 MHz, 14−j252 Ω at 915 MHz and 13−j242 Ω at 953 MHz, and the tag antenna has to be matched to it. A tag tuned for one region’s frequencies can be mismatched at another’s. UCODE 9’s “840–960 MHz” input frequency is the chip’s global span, not where a tag built on it is tuned, so ask for performance measured inside India’s licence-exempt band, 865–868 MHz under G.S.R. 853(E) of 2021.

Self-tuning. UCODE 9’s Self-Adjust selects one of three input-capacitance values at start-up, and Impinj describes AutoTune V2 as optimising performance to the tag’s environment. Treat them as narrowing the gap, not as validating a free-air figure on your product, and scope read range against your real environment first.

Where the reader comes in: e.r.p. ceiling, antenna gain and receive sensitivity

Forward link. Because India’s 2 W cap is on e.r.p., it applies to radiated power (conducted output plus antenna gain in dBd, which is dBi minus 2.15, minus cable loss), not the reader dial, so at the ceiling a higher-gain antenna means a lower conducted setting, not more power at the tag. Our tag-not-reading troubleshooting guide walks through the subtraction.

Reverse link. Nikitin and Rao define reader sensitivity as the weakest tag signal a reader can still pick out and decode, normally specified against a target signal-to-noise ratio or error rate; receiver design, protocol and interference all shift it. Inlay range plots assume a value (the M730/M750 inlay datasheet assumes 29 dBm reader power and −70 dBm antenna sensitivity); if your reader differs, the plot does not describe your system.

Which side limits? In Nikitin and Rao’s 2006 example (−10 dBm chip, 36 dBm e.i.r.p., −80 dBm reader sensitivity) the tag limited range to 20 ft against the reader’s 120 ft, and the authors concluded the tag typically limits. That chip was over 12 dB less sensitive than those tabled above, so check both sides on a modern link.

A datasheet, line by line

Datasheet lineWhat it really tells youWhat to ask for in writing
Read sensitivityTypical power at which the chip answers a queryExact IC part; chip-level, 2.15 dBi reference or finished tag; frequency
Write sensitivityTypical power to accept a write; less negative than readSame basis; the encoding distance your process needs
Operating frequency, e.g. 840–960 MHzThe chip’s global span, not the tag’s tuningMeasured performance in India’s licence-exempt band
Theoretical read rangeArithmetic from assumed reader power and receive sensitivityAssumed power, its scale (conducted, e.r.p., e.i.r.p.) and receive sensitivity
Mounting materialFree air or reference substratesA measured result on your substrate
OrientationWhich edge should lead; loss when rotatedYour process orientation; reader antenna polarisation used
TemperatureOperating and storage are different rangesBoth separately, and which applies while reading

Turning it into a testable tender line

  1. Name the exact IC part and variant, not a family.
  2. State read and write sensitivity with their basis: chip-level, tag on a 2.15 dBi reference antenna, or measured on the finished tag, plus the frequency.
  3. State the mounting material and orientation the figures must hold on.
  4. Add an acceptance test: read rate over a stated number of passes at a stated distance, tag on the real substrate, read by the offered reader and antenna. If you encode on site, add the same for writes.

ISO/IEC 18046-3:2020 defines tag performance test methods whose summarised reports form a unified tag datasheet. Clause 8 covers propagative tags, including ISO/IEC 18000-63, with tests for the minimum power threshold for identification, read and write. Clause 5.8 says that where the manufacturer specifies a mounting material for a UHF tag, the tests are run on that material and in free air. Ask bidders which method produced their figures. See our tender BOQ guide for the rest of the specification and the tag manufacturer checklist for sample rounds.

Where Identium fits

Identium Tech Solutions Pvt. Ltd. is an RFID manufacturer in New Delhi that designs and makes UHF tags, labels and inlays. Ask us, in writing, for the basis of any sensitivity figure on our datasheets. Passive tags need no WPC approval or BIS registration; our BIS and WPC buyer’s guide explains where approvals do apply. Custom tag orders typically turn around in 5–10 days, and every order is quoted against specification. Send the surface, reader and tender line through our contact page.

Further reading on Identium One

Frequently asked questions

Is a more negative dBm sensitivity figure better?

Yes, provided the two figures share a basis. A more negative number means the chip responds to less power: −24 dBm is about 4 microwatts, while −22 dBm is about 6.3 microwatts. Check first whether each figure is chip-level, referenced to a tag on a 2.15 dBi antenna or measured on a finished tag, because those presentations differ even for one chip.

Why is write sensitivity worse than read sensitivity?

The chip needs more power to write data than to answer a read, so its write threshold is less negative. NXP's UCODE 9 datasheet lists −24 dBm read and −22 dBm write, which in free space puts the write-limited distance at about 79% of the read-limited one. It matters wherever you encode or re-encode tags at a distance.

How much extra range does 1–2 dB of tag sensitivity buy?

In free space, 1.5 dB is worth about 19% more range and 2 dB about 26%, because range scales with 10 to the power of dB/20. On a real asset, orientation, mounting material and frequency tuning can cost far more than that, so the gain is real but easily swamped. Our chip guide covers when the choice of chip deserves scrutiny.

Can I put a dBm sensitivity figure straight into a tender?

Not on its own. Name the exact IC part, state read and write sensitivity with their basis and frequency, give the mounting material and orientation, and then add an acceptance test of read rate over a stated number of passes at a stated distance on the real substrate. The dBm figure helps you compare bids; the acceptance test is what you can verify at commissioning.

Why does a datasheet range plot not match my installation?

A range plot is arithmetic from an assumed reader power and receive sensitivity, drawn in free air or on reference substrates under laboratory conditions. Your reader's power, the scale it is quoted on (conducted, e.r.p. or e.i.r.p.), its receive sensitivity, the tag's orientation and the material it is fixed to will all differ. Ask for those assumptions and a measured result on your own substrate.

Need pricing, samples or a demo?

Talk to our RFID specialists — we manufacture in India and ship nationwide.