Technology 📅 September 25, 2026 ⏱ 18 min read

How Far Will a UHF RFID Tag Actually Read? The Calculation, Worked Three Times

Worked RFID link-budget calculation showing forward and return link range at three regulatory power ceilings

“What is the read range?” is the first question in every RFID enquiry we receive, and it is the question the published literature answers in the most fragmented way. Inlay datasheets print range as an unlabelled curve. Reader datasheets print conducted power in dBm. Regulations print radiated power in watts e.r.p. Each document is correct in its own unit, and the joining up is left to the buyer — so two quoted numbers, 9 m and 12 m, invite comparison while resting on different power levels, different units and different tags. This article does the arithmetic instead. One tag chip, one antenna, three legal power ceilings, and every step shown with its units, so you can re-run it on your own parts. The answer is 29.7 m in free space and about 9 m on a loading dock — and the gap between those two numbers is a ledger of losses you can itemise, predict and measure.

What an inlay datasheet tells you about range, and what you supply yourself

Start with the document a buyer is usually handed. The Avery Dennison Smartrac AD Trap M730/M750 product data sheet (document code AD-Trap-M730-M750-09-21-EN-DS) is a good, representative example from one of the largest inlay makers in the world. It is a thorough document: chip options, EPC and user memory, die-cut dimensions of 11 × 25 mm, PET substrate, standard pitch 32 mm, operating temperature −40 °C to 85 °C, ISO 18000-63 and EPC Class 1 Gen 2 conformance, RoHS and REACH statements.

On range, it carries exactly two chart headings — Read range with Impinj near field Mini-Guardrail antenna and Read range — under a single caption:

“All graphs are indicative: performance in real life applications may vary.”

That caption is honest, and the charts are genuinely useful for comparing one inlay against another from the same maker, measured the same way. Three further inputs turn a range curve into a site prediction: the reader transmit power, the antenna gain, and whether the axis is built on an e.r.p. or an EIRP basis. Those live in the reader datasheet, the antenna datasheet and the regulation, and once you supply them the curve becomes portable — to your site, your reader, your regulatory domain, and to a like-for-like comparison against any other inlay.

That is what the rest of this article assembles: one tag chip, one real antenna, three legal ceilings, three forward-link answers, three return-link answers, and then the itemised losses that close the gap between free space and a real dock door.

Putting the three legal ceilings into one unit

The first real obstacle is that each of the three big regulatory regimes states power in its own unit. Europe and India specify e.r.p. — effective radiated power, referenced to a half-wave dipole. The United States specifies conducted power plus an antenna gain allowance, which yields EIRP, referenced to an isotropic radiator. Friis needs EIRP. So everything converts first.

The conversion constant is 2.15 dB, the gain of a half-wave dipole over isotropic. India’s notification authorises the conversion explicitly. G.S.R. 853(E) of 10 December 2021, Rule 2(1)(c), defines the term as “the product of the power supplied to an antenna and its gain in a given direction relative to a half-wave dipole.” Because the reference is a dipole, dBm e.r.p. + 2.15 dB = dBm EIRP.

Working the three ceilings:

India and the ETSI lower band, 2 W e.r.p. 10 × log10(2000 mW) = 33.01 dBm e.r.p. Add 2.15 dB → 35.16 dBm EIRP = 3.28 W EIRP. The RAIN Alliance reaches the same constant independently, writing “2 Werp (EU lower band, 33 dBm erp, which is 35.15 dBm eirp).”
FCC, 902–928 MHz. 47 CFR 15.247(b)(2) allows “1 watt for systems employing at least 50 hopping channels” — 30 dBm conducted. Paragraph (b)(4) states that this limit “is based on the use of antennas with directional gains that do not exceed 6 dBi,” and that above 6 dBi the conducted power must be reduced “by the amount in dB that the directional gain of the antenna exceeds 6 dBi.” So 30 + 6 = 36.00 dBm EIRP = 3.98 W EIRP.
ETSI upper band, 915–921 MHz, 4 W e.r.p. 10 × log10(4000 mW) = 36.02 dBm e.r.p. Add 2.15 dB → 38.17 dBm EIRP = 6.56 W EIRP. EN 302 208 V3.3.1 (2020-08) is titled for exactly these two tiers and states in its scope: “Power limits up to a maximum of 2 W e.r.p. are specified for this equipment in the frequency band 865 MHz to 868 MHz and up to a maximum of 4 W e.r.p. in the frequency band 915 MHz to 921 MHz.”

Ceiling	W e.r.p.	dBm e.r.p.	dBm EIRP	W EIRP
India / ETSI lower band	2	33.01	35.16	3.28
FCC 902–928 MHz	— (30 dBm conducted + 6 dBi)	—	36.00	3.98
ETSI upper band 915–921 MHz	4	36.02	38.17	6.56

One clarification worth making, because it is routinely muddled: the 2 W e.r.p. figure in 865–868 MHz is the RFID interrogator tier specifically, and that is the row an RFID portal is designed to. The same Indian notification also carries 25 mW e.r.p. and 500 mW e.r.p. tiers, which belong to other short-range device categories sharing the same band.

Method check: reproducing RAIN’s published 40% figure before trusting it on India

Before running the method on a new case, run it on a published one. The RAIN Alliance’s System Design Guidelines V2, Table 3, compares Europe’s two bands and credits the upper band with “40% more read range” on the strength of its higher power tier. If our arithmetic reproduces that, the method is sound.

The two ceilings in EIRP: 38.17 dBm against 35.16 dBm. The difference is 3.01 dB. Received power falls as 1/r², so range scales as the square root of power, which means a dB difference converts to a range ratio through 10^(dB/20), not 10^(dB/10):

10^(3.01 / 20) = 10^0.1505 = 1.414
That is a 41.4% range increase — RAIN’s 40%, reproduced.

The same divisor-of-20 rule settles the most persistent question in the category. Doubling reader power is +3.01 dB, and +3.01 dB buys 41% more range; a doubling of range would take +6.02 dB, four times the power, which puts the useful levers in antenna gain, tag sensitivity and geometry rather than in transmit power. Impinj confirms the same law from the other direction in the M830/M850 datasheet, describing its short-range privacy mode as one that “reduces tag read range by roughly 10× (20 dB)”: 10^(20/20) = 10, exactly.

Two further rows of RAIN’s Table 3 matter when you read a range claim, because they change the rate rather than the distance: transmit channel width is 200 kHz in the lower band against 400 kHz in the upper, reader channel spacing 600 kHz against 1200 kHz, and the tag backscatter power allowance 10 µW against 100 µW. That last one is a clue we will come back to, because a tenfold backscatter allowance is a statement about the reader’s receiver rather than the tag’s range.

The forward link, worked three times

The forward link is the reader powering the tag. A passive tag wakes once the power arriving at its chip terminals reaches the chip’s sensitivity threshold, so the forward-link range is simply the distance at which incident power equals that threshold. Rearranging Friis for distance:

r = (λ / 4π) × √(EIRP × G_tag / P_th)

Four inputs, all of which come from a document. λ comes from the channel: 299 792 458 / 866.3 × 10^6 = 0.34606 m at India’s channel centred on 866.3 MHz, and 0.32764 m at 915 MHz. EIRP comes from the table above. G_tag is the tag antenna gain as a linear ratio. P_th is the chip read sensitivity.

For P_th we use a current part with a published figure: the Impinj M830/M850 Series tag chip datasheet v2.1, Table 8, gives read sensitivity −25.5 dBm and write sensitivity −20.0 dBm, both qualified “Measured in a 50-ohm system using a response to a Query command with a +2.15 dBi dipole.”

Working India in full, with a 0 dBi tag so the tag gain term is 1. Every intermediate is printed at the precision it is used, so keying these five lines into a calculator returns the same answer:

EIRP = 2000 mW e.r.p. × 10^(2.15/10) = 2000 × 1.6406 = 3281.2 mW
P_th = −25.5 dBm = 10^(−2.55) mW = 0.0028184 mW
EIRP / P_th = 3281.2 / 0.0028184 = 1 164 207
√1 164 207 = 1078.98
λ / 4π = 0.34606 / 12.56637 = 0.027539 m
r = 0.027539 × 1078.98 = 29.71 m

Repeating at each ceiling. Note the two right-hand column pairs: one computes each band at its own wavelength, which is the physically correct answer; the other holds the wavelength fixed at India’s to isolate the power variable alone. Reading the difference between those columns is the whole point of the next section.

Ceiling	dBm EIRP	Channel used	λ (m)	0 dBi tag	2.15 dBi tag (hypothetical)	Power term only, λ fixed at 0.34606 m
India / ETSI lower	35.16	866.3 MHz	0.34606	29.7 m	38.1 m	29.7 m
FCC 902–928	36.00	915 MHz	0.32764	31.0 m	39.7 m	32.7 m
ETSI upper 915–921	38.17	918 MHz	0.32657	39.7 m	50.8 m	42.0 m

A second current part for cross-reference: NXP’s UCODE 9 (SL3S1206 data sheet Rev. 3.5, 12 February 2025) publishes read sensitivity −24 dBm and write sensitivity −22 dBm, with footnote [1] reading “Tag sensitivity on a 2.15 dBi gain antenna.”

Here is the trap that invalidates most published range maths. Both vendors measure chip sensitivity through a 2.15 dBi reference dipole. That gain is already inside the published number. If you then add 2.15 dB again for “the tag antenna,” you have double-counted, and your answer is 28% long — exactly the 29.7 m to 38.1 m step in the 2.15 dBi column above, which is why that column is labelled hypothetical rather than offered as a recommendation. Use the published figure as it stands, with G_tag = 1, because the reference gain is already inside it; where you have a measured gain for your specific inlay, work from the chip’s conducted threshold instead.

The India-versus-FCC result nobody publishes

The received wisdom across the industry is that an ETSI-lower-band or Indian installation reads dramatically shorter than a North American one, because 2 W sounds like half of 4 W. Put both into EIRP and the gap almost vanishes.

India / ETSI lower: 35.16 dBm EIRP
FCC: 36.00 dBm EIRP
Difference: 0.84 dB — a fraction of a decibel either side of parity, rather than the factor of two the watt figures suggest

Converting that to range: 10^(0.84 / 20) = 1.102, a 10.2% advantage to the FCC ceiling. Then the wavelength term pushes back the other way, because r is directly proportional to λ and the lower band has the longer wavelength: 0.32764 / 0.34606 = 0.9468, or 5.32% against the FCC.

Net, same tag and same antenna: 1.102 × 0.9468 = 1.043. About 4.3%. Which is exactly the 31.0 m against 29.7 m in the table above, arrived at a second way.

Commercially that result is worth more than it looks. A 4.3% range difference sits inside the measurement noise of a real site survey. It means band choice is a regulatory and market-access question, and performance tracks geometry and tag choice — the same portal geometry, antenna count and dwell time will work in Delhi, Dubai, Frankfurt and Chicago, and what changes between them is the approval paperwork and the reader’s regional configuration. That is why we build the same reader platform in FCC-band and ETSI-band variants configured per order rather than treating them as different products, and it is the single most useful thing an exporter can know about this calculation. It also means India mirrors Europe’s lower band exactly: RAIN’s own regional settings table lists India’s IN8A profile and Europe’s EU8A with identical entries — 200 kHz transmit bandwidth, 600 kHz spacing, 320 kHz backscatter link frequency, dense reader mode M=4, adaptive frequency agility, max 2 W e.r.p.

Write range is the sharper gap, and it is the one that decides project budgets. On the Impinj part, read sensitivity −25.5 dBm against write sensitivity −20.0 dBm is a 5.5 dB penalty: 10^(−5.5 / 20) = 0.531, so writing reaches 47% less far than reading. Size an encoding station off the write figure.

Chip	Read sensitivity	Write sensitivity	Gap	Read range (India, 0 dBi tag)	Write range
Impinj M830 / M850 (datasheet v2.1, Table 8)	−25.5 dBm	−20.0 dBm	5.5 dB	29.7 m	15.8 m
NXP UCODE 9 (SL3S1206 Rev. 3.5)	−24 dBm typ	−22 dBm typ	2.0 dB	25.0 m	19.9 m

Read that table twice. The Impinj part reads 19% farther, and writes 21% shorter. If your application is inventory counting in a warehouse management system, the first column decides. If it is an inline encoding and print-apply station, the last column decides, and the ranking inverts. The better chip is the one matched to the operation that decides your project — and the ranking follows that operation, not the chip.

The return link, and when it governs

Everything so far is half the system. The tag also has to be heard. The return link is backscatter, so the signal traverses the path twice and falls as 1/r⁴ rather than 1/r² — which makes it far more sensitive to every decibel than the forward link is. Solving for distance:

r = [ EIRP × G_ant × G_tag² × M × (λ / 4π)^4 / P_rx ]^(1/4)

The parameter most often treated as unobtainable is published. M is the modulation factor, M = |ΔΓ|² / 4, and the Impinj M830/M850 datasheet v2.1, Table 10, gives “Change in Modulator Reflection Coefficient |ΔΓ| due to Modulation” as 0.8 typical, defined as |Γ_reflect − Γ_absorb| and explicitly tied to the same conditions as the sensitivity figures in Table 8. So M = 0.64 / 4 = 0.16.

P_rx rewards care, because it is three numbers rather than one. The Impinj R700 Series datasheet v2.1 gives receive sensitivity as Ideal −94 dBm, Typical −93 dBm, Challenging −88 dBm, and Table 5 defines each condition physically — Ideal at 4.5 ns time delay and 22+ dB return loss, Typical at 6.0 ns and 15 dB, Challenging at 38.2 ns and 12 dB. Those rows describe your installation rather than a marketing tier: a long cable run into a poorly matched antenna surrounded by steel racking is the Challenging row. The measurement condition is stated too — “Data collected using CISC conducted sensitivity test, 90% PSR. Using RF mode FCC DRM M8 @ 30 dBm” — following, as Impinj notes, “the RAIN Alliance and ISO/IEC 18046-2:2020 recommended receive sensitivity methodology.” A 6 dB spread between the best and worst published figure is a factor of 1.41 in return-link range, because the return link falls as 1/r⁴ and so scales as 10^(dB/40). Substituting a rounded −80 dBm because it looks conservative introduces a 13 dB error and a factor of 2.1.

Computing the return link with an 8.5 dBi reader antenna, 0 dBi tag and M = 0.16, alongside the forward-link results:

Ceiling	Forward link	Return, Typical (−93 dBm)	Return, Challenging (−88 dBm)	Which governs, Typical	Which governs, Challenging
India / ETSI lower	29.7 m	45.4 m	34.1 m	Forward	Forward
FCC 902–928	31.0 m	45.2 m	33.9 m	Forward	Forward
ETSI upper 915–921	39.7 m	51.0 m	38.2 m	Forward	Return

RAIN gives this the right vocabulary, naming the two shortfalls “reader gap — the range loss caused by the reader sensitivity limits” and “tag gap — the range loss caused by the tag minimum operating power limit,” and instructing designers to take “the minimum range on forward and return link.”

The bottom-right cell is the result worth carrying away. At the upper band’s ceiling in a reflective installation, the receiver becomes the governing term: the forward link reaches 39.7 m while the return link sets the system at 38.2 m, so the design range is the receiver’s number and reader receive sensitivity is where the next decibel is bought. Measured against India’s 29.7 m, the realised gain is 28.7%, against RAIN’s headline 41% for the power step alone. This is precisely the crossover Chakraborty, Roy and Jandhyala of the University of Washington predicted in Revisiting RFID Link Budgets for Technology Scaling: Range Maximization of RFID Tags, published in IEEE Transactions on Microwave Theory and Techniques, vol. 59, no. 2, pp. 496–503, February 2011: “a cross-over between uplink and downlink range emerges. In other words, future passive tags with improved sensitivity may be read range limited.” Their analysis tabulates tag sensitivities of 0, −10, −25 and −35 dBm; the −25 dBm row is today’s M830-class part. The future they modelled in February 2011 is the parts list you are buying from now. It also explains RAIN’s tenfold backscatter allowance in the upper band — the regulation anticipated that the receiver, rather than the tag, becomes the constraint.

From the RF port to the air

Every number above is EIRP in the air. What you actually configure is conducted power at a coaxial port, and three things sit between the two.

First, the reader’s own cap. The R700 datasheet gives transmit power as 10–30 dBm on PoE, and on PoE+ 10–31.5 dBm in the ETSI lower band and 10–33 dBm for FCC and the ETSI upper band. Whether the regulation or the reader sets the level depends entirely on your antenna gain, and the crossover is easy to locate. To radiate 35.16 dBm EIRP through an 8.5 dBi antenna with 0.75 dB of cable loss you need 35.16 − 8.5 + 0.75 = 27.4 dBm conducted — comfortably inside the reader’s range, so the regulation sets the level and you have 4 dB spare. Swap to a 2 dBi antenna and the same EIRP needs 33.9 dBm conducted, above the 31.5 dBm the reader offers in that band: the reader now sets the level, you radiate 2.4 dB under the legal ceiling, and 24% of your range has gone to an antenna choice.

Second, cable loss. Times Microwave publishes a closed-form fit for LMR-400 rather than making you read a graph:

Attenuation (dB per 100 ft) = 0.12229 × √F_MHz + 0.00026 × F_MHz

Verify it before trusting it. At 900 MHz: 0.12229 × 30 + 0.00026 × 900 = 3.6687 + 0.234 = 3.903 dB/100 ft, against the 3.9 printed in the same datasheet’s table. Now evaluate it where RFID actually lives, which the tabulated rows skip:

866.3 MHz: 3.825 dB/100 ft = 0.1255 dB/m
915 MHz: 3.937 dB/100 ft = 0.1292 dB/m
A 6 m portal run at 866.3 MHz therefore costs 6 × 0.1255 = 0.75 dB

Here is the part worth internalising: 0.75 dB of cable loss costs you nothing at all when you have conducted headroom. The compliance test measures radiated power, so with the 8.5 dBi antenna above you simply set the reader 0.75 dB higher and arrive at the same legal EIRP. Even a 20 m run (2.51 dB) only needs 29.2 dBm, still inside the cap. Cable loss becomes real range loss once the reader’s conducted ceiling sets the level — at which point every 0.75 dB is 8.3% of your range, or 2.5 m off the 29.7 m figure. Long runs are an argument for high-gain antennas.

Third, a real antenna to anchor the sum. The Times-7 A5010 SlimLine: 8.5 dBic typical far-field gain, 68° 3 dB beamwidth in both planes, right-hand circular polarisation, VSWR 1.3 typical across the band, 20 dB front-to-back ratio, axial ratio 1 dB typical, 50 Ω, IP67, SMA female side connector, 250 × 250 × 14 mm. Two notes on that gain figure. It is quoted in dBic, against a circularly polarised reference, and the polarisation loss factor between circular and linear polarisation is one half. As the MediaTek patent US12531585B2 puts it, “there is a 3 dB polarization loss when a circularly polarized signal is received by a linearly polarized antenna. Similarly, there is a 3 dB polarization loss when a linearly polarized signal is received by a circularly polarized antenna.” A circularly polarised reader antenna illuminating a linearly polarised dipole tag therefore gives up about 3 dB, so budget nearer 8.5 − 3 = 5.5 dBil where tag orientation is uncontrolled. And the 68° beamwidth is what turns a range number into a pallet-level portal geometry — at 3 m a 68° cone is roughly 4 m across, which is what decides antenna count, more than the maximum range does.

The loss ledger that takes 30 m down to 9 m

Time to close the loop. Times-7 markets the same A5010 at “Up to 9m / 29 ft. read range.” Our free-space calculation said 29.71 m. Both are correct, and the distance between them is arithmetic:

20 × log10(29.7 / 9.0) = 20 × 0.5186 = 10.37 dB

So the manufacturer’s honest field figure implies about 10.4 dB of cumulative derating against theory — a ledger, and after enough deployments you can predict each line of it. The table below is an Identium survey allowance table, drawn from our own UHF deployments rather than from a published standard. It is the range we budget with before a survey, and we publish it so that the aggregate can be checked: the 10.4 dB total is independently anchored by Times-7’s own published field figure against our computed 29.7 m, which is what the arithmetic above shows.

Loss (Identium survey allowance)	Allowance	Cause
Cable and connectors	1–3 dB	Run length and connector count, per the LMR-400 fit above
Polarisation mismatch	0–3 dB	Circular reader antenna against a linear tag dipole, per the 3 dB polarisation loss factor above
Tag orientation and tumble	3–6 dB	Tag gain is directional; a tumbling carton presents its null some of the time
Multipath fading and nulls	2–6 dB	Reflections from racking, floor and vehicles adding destructively
Material absorption and detuning	1–4 dB	Liquids, metals, dense cardboard, moisture shifting inlay resonance
Multi-tag contention	1–3 dB	Collisions and reduced per-tag dwell in a large population
Cumulative	8–25 dB	Totals land well below the sum of the maxima; 10–14 dB is a realistic dock-door figure

What each total does to the 29.71 m baseline, at 10^(−dB/20):

Total derating	Range factor	Range from 29.71 m
6 dB	0.501	14.9 m
8 dB	0.398	11.8 m
10.4 dB	0.302	9.0 m
12 dB	0.251	7.5 m
15 dB	0.178	5.3 m
20 dB	0.100	3.0 m

Two scope notes. The whole calculation is a far-field calculation. The standard boundary is the Fraunhofer distance: as Monemi, Bahrami, Rasti and Latva-aho state it, “The Fraunhofer distance on the antenna’s boresight is obtained as d = 2D2 /λ, where D is the maximum dimension of the antenna and λ is the wavelength.” For the A5010’s 250 mm aperture at 866.3 MHz that evaluates to 2 × 0.25² / 0.34606 = 0.36 m. The paper notes that for single-element thin-wire dipole antennas the expression “is accurate only for antennas where D ≥ 5λ”. The A5010’s 0.25 m aperture is smaller than the 0.34606 m wavelength, so we treat the wavelength as the governing scale and begin the working far field at a small multiple of it (3λ = 1.04 m), as a survey convention rather than a result carried over from that paper. Everything from about a metre outward is what this article computes, and closer work is a near-field problem with its own method. RAIN draws the same line, scoping its own guidance to the far field and describing the boundary in the same two terms — the wavelength and the maximum dimension of the antenna.

And this is why a site survey is the deliverable. The 10.4 dB is measurable and site-specific. A survey measures your racking, your product, your tag orientation and your floor, and replaces an 8–25 dB range with a number. Everything upstream of that — ceiling, antenna, cable, chip — you can now compute before anyone visits.

Run it on your own numbers

The worksheet, in order. Each line names the document the input comes from.

Ceiling → dBm EIRP. If your regulation states e.r.p., add 2.15 dB. Source: the notification itself — for India, G.S.R. 853(E) of 10 December 2021, Rule 2(1)(c) for the dipole reference and the RFID interrogator row for the 2 W limit, which also carries the channelling rules you must respect: maximum 200 kHz channels, transmission only on the four channels centred at 865.7, 866.3, 866.9 and 867.5 MHz, continuous transmission on a channel not exceeding 4 s with at least 100 ms between consecutive transmissions, tags replying at −20 dBm e.r.p., reference standard EN 302 208.
λ = 299 792 458 / f. Use your actual channel centre rather than the band edge.
Antenna gain and cable loss. Antenna datasheet, and the LMR-400 fit evaluated at your frequency. Check whether the reader’s regional cap or the regulation sets the level — conducted needed = EIRP − G_ant + cable loss.
Tag read and write sensitivity. Chip datasheet: Impinj M830/M850 v2.1 Table 8, or NXP SL3S1206 Rev. 3.5. Take the published figure as it stands — the 2.15 dBi reference gain is already inside it.
Forward range. r = (λ/4π) × √(EIRP × G_tag / P_th). Run it once on read sensitivity and once on write sensitivity.
Return range. r = [EIRP × G_ant × G_tag² × M × (λ/4π)^4 / P_rx]^(1/4), with M = |ΔΓ|²/4 from the chip datasheet and P_rx from the reader datasheet — choosing the Typical or Challenging row that honestly describes your site.
Take the minimum of forward and return, then subtract your loss ledger. That is your design range.

Two closing notes on scope. Every figure in this article is passive UHF. Our active 2.4 GHz line is a different link entirely, with its own budget: up to 100 m typical and 400 m line-of-sight, confirmed by site survey. And where you want the link budget monitored after commissioning rather than calculated once, reader-side telemetry is the place it lives — ReaderSense Edge MDM reports the configured power, the realised read rate and the per-antenna behaviour that tells you when a portal has drifted from its design.

Next in this series, the two calculations that sit either side of this one: a dock-door portal worksheet that turns a range number into antenna count, mounting height and dwell time, and a settings-level guide to why a portal misses tags and why it reads too many — session, Q, target flags, power and antenna geometry. If you would rather hand the arithmetic over, send us the geometry, the product and the tag population and we will return the budget with every assumption named, the way it is laid out above. We design and manufacture the UHF readers and write the software, so the numbers in a quotation come from the same worksheet as the numbers here.

Frequently asked questions

How far can a passive UHF RFID tag actually be read?

In free space, further than most people expect: with an Impinj M830-class chip at −25.5 dBm read sensitivity, a 0 dBi tag and India’s 2 W e.r.p. ceiling (35.16 dBm EIRP), Friis gives 29.7 m. In a real installation, expect roughly 6–9 m at a dock door. The difference is about 10–14 dB of cumulative loss — cable, polarisation mismatch, tag orientation, multipath, material absorption and multi-tag contention — and because range scales as 10^(−dB/20), 10.4 dB alone takes 29.7 m down to 9.0 m.

Why is my read range shorter than the tag datasheet suggests?

Because a range curve becomes a site prediction once three more inputs are supplied: the reader transmit power, the antenna gain, and whether the axis is e.r.p. or EIRP. Inlay datasheets are built for comparing inlays measured the same way — the Avery Dennison AD Trap M730/M750 sheet says so itself, captioning its charts “All graphs are indicative: performance in real life applications may vary.” Supply the three inputs from your reader datasheet, your antenna datasheet and your regulation, compute the link budget from the chip sensitivity figure, and you get a number that transfers to your reader, your power ceiling and your site.

Does an RFID reader in Europe or India read shorter than one in the United States?

Barely. Put both into the same unit and India’s 2 W e.r.p. is 35.16 dBm EIRP against the FCC’s 36.00 dBm EIRP — a difference of 0.84 dB. That is a 10.2% range advantage to the FCC ceiling, which the longer wavelength at 866.3 MHz then claws back by 5.32%. Net difference on the same tag and antenna: about 4.3%, or 31.0 m against 29.7 m. Band choice is a market-access and approvals question; performance tracks the geometry, the antenna and the tag.

What is the difference between the forward link and the return link in RFID?

The forward link is the reader powering the tag, and it falls as 1/r²; its range is set by tag chip sensitivity. The return link is the tag backscattering to the reader, and because the signal traverses the path twice it falls as 1/r⁴; its range is set by reader receive sensitivity and the chip’s modulation coefficient. RAIN calls the two shortfalls the “tag gap” and the “reader gap” and instructs designers to take the minimum of the two. With today’s parts the forward link usually governs — and at the ETSI upper band’s 4 W e.r.p. ceiling in a reflective installation the receiver governs instead, setting the system at 38.2 m against 39.7 m forward.

Should I use read sensitivity or write sensitivity to size an encoding station?

Always write sensitivity, and the gap is large. The Impinj M830/M850 datasheet gives read sensitivity −25.5 dBm and write sensitivity −20.0 dBm — 5.5 dB, which is 47% less range. At India’s ceiling that is 15.8 m for writing against 29.7 m for reading. The ranking between chips can also invert: NXP’s UCODE 9 reads shorter (25.0 m) and writes further (19.9 m), because its read-to-write gap is only 2 dB. Choose the part for the operation that decides your project.

How much read range does coax cable loss cost me?

Often nothing. Times Microwave’s fit for LMR-400 gives attenuation (dB/100 ft) = 0.12229 × √F_MHz + 0.00026 × F_MHz, which is 0.1255 dB/m at 866.3 MHz, so a 6 m run costs 0.75 dB. But the compliance test measures radiated power, so where you have conducted headroom you simply turn the reader up 0.75 dB and radiate the same legal EIRP. Cable loss becomes real range loss once the reader’s own conducted cap sets the level — and then 0.75 dB is 8.3% of your range. Long runs argue for higher antenna gain.

Does doubling reader power double read range?

It buys 41% more. Doubling power is +3.01 dB, and because received power falls as 1/r², range scales as 10^(dB/20): 10^0.1505 = 1.414. This is exactly why RAIN credits the ETSI upper band’s 4 W e.r.p. with “40% more read range” over the lower band’s 2 W. Doubling range takes +6.02 dB, a four-fold power increase, which is why the useful levers are antenna gain, tag sensitivity and geometry rather than transmit power. Impinj confirms the same law inversely, describing its short-range privacy mode as reducing range “by roughly 10× (20 dB).”