Technology 📅 September 6, 2026 ⏱ 12 min read

RFID Reader Power Budget: From dBm at the Port to Legal e.r.p.

A UHF RFID reader on a bench with its coaxial feeder, connectors and a circularly polarised panel antenna, annotated with the decibel losses and gains between the reader port and the radiated field

“What should I set the reader to?” is the most common question we get after a shipment lands, and the answer sits one step past the number printed on the reader. That number is conducted power at the antenna port. What the regulator limits is power leaving the antenna — after your coax has taken its share and your antenna has multiplied what survives. Between the two sits a chain of four or five decibel terms, including one 2.15 dB reference step that decides whether a build clears type testing on the first attempt. This article walks the whole chain in both directions: forward, from a port setting to a radiated figure; and backward, from a legal ceiling and a chosen antenna to the maximum port power you are allowed to set. Every cable figure below comes from the cable manufacturer’s own datasheet, every ceiling from the rule that sets it.

The chain, in one line

Every power argument — with a customer, a test house, or a commissioning engineer holding a laptop — reduces to one line:

Port power (dBm) − cable loss (dB) − connector loss (dB) + antenna gain (dBi) = EIRP (dBm)

And then one more:

EIRP (dBm) − 2.15 dB = e.r.p. (dBm)

That is the whole calculation. Everything else in this article is either a number to put into it or a consequence of getting it wrong.

Why regulators write limits two different ways

EIRP is referenced to a theoretical isotropic radiator. Effective radiated power is referenced to a half-wave dipole. India’s own rules define the term precisely: e.r.p. is “the product of the power supplied to an antenna and its gain in a given direction relative to a half-wave dipole.” A half-wave dipole has 2.15 dBi of gain over isotropic, so the identical radiated field is always 2.15 dB larger when you express it as EIRP than when you express it as e.r.p.

European and Indian short-range-device rules are written in e.r.p. The FCC works in conducted power with an antenna-gain condition, which people then quote as an EIRP number. Mix the two references and you pay for it in one of two ways. Treat a 2 W e.r.p. ceiling as if it were 2 W EIRP and you set the reader 2.15 dB low, which costs 22% of your free-space read distance. Treat a 2 W EIRP ceiling as if it were e.r.p. and you transmit 2.15 dB over the limit, which is the version that shows up in a test report.

What the number on the box actually is

The number on the box — “33 dBm reader”, “2 W reader” — is the input to this chain, and the output comes later. It is conducted power available at the RF port, measured into 50 ohms. It sets the ceiling of what you can ask for; what you may legally use follows from what you bolt to the port.

The four conversions, with tables

dBm to watts

These are the values you actually meet on a reader’s power menu and in a rule book.

dBm	Watts	Where you see it
20.0	0.10 W	Handheld and low-PoE-class fixed readers
25.0	0.32 W	Short-range desktop and kiosk readers
27.0	0.50 W	Half-watt class
30.0	1.00 W	FCC conducted ceiling; common PoE cap
31.5	1.41 W	Typical ETSI lower-band reader maximum
33.0	2.00 W	Typical FCC / upper-band reader maximum
36.0	4.00 W	The FCC 4 W EIRP figure, as a radiated total

Two ceilings expressed both ways, because you will need them constantly: 2 W e.r.p. = 33.0 dBm e.r.p. = 35.2 dBm EIRP, and 4 W e.r.p. = 36.0 dBm e.r.p. = 38.2 dBm EIRP.

dBi, dBd and dBic

dBi is gain over isotropic. dBd is gain over a half-wave dipole. dBi = dBd + 2.15. It is the same 2.15 dB as the EIRP-to-e.r.p. step, for the same reason, and holding that in mind makes the whole subject straightforward.

dBic is the gain of a circularly polarised antenna measured against a co-polarised circular reference — the figure almost every RFID panel antenna is sold on. Two things follow. For the regulatory sum, put the full dBic figure in as your gain; that is the conservative position and it keeps you under the ceiling whichever measurement convention your test house applies. For the link budget, a circular antenna illuminating a linearly polarised tag delivers about 3 dB less into that tag than the gain figure implies, because only the matched component of the field couples. Some integrators net that 3 dB off the antenna gain before doing the regulatory sum — ask which convention your declaration used before you spend the decibel.

Decibels to distance

Free-space field strength falls with distance, so power scales with the square of range. Every 6 dB doubles the theoretical distance and every 6 dB lost halves it. Below that, the useful shorthand is 20·log: 1.5 dB is about 1.19× range, 3 dB is about 1.41×, 10 dB is about 3.16×, 13 dB is about 4.47×. Real installations track the free-space curve loosely, but the ratio between two settings on the same site holds, which is what makes this arithmetic worth doing.

Coax loss at 866 and 915 MHz

Cable loss is the term most often estimated, and the one where the manufacturer has already published the answer. Times Microwave gives loss in dB per 100 ft as A·√F + B·F, with F in MHz. Evaluated at 866 MHz and 915 MHz and converted to metres, the LMR family lands as follows. Belden’s RG-58 type 8259 is tabulated directly by the manufacturer at 21.1 dB/100 ft at 900 MHz.

Cable	dB/m @ 866 MHz	dB/m @ 915 MHz	3 m run	7.5 m run	15 m run
RG-58 type (Belden 8259)	0.69 (at 900 MHz)	0.69 (at 900 MHz)	2.08 dB	5.19 dB	10.38 dB
LMR-195	0.358	0.368	1.07 / 1.10 dB	2.68 / 2.76 dB	5.37 / 5.52 dB
LMR-240	0.243	0.250	0.73 / 0.75 dB	1.82 / 1.88 dB	3.65 / 3.75 dB
LMR-400	0.126	0.129	0.38 / 0.39 dB	0.94 / 0.97 dB	1.88 / 1.94 dB

Read the RG-58 row twice. A 15 m RG-58 patch between a reader and a dock-door antenna spends 10.4 dB — more than 90% of the power, and a range factor of about 0.30. The same run in LMR-400 costs under 2 dB. Cable grade is the single most valuable line to specify on a purchase order for fixed warehouse RFID installations, and naming it there is what settles the question before the van is loaded.

Connectors, adapters and hubs

Beyond the cable itself, budget the interfaces. The three figures below are our own bench measurements at these frequencies, and we quote them as ours: a correctly crimped quality N or RP-TNC connection costs roughly 0.05–0.1 dB, so a normal two-ended jumper is about 0.2 dB; a screw-on adapter to change connector family is worth budgeting about a third of a decibel; and an antenna hub or multiplexer is worth about 1 dB in the budget, then confirmed against its own datasheet, since a four-way hub is an active loss element in every path. Adapters multiply quickly — a reader with RP-TNC ports feeding an antenna with an N pigtail through two adapters has spent the better part of a decibel before the cable starts.

Short beats expensive

Loss is per metre, so halving the run halves the loss regardless of cable grade. A 3 m LMR-240 jumper (0.73 dB) beats a 15 m LMR-400 run (1.88 dB) by more than a decibel, at a fraction of the cost and with a far easier bend radius. Where the reader can be mounted near the antennas — a portal frame, a parking lane gantry, a conveyor upright — move the reader rather than upgrading the cable.

The reverse calculation buyers need

Forward arithmetic answers “what am I radiating?”. The question that actually arrives in a bid is the reverse: given a ceiling and an antenna, what is the highest port power I am allowed to set? Rearranging the chain:

Max port power (dBm) = ceiling as EIRP (dBm) + total feeder loss (dB) − antenna gain (dBi)

If your ceiling is quoted in e.r.p., add 2.15 dB first to get it into EIRP. So for a 2 W e.r.p. limit, start from 35.16 dBm; for 4 W e.r.p., start from 38.17 dBm.

Worked as a lookup across common antenna gains and three feeder-loss cases:

Antenna gain	2 W e.r.p. — 1.0 dB feeder	2 W e.r.p. — 2.0 dB feeder	2 W e.r.p. — 3.5 dB feeder	4 W e.r.p. — 2.0 dB feeder
3 dBi	33.2 dBm	34.2 dBm	35.7 dBm	37.2 dBm
6 dBi	30.2 dBm	31.2 dBm	32.7 dBm	34.2 dBm
8.5 dBi	27.7 dBm	28.7 dBm	30.2 dBm	31.7 dBm
9 dBi	27.2 dBm	28.2 dBm	29.7 dBm	31.2 dBm
12 dBi	24.2 dBm	25.2 dBm	26.7 dBm	28.2 dBm

Three things fall straight out of that table. Higher gain lowers your permitted port power one-for-one. Feeder loss raises it one-for-one, because loss between the port and the antenna is loss the regulator never sees. And in the low-gain, low-loss corner the ceiling leaves headroom to spare — at 3 dBi and 1 dB of feeder, a 2 W e.r.p. limit permits 33.2 dBm, which is above what most readers will produce. There the reader sets the working ceiling and the rule sits above it.

When the arithmetic lands near the reader’s floor

Fixed readers typically bottom out around 10 dBm, so this is rare, but tightly specified installations do get close — a 12 dBi antenna on a very short jumper under a low ceiling can push the permitted setting into the low twenties while the application only ever needed a metre of range. When the arithmetic gets tight, the fix is upstream of the power menu: choose a lower-gain, wider-beam antenna and recover the coverage you actually wanted, or accept the longer feeder run you were trying to avoid. Both put you back in the middle of the reader’s usable range, which is also where its power control is best behaved.

Recording it so it survives commissioning

A calculated port setting holds its value when the next engineer on site can see the reasoning behind it. Put the permitted port power for each antenna port in the commissioning record next to the antenna model, cable type and run length that justify it, and set it in the reader’s persistent start-up configuration rather than in a session command, so a power cycle restores the compliant value. Where the reader software allows a per-region profile to be locked, use it — the region setting is what constrains the power list the operator can even see.

Worked four ways: one antenna, four ceilings

Take one physical install and ship it to four markets. Antenna: an 8.5 dBic circularly polarised panel — the Laird S8658WPR class, 65° azimuth beamwidth. Feeder: 7.5 m of LMR-240 plus three connector interfaces at 0.1 dB each. That gives 2.12 dB of total loss in the lower band and 2.18 dB in the upper band.

Market and band	Ceiling, and how it is written	Feeder loss	Port setting	Resulting EIRP	Resulting e.r.p.
India, 865–868 MHz	2 W e.r.p. (G.S.R. 853(E), 10 Dec 2021)	2.12 dB	28.75 dBm	35.13 dBm	32.98 dBm = 1.98 W
EU lower band, 865–868 MHz	2 W e.r.p. (EN 302 208)	2.12 dB	28.75 dBm	35.13 dBm	32.98 dBm = 1.98 W
EU upper band, 915–921 MHz	4 W e.r.p. (EN 302 208)	2.18 dB	31.75 dBm	38.07 dBm	35.92 dBm = 3.91 W
United States, 902–928 MHz	1 W conducted, reduced above 6 dBi (47 CFR 15.247)	2.18 dB	27.5 dBm	33.82 dBm = 2.41 W	31.67 dBm

Same hardware, same mounting, four different numbers in the power field — a 4.25 dB spread, which is a range ratio of about 1.6× between the strongest and weakest configuration. This is why a reader has a region setting and why the region setting matters more than the model number.

The India row in brief

India’s licence-free UHF allocation is 865–868 MHz under G.S.R. 853(E) of 10 December 2021, made expressly in supersession of the 2005 rules that covered 865–867 MHz, and interrogators are permitted 2 W e.r.p. on channels of 200 kHz or less with EN 302 208 as the reference standard. India therefore mirrors the ETSI lower band, which is why one 865–868 MHz build serves both markets and why the India and EU lower-band rows of the table above carry identical numbers.

Why the FCC row behaves differently

The other three ceilings are radiated limits, so feeder loss buys you port power. The FCC one works on a different quantity. 47 CFR 15.247 caps conducted output at 1 W and states that this limit assumes antennas of 6 dBi or less; above that, conducted power is reduced by the amount in dB by which the gain exceeds 6 dBi. With an 8.5 dBi antenna that is 30 − 2.5 = 27.5 dBm, and under a strict reading the reduction is taken on conducted power, so the port setting stands at 27.5 dBm regardless of feeder length. The practical consequence is that the same install radiates 2.41 W EIRP in the US against a nominal 4 W headline, and that is entirely normal.

The low-power case, and a different way to write a rule

Different device categories in the same band carry different ceilings. Under the same Indian rules, tracking, tracing and data-acquisition devices in 865–868 MHz sit at 500 mW e.r.p. with adaptive power control and duty-cycle limits — 29.14 dBm EIRP, which on the install above works out to a 22.8 dBm port setting. Which table your equipment falls under changes the answer by 6 dB before any hardware choice is made.

Japan writes the rule in a third form entirely. ARIB STD-T107, covering 920 MHz-band RFID for specified low-power radio stations, limits interrogator antenna power to 250 mW with antenna gain of 3 dBi or less, and permits antenna power up to 500 mW where the equipment is in a case that cannot easily be opened and EIRP stays below 27 dBm. That category is defined around the antenna as well as the power figure, so a Japan-bound build is specified antenna-first, starting from a 3 dBi class part and working the port power to suit. Always read which quantity a market’s limit constrains — conducted power, radiated power, or the antenna itself — before converting it.

When the power source sets the ceiling before the regulator does

Regulatory arithmetic sets the ceiling; the power source decides how much of that ceiling the reader can reach. Fixed readers are increasingly PoE-powered, and published datasheets cap transmit power by PoE class in black and white.

The Impinj R700 series datasheet is explicit: the R700 offers 10–30 dBm on PoE (802.3af) against 10–33 dBm on PoE+ (802.3at), and the reader “will limit the maximum transmit power to 30 dBm while configured to use PoE power.” Its R720 sibling is capped at 10–20 dBm on PoE and reaches the full regional range on PoE+. Power negotiation is by LLDP. Those are published figures for one widely deployed family, and the pattern is general.

Shortfall	Typical cause	Range factor
1.5 dB	31.5 dBm ETSI-lower maximum reached on PoE+ but only 30 dBm available on PoE	×0.84 (1.19× to recover)
3 dB	33 dBm on PoE+ against 30 dBm on PoE	×0.71 (1.41× to recover)
13 dB	33 dBm on PoE+ against a 20 dBm PoE cap	×0.22 (4.47× to recover)

How to catch a fallback to the lower class

The failure mode that costs projects a week is a port that works. When LLDP negotiation falls back, the reader boots, links, answers on the network and reads tags, while presenting the lower class’s power list — silently. The tell is that the power values you were told to set are absent from the menu. Long or marginal Cat 5e runs, older switches without LLDP support, and mid-span injectors that pass class negotiation imperfectly all produce it. Verify the negotiated class on the reader itself during commissioning and record it, in preference to the switch port’s configured class.

Budget the switch, not just the ports

IEEE 802.3at delivers 30 W at the PSE port and guarantees 25.5 W at the powered device. A twelve-reader pallet tracking deployment on PoE+ therefore wants headroom for roughly 360 W of PoE budget before anything else on that switch is counted. Switch vendors publish a total budget alongside the per-port maxima, and the two are different numbers: Ubiquiti’s UniFi Pro 24 PoE, for example, publishes 400 W of total PoE availability across 16 PoE+ ports rated 30 W and 8 PoE++ ports rated 60 W. Where demand exceeds the total, a switch resolves it by allocating to a lower class. Build the worksheet before you buy: readers × class wattage × port count, against the switch’s published total PoE budget, with every other powered device on the same switch included.

Choosing antenna gain when the ceiling is fixed by law

When the radiated ceiling is fixed, gain becomes a way to choose the shape of the power you are already allowed. Every decibel of gain you add is a decibel you subtract from the port, so the two cancel at the ceiling. Beamwidth is what changes.

One manufacturer’s catalogue makes the trade visible in a single column. Across Laird’s UHF panels: 5.5 dBic comes with a 100° 3 dB beamwidth, 6 dBic with 80°, 8.5 dBic and 9 dBic with 70°, and the wideband 8.5 dBic part with 65°. Gain up, beam narrow — always.

Footprint from beamwidth

For a downward-facing antenna the illuminated width at floor level is approximately 2 × mounting height × tan(half beamwidth).

Gain / beamwidth	2.5 m mounting	4 m mounting	6 m mounting
5.5 dBic / 100°	5.96 m	9.53 m	14.30 m
6 dBic / 80°	4.20 m	6.71 m	10.07 m
9 dBic / 70°	3.50 m	5.60 m	8.40 m
8.5 dBic / 65°	3.19 m	5.10 m	7.64 m

A 3.5 dB gain step from 5.5 to 9 dBic tightens the floor footprint at 4 m from 9.53 m to 5.60 m — a 41% reduction in illuminated width. On a wide dock door that is the difference between one antenna and two. Height helps: the same 70° antenna covers 3.50 m at 2.5 m and 8.40 m at 6 m, which is why raising an antenna is often a better answer than widening its beam.

Where gain becomes part of your compliance position

Under e.r.p.-based rules, gain and port power trade one-for-one and the choice stays an engineering one. Under the FCC form, gain above 6 dBi is a condition on your conducted power, so it forms part of your compliance position rather than a tuning step. And under a rule written like Japan’s, the antenna gain figure itself is part of the equipment category. Fix the destination market before you fix the antenna.

Polarisation, and the 3 dB it costs

Where tag orientation is known and controlled — cartons on a conveyor, plates on vehicles, files in a drawer — a linear antenna matched to the tag delivers the full gain into it. Where orientation is unknown, circular polarisation reads a tag at any rotation, at the cost of roughly 3 dB into any single linear tag. That 3 dB is a 0.71× range factor, and it is almost always the right trade: reading a tag at every rotation is worth more on a real site than reading it strongly at one rotation only.

From a theoretical number to a site number

Everything above computes what leaves the antenna. What comes back from a tag depends on a further ladder of derating that a site survey pins down:

Feeder — cable, connectors, adapters, hubs, as tabulated above.
Polarisation mismatch — about 3 dB for circular against linear.
Tag orientation — a tag presented edge-on to the field can lose far more than the polarisation term alone.
Substrate — a general-purpose inlay on metal or against liquid detunes badly. On-metal tags exist precisely because this term is large.
Environment — racking, mesh, wet cardboard, forklift bodies and other readers all take their cut, and multipath gives as often as it takes.

A power budget therefore lands as a range, confirmed by site survey. We quote the band and then walk the site, so the number you receive already accounts for tags, substrate and mounting.

What to measure, and on what sheet

Commissioning should log, per antenna port: negotiated PoE class and available power list; port power actually set; antenna model and gain; cable type and measured run length; connector and adapter count; measured RSSI on a reference tag at three fixed marked positions; and read rate over a fixed interval on a known tag population. Marked positions matter more than absolute values — the number that tells you something six months later is the change at the same spot, not the reading itself. Keep the sheet with the site file, so the next engineer diagnosing a drop-off compares against a baseline.

A one-page worksheet to attach to your purchase order

Make the arithmetic part of the bid, and every quotation becomes comparable. Attach a single page with these fields.

Inputs you supply

Destination market and band — the country, not the region.
The ceiling as written in the applicable rule, and whether it is e.r.p., EIRP or conducted.
Antenna model, gain figure and whether it is quoted in dBi, dBd or dBic.
Cable type and run length per port, in metres.
Connector, adapter and hub count per path.
Reader power source: PoE class, PoE+, or external DC.

Outputs the supplier returns

Total feeder loss per port, in dB, with the cable manufacturer’s datasheet figure cited.
Maximum permitted port power, in dBm, to the reader’s power resolution.
Calculated EIRP and e.r.p. at that setting.
Margin remaining below the ceiling, in dB.
Whether the reader can actually produce that setting on the specified power source.
The reader region setting to be applied, by name.

Six numbers, and the bid stops being a comparison of headline power figures and becomes a comparison of installations. Because we build both the reader hardware and the software that configures it, we ship FCC-band 902–928 MHz and ETSI-band 865–868 MHz variants configured per order, with the region profile and per-port power set before the unit leaves — so the sheet that comes back with the quotation is the sheet the commissioning engineer works from.

Frequently asked questions

How do I convert 2 W e.r.p. to dBm?

2 W is 33.0 dBm, so the ceiling is 33.0 dBm e.r.p. To use it in an EIRP calculation, add 2.15 dB: 2 W e.r.p. = 35.16 dBm EIRP. Then subtract your antenna gain in dBi and add back your total feeder loss in dB to get the maximum permitted port power. For an 8.5 dBi antenna on a feeder losing 2.0 dB, that is 35.16 + 2.0 − 8.5 = 28.7 dBm at the port.

What is the difference between ERP and EIRP?

They describe the same radiated field against two different reference antennas. EIRP is referenced to an isotropic radiator; e.r.p. is referenced to a half-wave dipole, which itself has 2.15 dBi of gain. So EIRP is always 2.15 dB numerically larger than e.r.p. for the same transmission. European and Indian short-range-device rules are written in e.r.p.; the FCC works in conducted power with an antenna-gain condition. Converting between them wrongly costs you 2.15 dB in one direction or puts you 2.15 dB over the limit in the other.

What transmit power is allowed for a UHF RFID reader in Europe and in the United States?

Under EN 302 208 the European lower band, 865–868 MHz, permits up to 2 W e.r.p., and the upper band, 915–921 MHz, permits up to 4 W e.r.p. — and adoption of the upper band is set country by country, so confirm the destination country before specifying. In the United States, 47 CFR 15.247 caps conducted output power at 1 W for 902–928 MHz digitally modulated systems, on the assumption of an antenna of 6 dBi or less; above 6 dBi the conducted power must be reduced by the amount in dB that the gain exceeds 6 dBi. That combination is what produces the familiar 4 W EIRP figure.

How much read range does coax cable loss cost me?

It depends entirely on the cable. At 915 MHz, a 15 m run costs about 1.94 dB in LMR-400, 3.75 dB in LMR-240, 5.52 dB in LMR-195, and about 10.4 dB in RG-58 type cable such as Belden 8259. Since 6 dB halves free-space distance, that RG-58 run leaves theoretical range at roughly 38% of what the same reader achieves on LMR-400. Under an e.r.p.-based rule you can raise the port power to compensate up to the reader’s own maximum, and the shorter or better run is nearly always cheaper than the reader upgrade.

Does a higher-gain antenna always give more range?

It gives more range while there is still headroom below the radiated ceiling. Once the ceiling binds, every decibel of gain you add is a decibel you remove from the port to stay at the same e.r.p., so the two cancel and what changes is beam shape: gain and beamwidth move in opposite directions. Across one manufacturer’s UHF panel range, 5.5 dBic comes with a 100° beamwidth and 9 dBic with 70°. At 4 m mounting height that tightens the floor footprint from about 9.5 m to about 5.6 m. Higher gain reaches further along the boresight and concentrates the beam — the right choice for a lane, while a wide doorway is better served by a wider-beam panel.

Why is my reader limited to 30 dBm?

Most often the PoE class rather than the regulator. Published datasheets cap transmit power by power source: the Impinj R700, for example, offers 10–30 dBm on PoE (802.3af) against 10–33 dBm on PoE+ (802.3at), and its R720 sibling is capped at 10–20 dBm on PoE. Negotiation is by LLDP, and when it falls back the reader still boots and reads while presenting the lower class’s power list, silently. Check the negotiated class on the reader itself, verify the switch’s total PoE budget covers every powered device on it, and then look at the region setting.

How do I calculate the maximum port power for a given antenna?

Use: maximum port power (dBm) = ceiling expressed as EIRP (dBm) + total feeder loss (dB) − antenna gain (dBi). Convert an e.r.p. ceiling to EIRP first by adding 2.15 dB. Total feeder loss is cable loss per metre from the manufacturer’s datasheet times run length, plus the interface terms — on our bench, roughly 0.05–0.1 dB per quality connector, about a third of a decibel per adapter, and about 1 dB for an antenna hub. Round the answer down to the reader’s power resolution. Under the FCC form the calculation works on a different quantity: conducted power itself is capped at 1 W and reduced dB-for-dB above 6 dBi of antenna gain.