dBi or dBd? Why I Prefer the Dipole Reference

My starting question was simple: why do Comet and Diamond keep using dBi when dBd is such a useful reference for the antennas we actually compare? The technical answer begins with the manufacturers’ actual specifications: professional antennas use both conventions. My preference for dBd is a preference for a meaningful comparison baseline; the suffix alone cannot establish the quality of an antenna or its measurement.

What I want a gain number to tell me

A lossless half-wave dipole already concentrates radiation broadside to its wire. Calling its maximum gain 0 dBd makes that familiar performance the starting point. An antenna specified at 3 dBd offers twice the radiation intensity in its stated direction for the same accepted power, relative to that ideal dipole reference. At 6 dBd, the factor is approximately four.

That is why I like dBd for practical comparisons. A specification of 2.15 dBi describes the reference dipole's own maximum gain; expressed as 0 dBd, its relationship to the baseline is immediately visible. Neither label changes the field, efficiency or pattern.

The conventional spelling is dBd: decibels relative to a dipole. In dBi, the final letter identifies an isotropic reference. A bare dB identifies a ratio but leaves the reference unstated unless the accompanying specification supplies it.

The manufacturers do not all use the same convention

Here are concrete examples from manufacturers' own specifications. The converted column applies the conventional 2.15 dB offset to the published number; it is not an independent measurement.

Diamond's X-series page specifically explains that its mixed dB/dBi labels both denote isotropic gain. That clarification applies to those specifications; it does not make every unqualified dB claim from every manufacturer a dBi claim. Amphenol Procom's dual labeling is a useful practice I would like to see more often.

These examples disprove the idea that professional antennas universally use dBd. They do not establish why a particular company chose its convention. dBi has a direct role in link budgets and effective-aperture calculations, and its use is technically legitimate. Its numerical value is 2.15 higher for the same gain, which can affect a casual comparison, but that arithmetic does not prove a marketing motive.

The extra decimal places in the converted column come from subtraction. They do not improve the accuracy of the original specifications. Nor does the table establish that the manufacturers used identical test environments, port definitions or uncertainty budgets.

Gain begins with power per unit solid angle

Let U(θ, φ) be far-field radiation intensity in watts per steradian. It is r² times the outward, time-averaged power density in watts per square meter, once the angular far-field pattern is established. Total radiated power is the integral over the sphere:

Directivity D compares radiation in one direction with the spherical average. Gain also charges the antenna for dissipative loss. If P_acc is the net power accepted at the antenna port and η_rad = P_rad/P_acc, then:

These equations use linear power ratios for D and G. For a passive antenna, efficiency cannot exceed one, so gain cannot exceed directivity. Total gain includes both orthogonal polarization components; a specified partial or co-polar gain includes only the stated component. Compare like quantities.

ITU-R P.341-7, Annex 1 distinguishes isotropic gain from gain relative to a half-wave dipole isolated in space. The dipole is oriented with its equatorial plane containing the comparison direction. This is a fixed reference to its broadside maximum, not division by the dipole's directional null.

Where the 2.15 dB comes from

Take the ideal thin, lossless, half-wave dipole along the z axis, with sinusoidal current and no ground or nearby objects. Its normalized far-field electric-field pattern is:

This is the half-wave case of the finite-dipole field derived in MIT's antenna notes, equations 9.37–9.39. Its limits at θ = 0 and π are zero. Integrating its power pattern gives:

The familiar offset is this result rounded to 2.15 dB. Because the reference is lossless, its gain equals its directivity. The conversion for the same antenna gain is therefore:

An electrically short dipole has a different limiting pattern and a maximum directivity of 1.5, or about 1.76 dBi. It is not the dBd reference. A physically constructed resonant dipole also has finite diameter, loss, a feed arrangement and a length adjusted for resonance. The conventional reference does not acquire those installation details merely because both antennas are called dipoles.

Gain is a power ratio, hence 10 log. At equal distance, accepted power, medium impedance and polarization, the corresponding field-amplitude ratio can be written with 20 log. Mixing a power ratio with 20 log doubles the claimed decibel advantage.

The suffix does not tell you which losses were included

For a single-port antenna referenced to a real impedance Z₀, such as 50 Ω, let Γ be its input reflection coefficient. Incident power and accepted power differ by the mismatch factor 1 − |Γ|². Realized gain includes that factor:

Rohde & Schwarz's OTA measurement white paper distinguishes directivity, gain and realized gain on this basis. A good impedance match does not establish good radiation efficiency.

Consider a calculated example with peak directivity 8.00 dBi, radiation efficiency 70%, and VSWR 2:1 at the same frequency and port:

This example loses approximately 1.55 dB through dissipation and another 0.51 dB through mismatch. Changing the reference subtracts 2.15 from either gain number. It cannot turn directivity into gain or gain into realized gain. These values illustrate the definitions; they are not test results for any product in the table.

The reference plane matters just as much. A specification at the antenna connector normally excludes the user's external feedline. If a cable or feed network is included in the measured assembly, its losses belong inside that assembly's gain definition. With mismatch, simply subtracting a nominal matched cable loss may miss the interactions between the cable, antenna and source.

How mixed references distort a purchase comparison

Suppose two otherwise comparable specifications state 9 dBi and 6 dBd. Subtracting the printed numbers suggests a 3 dB advantage. Converting first gives 9 dBi versus 8.15 dBi: a difference of just 0.85 dB.

Under equal conditions, that is roughly 22% more power density in the specified direction. The comparison remains conditional on compatible quantities and measurement conditions. If one figure is lossless simulated directivity and the other is measured realized gain, changing the suffix does not fix the comparison.

ERP and EIRP carry the same reference distinction

For a stated direction, equivalent isotropically radiated power uses isotropic gain. Effective radiated power uses the half-wave dipole reference. With P_acc in watts at the antenna port:

For 50 W accepted power and 6 dBd gain, ERP is approximately 199 W and EIRP approximately 327 W. Both describe the same directional radiation. The antenna has not created additional watts: these are the powers the specified lossless reference radiators would require to produce the same far-field intensity.

If you instead start from incident power at a defined real-impedance port, use the compatible realized gain. Starting with accepted power and realized gain would count mismatch twice. Starting from transmitter output additionally requires the intervening feed system to be accounted for.

Why dBi fits naturally into link equations

For reciprocal antennas in free space, in each other's far field, with matched polarization, the Friis relation is:

Here P_r,av is the power available from the receiving antenna to a conjugately matched load, P_t,acc is accepted transmit power, R is separation and λ is wavelength. Both linear gains are relative to isotropic in the relevant directions. A_e is effective receiving aperture for the matched polarization. The assumptions exclude ground-reflected interference and other multipath.

In decibel form, using dBm for both powers and meters for both R and λ:

Using dBd for both gain terms requires adding 4.30 dB overall. Using dBd for one gain term requires adding 2.15 dB. In linear calculations, convert with G = 10^{(GdBd + 2.15)/10}. Putting 10^GdBd/10 straight into the aperture equation underestimates aperture by approximately 1.64.

This mathematical convenience is a sound reason for engineers to use dBi. I can prefer dBd when communicating an antenna comparison while using isotropic gain inside a calculation.

A dipole in your garden is not the reference definition

The 2.15 dB offset belongs to the ideal reference. A real dipole above ground has interference between direct and reflected fields. Height, orientation, soil and nearby structures change its lobes, nulls and losses. A measurement against that installed dipole establishes a difference between two installed systems.

If the reference antenna's installed gain is unknown, calling that measured difference dBd silently assumes the reference has exactly the ideal free-space dipole's gain in the measurement direction. That assumption can be seriously wrong. Reporting “2 dB above this reference dipole in this setup” preserves the actual result.

A useful related example is an ideal lossless quarter-wave monopole above an infinite perfectly conducting plane. Image theory gives approximately 5.15 dBi maximum directivity, equivalent to 3.00 dBd, because radiation is confined to the upper hemisphere. That ideal result does not automatically describe a whip on a finite vehicle roof, a few radials or lossy earth.

The conversion between dBi and dBd remains 2.15 dB for a properly defined installed gain quantity. What changes with installation is the antenna's gain and pattern, not the reference offset. There is no universal ground-gain bonus for practical antennas.

The useful direction matters more than the peak

A vertical collinear can be approximately omnidirectional in azimuth while concentrating its elevation pattern into a narrow region. A high peak can coexist with weak coverage above or below it. Mast coupling, mounting tilt and frequency can move the useful lobe relative to the required path.

Comet itself discusses this tradeoff on its GP-9M page: a narrow elevation pattern can be unsuitable for some valley-to-mountaintop paths, while a wider pattern may serve them better. That is an application question even after every gain figure has been converted correctly.

A plot normalized to its own peak at 0 dB shows relative pattern shape. It cannot establish absolute gain without a calibrated peak value. Two normalized plots can look identical while their efficiencies and realized gains differ substantially.

On receive, more signal gain also does not automatically mean the same improvement in signal-to-noise ratio. The antenna weights noise from different directions through its whole pattern; receiver noise and losses contribute too. The gain suffix supplies no information about that noise environment.

You can measure dBi without building an isotropic antenna

A mathematical reference does not require a physical isotropic radiator on the test range. A calibrated antenna with known gain can transfer that reference through a substitution measurement. Absolute methods can also solve for the gains of three antennas from their pairwise transmissions; NIST describes this three-antenna calibration approach.

For an idealized reciprocal, matched, polarization-aligned far-field measurement, remove the known free-space spreading term and express each pair's transfer in decibels. Calling the corrected pair sums C gives:

This algebra shows how an absolute reference can be recovered without assuming one antenna's gain. A real calibration needs corrections for mismatch, separation, alignment, reflections and instrumentation; the equations alone are not a complete range procedure.

For a receive substitution into the same matched real-impedance receiver, corrected received-power differences correspond to differences in realized gain. Recover gain by accounting for the antennas' mismatch factors separately. If the receiver is not matched, the complete power-transfer mismatch must be included. A raw S₂₁ trace or an S-meter reading does not supply those corrections by itself.

Keep position, polarization, illumination and receiver linearity controlled, account for cable loss, and characterize unwanted feedline common-mode current. Otherwise the feedline may become part of the receiving structure being compared. A dBd label does not prove that a physical reference dipole was used, any more than a dBi label proves that the result was simulated.

Precision and uncertainty survive the conversion

Subtracting a conventional reference constant does not reduce measurement uncertainty. A gain stated as 8.0 dBi with a specified uncertainty retains that uncertainty when expressed as 5.85 dBd. The additional displayed decimal comes from arithmetic.

For a difference between two gain estimates in decibels, uncertainty also depends on correlation:

Two independent standard uncertainties of 0.6 dB give approximately 0.85 dB standard uncertainty in their difference. That is already the entire apparent advantage in the 9 dBi versus 6 dBd example. These illustrative uncertainty values must not be assigned to a manufacturer without evidence. Shared calibration errors can partly cancel in a controlled comparison; unrelated specifications may have no common uncertainty basis at all.

How I want antenna gain to be stated

I prefer dBd first for practical dipole-based comparisons, with dBi alongside it for calculations. The accompanying specification should state:

• The frequency or frequency range, and whether the value is typical, minimum or peak.
• Whether the quantity is directivity, gain or realized gain.
• The port/reference plane and which feed or matching losses are included.
• The direction, polarization component and relevant radiation patterns.
• The environment: free space, specified ground, mast, vehicle or another defined installation.
• Whether the result is calculated or measured, with method and uncertainty where available.

My practical rule is straightforward: convert every claim to a common reference before comparing it, then compare the same physical quantity in the directions the link needs. dBd makes the dipole baseline visible. Paired with dBi and a complete specification, it gives readers both an intuitive comparison and the numbers needed for a sound link budget.

Questions, antenna-factor records or height trials to share? Contact RF.Guru.