100nF

Free-space path loss formula and RF link budget

Free-space path loss by ITU-R P.525: the 32.45 dB constant, why it rises with frequency, a 2.4 GHz link budget to margin and range, and where free space fails.

Free-space path loss between two isotropic antennas is Lbf = 20 log(4πd/λ), with d and λ in the same unit. At 2.4 GHz over 100 m that is 80.05 dB, and it rises 6.02 dB every time either the distance or the frequency doubles. With f in MHz and d in km the formula is 32.45 + 20 log f + 20 log d; the Recommendation that defines it, ITU-R P.525-5, prints the constant rounded to 32.4. A link budget puts that loss between the transmitter’s EIRP and the receive antenna’s gain, and the margin is what is left above the receiver’s sensitivity. For a 10 dBm radio with 2 dBi antennas and half a decibel of cable at each end, 100 m at 2.44 GHz delivers −67.2 dBm, 22.8 dB above a −90 dBm sensitivity, and free space would carry the link to 1.38 km.

The rest of this article is where those numbers come from, which of them P.525 states and which are derived from its definitions, and the conditions under which a free-space figure describes a real link at all. The RF link budget calculator runs the same equations; every derived number below comes from its module.

A link budget drawn as a level diagram in dBm for a 2.44 GHz link over 100 metres. The signal starts at 10 dBm, loses 0.5 dB in the transmit cable, gains 2 dBi at the transmit antenna to an EIRP of 11.5 dBm, then falls 80.2 dB across the free-space path to minus 68.7 dBm, gains 2 dBi at the receive antenna and loses 0.5 dB in the receive cable, arriving at minus 67.2 dBm. A line at the minus 90 dBm receiver sensitivity sits 22.8 dB below the received power: that gap is the link margin.
Fig 1 — The calculator's default link (illustrative): 10 dBm out, 0.5 dB of cable and 2 dBi of antenna at each end, 100 m at 2.44 GHz. The free-space term is 80.2 dB; every other term together moves the level by 3 dB. The margin is what is left above the sensitivity.

What free-space path loss is

P.525 starts from a definition that excludes everything a real path contains:

For the purposes of radiocommunication, free space is defined as a perfect vacuum which may be considered of infinite extent in all directions, so that free space propagation is the propagation of a radio wave radiating in free space.

That is on page 1, and the Recommendation states its reason for bothering just above it: “free-space propagation is a fundamental reference for radio-engineering”. Nothing in a vacuum absorbs energy. The loss is geometry, and P.525 builds it from two factors.

The first is spreading. A transmitter radiating p watts from an isotropic antenna spreads them over a sphere, so at distance d the power flux-density is p/(4πd²) watts per square metre (Eq 3, p. 2). The second is collection. A receiving isotropic antenna intercepts that flux over its effective aperture, which P.525 gives as λ²/4π, so the power at its output is the product (Eq 4, p. 3):

pr=s×a=p4πd2×λ24πp_r = s \times a = \frac{p}{4\pi d^2} \times \frac{\lambda^2}{4\pi}

The free-space basic transmission loss, symbol Lbf (or Abf), is the ratio of p to pr in decibels, P.525’s Eq 5:

Lbf=−10log⁡10(14πd2×λ24π)=20log⁡10(4πdλ) dBL_{bf} = -10\log_{10}\left(\frac{1}{4\pi d^2}\times\frac{\lambda^2}{4\pi}\right) = 20\log_{10}\left(\frac{4\pi d}{\lambda}\right)\ \text{dB}

The where-list under it is short and matters: “d and λ are expressed in the same unit” (p. 3). The wavelength is c/f, and the calculator uses the exact SI value, which the BIPM SI Brochure states as “the exact relation c = 299 792 458 m s⁻¹” (p. 127). At 2.44 GHz, λ = 12.29 cm.

Why path loss increases with frequency

Look at the two factors again. The spreading term 1/(4πd²) has no frequency in it. A wave at 433 MHz and a wave at 12 GHz thin out over the sphere at exactly the same rate. All of the frequency dependence sits in the second factor, the aperture λ²/4π of the receiving antenna, which shrinks as the square of the wavelength.

Free-space path loss at a fixed 100 metres plotted against frequency from 100 MHz to 10 GHz on a logarithmic axis, split into its two factors from ITU-R P.525. The spreading term, 10 log of 4 pi d squared, is a flat 51.0 dB at every frequency. The isotropic receiving aperture term, minus 10 log of lambda squared over 4 pi, rises 20 dB per decade. Their sum, the path loss, rises from 52.4 dB at 100 MHz to 92.4 dB at 10 GHz, so all of the frequency dependence comes from the shrinking aperture of the receiving antenna, not from the space between the antennas.
Fig 2 — Lbf at 100 m split into the two factors of P.525 Eq (5). The spreading of the power over a sphere does not depend on frequency; the effective aperture of a receiving isotropic antenna, λ²/4π, does: 2.1 cm² at 5.8 GHz, 94.9 cm² at 868 MHz.

At 100 m the spreading term is 51.0 dB at every frequency. The aperture term adds 29.2 dB at 2.44 GHz, for the 80.2 dB total, and climbs 20 dB per decade of frequency. An isotropic antenna at 868 MHz has an effective aperture of 94.9 cm²; at 2.44 GHz it is 12.0 cm², and at 5.8 GHz 2.1 cm². The higher frequency does not fade faster in the space between the antennas. A receiving antenna of the same gain simply catches less of it.

That is why the statement “path loss rises with frequency” carries a condition: it holds for antennas whose gain stays fixed as the frequency changes, which is what Lbf assumes, since it is defined between isotropic antennas at both ends. Whether a particular antenna keeps its gain across a band is a property of that antenna, to be read from its datasheet at the link frequency. P.525 states the aperture of the isotropic antenna and nothing else; it gives no relation between a real antenna’s gain and its physical size, and the calculator does not invent one.

The free-space path loss formula in dB, and its constant

“Equation (5) can also be written using the frequency instead of the wavelength”, and P.525 does so as Eq 6 (p. 3):

Lbf=32.4+20log⁡10f+20log⁡10d dB,f in MHz, d in kmL_{bf} = 32.4 + 20\log_{10} f + 20\log_{10} d\ \text{dB}, \qquad f\ \text{in MHz},\ d\ \text{in km}

The constant is 20 log(4π × 10⁹/c). With the exact c it is 32.448 dB, so the printed form sits 0.048 dB below Eq 5 at every frequency and distance. That is far below anything a link budget can resolve, but it is the reason two calculators can disagree in the second decimal place: at 2.4 GHz and 100 m, Eq 5 gives 80.05 dB and Eq 6 as printed gives 80.00 dB.

The constant belongs to the units, and changing the units changes it:

f ind inConstant, from Eq 5 with exact c
MHzkm32.45 dB (P.525 prints 32.4)
GHzkm92.45 dB
MHzm−27.55 dB
GHzm32.45 dB

The last row is a trap. GHz and metres give the same constant as MHz and kilometres, because both pairs multiply to the same power of ten. A formula quoted with “32.45” and no units is correct for two unit systems and 60 dB off in the other two.

P.525’s own conversion formulae in section 4 switch to f in GHz, and the Recommendation ends with a line that is easy to misread:

Note that equations (8) and (10) can be used to derive equation (6).

Substituting Eq 8 into Eq 10 gives 20 log d + 20 log f + 167.2 − 74.8, which is 92.4 + 20 log f + 20 log d with f in GHz. It becomes Eq 6’s 32.4 only after f is converted to MHz, which takes 60 dB off the constant. The two sections of the same Recommendation use different frequency units, and the note does not say so.

Free-space path loss at common frequencies and distances

Eq 5 for three bands short-range radios use. At exactly 2.4 GHz, 100 m is 80.05 dB; the table’s 2.44 GHz row is the calculator’s default frequency.

Frequency10 m100 m1 km10 km
868 MHz51.2 dB71.2 dB91.2 dB111.2 dB
2.44 GHz60.2 dB80.2 dB100.2 dB120.2 dB
5.8 GHz67.7 dB87.7 dB107.7 dB127.7 dB

Every row climbs 20 dB per decade of distance. Moving the same link from 868 MHz to 2.44 GHz costs 20 log(2440/868) = 8.98 dB at any distance.

A link budget follows the signal from the transmitter’s output connector to the receiver’s input and adds what happens to it, in decibels. P.525 supplies the middle term and a definition for the first. Its e.i.r.p. is the “equivalent isotropically radiated power (e.i.r.p.) of the transmitter in the direction of the point in question” (p. 2): the power an isotropic antenna would have to radiate to put the same flux at the receiver. P.525 prints no expression for it. The sum below is what a gain stated in dBi means, and the calculator labels it as derived:

EIRP=Ptx−Ltx+Gtx\text{EIRP} = P_{tx} - L_{tx} + G_{tx}

The receive side adds its gain on the same dBi basis to P.525’s loss between isotropic antennas, and subtracts the loss between the antenna and the receiver:

Prx=EIRP−Lbf+Grx−Lrx−LotherP_{rx} = \text{EIRP} - L_{bf} + G_{rx} - L_{rx} - L_{other}

The margin is Prx less the sensitivity. Lother is where every allowance goes that free space does not include: a wall, foliage, polarisation mismatch, a fade reserve. P.525 gives no values for any of them.

Each term has a usual source. Ptx is the radio’s output power from its datasheet, at the connector. Ltx and Lrx are cable, connector and filter losses at the link frequency; the coax impedance calculator shows how a line’s geometry sets its impedance and conductor loss, and the cable’s datasheet gives the attenuation per metre. Gtx and Grx are the antennas’ gains in dBi in the direction of the other end, not their peak gains, because the EIRP is defined in the direction of the point in question.

P.525 prints no numeric example, so this one is illustrative: the calculator’s defaults, a short-range link that does not come from a source.

λ        299 792 458 / 2.44 × 10⁹                = 12.29 cm
L_bf     20 log(4π × 100 / 0.1229)                = 80.20 dB
         Eq 6: 32.4 + 20 log 2440 + 20 log 0.1    = 80.15 dB
EIRP     10 dBm − 0.5 dB + 2 dBi                  = 11.5 dBm  (−18.5 dBW, 14.1 mW)
P_rx     11.5 − 80.20 + 2 − 0.5                   = −67.20 dBm
margin   −67.20 − (−90)                           = 22.80 dB
range    100 m × 10^(22.80/20)                    = 1.38 km

Fig 1 at the top of the page is this budget as a level diagram. The cables and antennas together move the level by 3 dB; the path takes 80.2 dB. The gap widens with distance, since only the path term grows. Write every term down anyway: the small ones are easy to forget, and each counts at the margin.

The −90 dBm sensitivity is a property of the receiver in a particular mode, not a constant. It is the receiver’s noise floor plus the signal-to-noise ratio its demodulator needs, and the floor is kT0B plus the noise figure. The noise figure calculator puts kT0B at −114.0 dBm in 1 MHz; with an illustrative 6 dB noise figure the floor is −108.0 dBm, and a −90 dBm sensitivity then means the demodulator needs 18.0 dB of SNR. A receiver quoted at a lower data rate in a narrower bandwidth has a lower floor and a better sensitivity, which is why a sensitivity copied from a datasheet has to be the figure for the mode the link will run.

Range for a given receiver sensitivity

The range the calculator reports is Eq 5 inverted: the distance at which Prx falls to the sensitivity, which is where Lbf equals EIRP + Grx − Lrx − Lother − S.

dmax=λ4π 10(EIRP+Grx−Lrx−Lother−S)/20d_{max} = \frac{\lambda}{4\pi}\,10^{(\text{EIRP} + G_{rx} - L_{rx} - L_{other} - S)/20}

For the worked link the loss the budget can absorb is 103 dB, reached at 1.38 km. Because the loss grows as 20 log d, margin converts to distance at a fixed rate whichever term it comes from:

Margin at dFree-space range ÷ d
3 dB1.41
6 dB2.00
10 dB3.16
20 dB10.0

Read the other way, the table is the price of an allowance. Put 10 dB in Lother for walls or fading and the 2.44 GHz link’s free-space range falls from 1.38 km to 437 m, 0.32 of what it was.

Received power in dBm against distance from 1 metre to 10 kilometres on a logarithmic axis, for the same link budget at 868 MHz, 2.44 GHz and 5.8 GHz. Each is a straight line falling 20 dB per decade of distance, offset by the frequency. Where each crosses the minus 90 dBm sensitivity line is its free-space range: 3.88 km at 868 MHz, 1.38 km at 2.44 GHz and 581 m at 5.8 GHz.
Fig 3 — The budget of Fig 1 swept over distance at three frequencies, antenna gains held at 2 dBi. Each line falls 20 dB per decade, 6.02 dB per doubling; the crossing with the sensitivity is the free-space range, and the dot is the worked example, −67.2 dBm at 100 m. With the gains fixed, the lower frequency reaches further.

Fig 3 sweeps the same budget over distance at three frequencies. With the 2 dBi gains held fixed, 868 MHz has 31.8 dB of margin at 100 m and reaches 3.88 km; 5.8 GHz has 15.3 dB and reaches 581 m. The lines are parallel, spaced by 20 log of the frequency ratio, and that spacing is the aperture term of Fig 2.

Field strength and the conversion formulae

For a transmitter serving many receivers, P.525 works in field strength rather than loss. Eq 1 gives the r.m.s. field at d metres from an e.i.r.p. of p watts as e = √(30p)/d in V/m, and Eq 2, which “uses practical units”, restates it as 173 √p/d in mV/m, kW and km (p. 2). Section 4 then gives the decibel forms, with Pt in dB(W), E in dB(µV/m), f in GHz and d in km:

E=Pt−20log⁡10d+74.8,Pr=E−20log⁡10f−167.2,S=E−145.8E = P_t - 20\log_{10} d + 74.8, \qquad P_r = E - 20\log_{10} f - 167.2, \qquad S = E - 145.8

Each printed constant is a rounding of a value that follows from Eq 1, 3 and 4 and the exact c. The calculator’s tests rebuild them: 173.21 for Eq 2, 74.77 for Eq 8, 167.22 for Eq 9, 145.76 for Eq 11, and 103.44 for the radar equation. For the worked link, the 11.5 dBm EIRP gives 6.51 mV/m at 100 m, or 76.27 dB(µV/m); Eq 8 with its rounded constant gives 76.3. The calculator reports the field strength and flux-density at the receiver alongside the budget, from the EIRP, which is the p that Eq 1 takes.

P.525’s scope is one sentence: “This Recommendation provides methods to calculate the attenuation in free space” (p. 1). It is a reference, and its own notes mark the edges.

No ground, no atmosphere, no second path. The vacuum of the definition has no earth under it. P.525 deals with one case of ground, in Note 2 to Eq 1: an antenna at ground level with vertical polarisation radiates only into the upper half-space, and

When the ground is assumed to be plane and perfectly conducting, the power flux-density for a given radiated power is doubled, as compared with an antenna in free space.

On field strength P.525 puts the same effect as 3 dB. That is the ground alone, at its most idealised, and the note says it is already included in Recommendation ITU-R P.368 and in Annex 3 to Recommendation ITU-R P.341. Terrain, a reflected path, rain, foliage and walls are all outside P.525. A link at 10 km with 14 dBm, 2 dBi at each end and an illustrative −120 dBm narrow-band receiver has 25.8 dB of free-space margin at 868 MHz, which the range formula turns into 195 km. No link along the ground behaves like that, and nothing in P.525 says it would.

Far field only. The 1/d behaviour of Eq 1 is not unconditional. P.525’s footnote to its definition quotes the electrotechnical vocabulary:

NOTE – For propagation in free space, the magnitude of each vector of the electromagnetic field in any given direction from the source is proportional to the reciprocal of the distance from the source beyond a distance determined by the size of the source and the wavelength.

P.525 gives no number for that distance. Eq 5 itself reaches 0 dB at d = λ/4π, 9.8 mm at 2.44 GHz, where it plainly no longer describes anything physical; the calculator flags a path shorter than one wavelength.

Matched, co-polarised, pointed. Eq 9 is headed “Available power received through a conjugately matched isotropic receiving antenna for a given field strength” (p. 4); its own where-list drops the word isotropic, but the equation’s constant is the isotropic aperture’s. A mismatch between antenna and feed is a further loss. Note 1 covers polarisation for the field-strength equation only: for an elliptically polarised wave e becomes √(ex² + ey²), and “e should be replaced by e√2 in the case of circular polarization” (p. 2). The budget assumes both antennas share a polarisation and face each other.

One way. A radar sees the path twice:

Radar systems represent a special case because the signal is subjected to a loss while propagating both from the transmitter to the target and from the target to the receiver.

Its Eq 7 goes as 40 log d with a target cross-section term, and it is not what a one-way link budget computes.

  • The wrong constant for the units. Put 2.44 into the MHz form for 100 m and the loss comes out as 20.15 dB, 60 dB low. Put 2440 into the GHz form and it is 140.15 dB, 60 dB high.
  • Mixing dBm and dBW. They differ by 30 dB. P.525’s section 4 works in dB(W); the worked EIRP of 11.5 dBm is −18.5 dBW. The dB calculator converts between dBm, dBW and watts.
  • Adding powers in decibels. A budget adds gains and losses in dB to one power level. Two powers add as watts.
  • Peak gain instead of gain toward the other end. The EIRP is defined in the direction of the receiver.
  • Leaving out the cable. Transmit-side loss comes straight off the EIRP; receive-side loss comes straight off the received power.
  • Treating the free-space range as the range. It is the range in a vacuum. Size Lother for the real path before quoting a distance.

Sources