Published Nov 7, 2026 · Last updated Nov 7, 2026 · 8 min · IndieRF

LoRa and Meshtastic Range: Horizon Caps It Near 12 km

A Meshtastic node on Long Fast has the budget for about 166 km of free space with a 20 dB fade margin, yet two nodes on 2 m masts lose sight of each other at 11.7 km. On short masts the radio horizon sets LoRa range, and antenna height is the input that moves it.

Two Meshtastic nodes on Long Fast, 22 dBm into 2 dBi antennas, have enough budget to reach about 166 km in free space and still keep a 20 dB fade margin. Put both antennas on 2 m masts and they lose sight of each other at 11.7 km, the radio horizon. Over flat open ground the ground reflection pulls that in to about 5.0 km. On short masts, LoRa range is set by height, and the link budget has dB left over.

The budget: 136 dB to spend

Free-space path loss at distance dd and frequency ff is

FSPL=20log⁡10dkm+20log⁡10fMHz+32.45 dB\mathrm{FSPL} = 20\log_{10} d_\mathrm{km} + 20\log_{10} f_\mathrm{MHz} + 32.45\,\mathrm{dB}

Long Fast is SF11 at 250 kHz on 906.875 MHz, the US default channel. EIRP is 24 dBm, the receive antenna adds 2 dBi, and the sensitivity is −130 dBm, so the link survives 156 dB of path loss. Keep 20 dB for fading and 136 dB is left to spend, which free space stretches to 166 km. The SX1262 preset on the same page (22 dBm, SF12 at 125 kHz, −137 dBm from Semtech's Table 3-8, 915 MHz) has 143 dB to spend and reaches 368 km.

That −130 dBm comes from the Semtech SX1276 data sheet, Table 10, at the same spreading factor and bandwidth, measured at coding rate 4/6. Meshtastic publishes its modem settings and leaves sensitivity to the chip, and Long Fast runs CR 4/5, so every Meshtastic sensitivity in this note is the chip's figure at a neighboring coding rate.

The 20 dB is the Rayleigh rule. With no line of sight, the chance a fade dips more than FF dB under the mean is 1−e−10−F/101 - e^{-10^{-F/10}}, close to 10−F/1010^{-F/10}, so 20 dB holds 99% of the time and 30 dB holds 99.9%. In free space 20 dB of margin divides the range by ten. A strong line of sight (Rician) needs less, and terrain or a hand around the radio is a separate allowance on top.

On 2 m masts the horizon is 11.7 km

The page's horizon is the 4/3-earth model, with hh in m and dd in km. 4.12 is 2⋅43⋅6371 km×0.001 km/m\sqrt{2 \cdot \tfrac{4}{3} \cdot 6371\,\mathrm{km} \times 0.001\,\mathrm{km/m}}, rounded the way textbooks print it:

dkm=4.12 (h1+h2),h in md_\mathrm{km} = 4.12\,\left(\sqrt{h_1} + \sqrt{h_2}\right), \quad h \text{ in m}

Each 2 m mast sees 5.8 km to its horizon, and two of them see each other out to 11.7 km. Past that the earth itself is in the path, and the dB left in the budget only add margin. The spreadsheet closes the link at 166 km; the curvature of the earth closes it at 11.7.

The page's range readout is the lesser of the free-space range and the horizon. On 2 m masts the horizon wins for every row of the SX1262 and Meshtastic rate tables, which is why the page marks them all with a star.

CaseFree-space range (km)Radio horizon (km)Range readout (km)
SX1262, SF12 at 125 kHz, 2 m and 2 m368.2911.6511.65
Meshtastic Long Fast, 2 m and 2 m165.9811.6511.65
Long Fast, 10 m roof and 2 m165.9818.8618.86
Long Fast, 100 m above the ground around it, and 2 m165.9847.0347.03

Height is the input that moves it

The horizon grows with the square root of height. Raise one end to a 10 m roof and it is about 19 km. Put that end on a hill 100 m above the ground around it and it is about 47 km, still well inside the 166 km budget, so height is still what sets the range.

I spend on the mast before the radio. On 2 m masts every change on the radio side moves only the free-space number, while 8 m more mast at one end adds about 7 km to the range readout.

In the clip, the TX antenna height goes from 2 to 10 m and the range readout grows from 11.7 to 18.9 km, the radio horizon at each height, while free space alone still reaches 166 km.

Meshtastic Long Fast on 2 m masts from the first link in the Link Budget Calculator, with the TX antenna height typed up to 10 m.

Spreading factor trades rate for range

Each step of spreading factor doubles the symbol time, 2SF/BW2^{\mathrm{SF}}/\mathrm{BW}, and from SF7 up buys 1 to 3 dB of sensitivity in the SX1276 table. On the SX1262, SF12 over SF7 is 13 dB, which in free space is 4.5 times the range: 368 km against 82 km, for symbols 32 times longer.

The Meshtastic presets make the same trade. The range column is the page's rate table with Long Fast's radio, free space at a 20 dB fade margin. Sensitivities are SX1276 Table 10 at CR 4/6, as above.

PresetModemSensitivity (dBm)Symbol (ms)Free-space range
Short TurboSF7, 500 kHz, CR 4/5−1160.2633.1 km
Short FastSF7, 250 kHz, CR 4/5−1200.5152.5 km
Short SlowSF8, 250 kHz, CR 4/5−1231.0274.1 km
Medium TurboSF9, 500 kHz, CR 4/5−1221.0266.1 km
Medium FastSF9, 250 kHz, CR 4/5−1252.0593.3 km
Medium SlowSF10, 250 kHz, CR 4/5−1284.10132 km
Long TurboSF11, 500 kHz, CR 4/8−1284.10132 km
Long FastSF11, 250 kHz, CR 4/5−1308.19166 km
Long ModerateSF11, 125 kHz, CR 4/8−13316.38234 km

Even Short Fast reaches 52 km in free space, 4.5 times the 2 m horizon. On short masts a slower preset buys fade margin, and the distance stays at the horizon. Long Fast carries 10 dB more margin than Short Fast at the same 11.7 km, and pays for it with symbols 16 times longer. Each symbol carries SF bits, 11 against 7, so a packet holds the channel about 10 times as long. Pick the slow preset for links that fade, and the fast one when the mesh is busy and the hops are short.

Where free space plus a horizon stops

Both numbers assume a clear path over a smooth earth, with terrain, trees and buildings left out. The page has two path models that show how far a real path falls short.

Two-ray ground adds the reflection off flat earth with a reflection coefficient of −1. Past the crossover distance the loss rises 40 dB per decade, twice the free-space slope, and Long Fast on 2 m masts closes at about 5.0 km, under half the horizon.

3GPP TR 38.901 RMa NLOS is the rural macrocell model with no line of sight, using the page's default 5 m buildings and 20 m streets. With a 10 m base and a 2 m node it gives about 1.3 km, a fifteenth of the 18.9 km horizon. That is the mean loss; the TR quotes 8 dB of shadow-fading σ around it, and the page shows that as a note under the path loss. The compare link below opens free space as link A and RMa NLOS as link B on the same masts, with both ranges marked on the margin plot.

Path, Long FastModel range (km)Radio horizon (km)Range readout (km)
Two-ray ground, 2 m and 2 m5.0211.655.02
Free space, 10 m and 2 m (link A)165.9818.8618.86
3GPP RMa NLOS, 10 m and 2 m (link B)1.2818.861.28

A real hill between the nodes needs a terrain-profile model with elevation data, such as Longley-Rice or ITU-R P.1812, which is the next tool after this page. Treat the free-space and horizon numbers as the ceiling, two-ray as open farmland, and RMa NLOS as a village with the node indoors or behind a barn.

References

  • Semtech, SX1276/77/78/79 data sheet, Rev. 5 (August 2016), Table 10 (sensitivity) and Table 13 (demodulator SNR).
  • Semtech, SX1261/2 data sheet DS.SX1261-2.W.APP, Rev. 1.2 (June 2019), Table 3-8 (sensitivity) and Table 3-9 (output power).
  • Meshtastic documentation, Radio Settings: spreading factor, bandwidth and coding rate per modem preset.
  • 3GPP TR 38.901 V17.0.0 (ETSI TR 138 901, 2022-04), Table 7.4.1-1: RMa, UMa and UMi path loss and shadow fading.
  • T. S. Rappaport, Wireless Communications: Principles and Practice, 2nd ed.: the two-ray ground reflection model and Rayleigh fading.

Questions? Contact

Discussion

No comments yet. Start the thread with a measurement, a correction, or a worked example.