How to Calculate an RF Link Budget by Hand

Back to Blog

List Every Gain and Loss in the Chain

A link budget is just a sum of decibel terms, so the first task is to enumerate the chain end to end. On the transmit side you have the radio's output power in dBm, the loss of the coax and connectors running up to the antenna, and the transmit antenna's gain in dBi. Across the air you have the free-space path loss, plus any extra losses for obstructions, rain, or foliage. On the receive side you have the receive antenna gain, then the coax and connector loss down to the radio. Everything expressed in decibels adds and subtracts, which is the whole reason the industry works in dB rather than watts.

Pull the transmit power and receiver sensitivity from the radio's datasheet, because those are device-specific and you must not invent them. Antenna gains come from the antenna datasheet in dBi. Cable loss depends on the coax type, length, and frequency and comes from the cable manufacturer's loss-per-length table; a long run of thin coax at a high frequency can throw away much of your budget before the signal ever leaves the mast. Connector losses are small but real and worth a fixed allowance per connector pair.

Keep transmit and receive antenna choices consistent with the network topology. A point-to-point link can use a high-gain directional antenna at both ends, while a master serving many remotes usually runs an omnidirectional antenna at the master and a Yagi at each remote. The master's omni gain applies to every link in the budget, so a weak master antenna quietly penalizes the whole network.

Compute Free-Space Path Loss

The dominant loss on most clear paths is free-space path loss, the spreading of energy as the wave expands. In its everyday engineering form, FSPL in dB = 20 log10(d) + 20 log10(f) + 32.44, where d is the path length in kilometers and f is the frequency in megahertz, and the 32.44 constant bundles the unit conversions. Read the structure: loss rises by 20 dB per decade of distance and 20 dB per decade of frequency, so both doubling distance and doubling frequency each add about 6 dB of loss. That 6-dB-per-doubling behavior is worth memorizing as a field sanity check.

Free-space loss assumes an unobstructed path, which is why you verify clearance separately with Fresnel-zone geometry. If the path clips an obstruction, add a diffraction-loss allowance on top of FSPL, because the free-space number alone will be optimistic. On longer or higher-frequency links, atmospheric and rain attenuation also add terms; for the bands and short hops typical of SCADA telemetry these are usually small, but they belong in the ledger for completeness on marginal paths.

Because FSPL is the biggest single term, it is also the biggest lever. Shortening the path with a mid-point repeater, as covered in the guide to commissioning a store-and-forward repeater, cuts each hop's distance and therefore its path loss, which is often cheaper than chasing a few more dB out of antennas and coax on one long hop.

Solve for Received Power and Fade Margin

Now assemble the ledger into one equation. Received power in dBm equals transmit power, plus transmit antenna gain, minus transmit cable and connector loss, minus free-space path loss, minus any obstruction and atmospheric losses, plus receive antenna gain, minus receive cable and connector loss. Work it as a running total, starting from transmit power in dBm and adding each signed term, and the answer lands directly in dBm as the signal level presented to the receiver.

The number that decides the link is not received power by itself but the fade margin: received power minus the receiver's sensitivity, both in dBm. Sensitivity is negative and large in magnitude, so subtracting it from a received level that is also negative yields a positive margin when the link closes. That margin is the buffer that keeps the link up when rain, foliage growth, atmospheric fading, or antenna misalignment temporarily rob signal. The concept and typical target ranges live in the reference on fade margin.

Do not confuse this signal-budget margin with the geometric clearance from the Fresnel check; they are independent gates a path must both pass. A path can have generous fade margin yet fail Fresnel clearance and suffer diffraction, or clear every Fresnel zone yet have too little fade margin to survive a rainstorm. Build the full picture by pairing this ledger with the RF link budget concept page and the path-verification how-to.

Verifying the Result

Cross-check the arithmetic with the 6-dB rules. Halving the distance should reduce FSPL by about 6 dB; doubling frequency should increase it by about 6 dB. If a change of that size in an input does not move your total by roughly that amount, you have a sign error or a units slip in the FSPL term, which is by far the most error-prone line in the ledger.

Sanity-check the final received power against the receiver's noise floor and sensitivity. A received level that lands above sensitivity by a comfortable margin predicts a solid link; a level below sensitivity predicts no link at all, no matter how the rest of the network is built. If the paper margin is thin, the fix menu is the same list you built in step one: more antenna gain, lower-loss coax, a shorter hop via a repeater, or a higher-power radio within regulatory limits.

Confirm the paper budget in the field at commissioning by reading the received signal level the radio reports and comparing it to your predicted received power. A field reading close to prediction validates the whole budget; a reading many dB weaker than predicted means a real-world loss you did not account for - a bad connector, water in the coax, a misaimed antenna, or a missed obstruction - and it points you straight at the term to investigate.

Common Mistakes

The classic mistake is a units error in the FSPL formula. The 32.44 constant is specific to kilometers and megahertz; use miles or gigahertz and the constant is wrong and the entire budget is off by tens of dB. Fix your units to km and MHz before anything else, or use a form of the constant matched to whatever units you prefer, but never mix them.

A second recurring error is forgetting coax and connector losses, especially on tall masts with long runs of lossy cable at higher frequencies. Those losses hit twice, once on transmit and once on receive, and on a high-band link they can quietly consume more budget than the antennas provide. Always pull the real loss-per-length figure for the actual cable at the actual frequency rather than assuming it is negligible.

The third is treating the raw received power as the pass criterion instead of the fade margin. A link that just barely reaches sensitivity with zero margin will fail the first time the weather turns, foliage grows, or an antenna shifts. Require a positive fade margin sized for the path's conditions, and remember it must clear alongside, not instead of, the Fresnel-zone geometry.

Frequently Asked Questions