How to Work an In-Building DAS Link Budget (Before the Walk Test Works You)
The cascade, end to end
Every in-building distributed antenna system answers the same question: how much signal survives the trip from the source to the phone or portable radio at the worst corner of the floor? The answer is an accounting exercise called a link budget, and it runs in two halves. The wired half starts at the signal source — a bidirectional amplifier or remote unit — and subtracts everything the distribution network eats: coaxial cable loss, splitter division, connector and jumper losses. Add the antenna gain and you have EIRP, the effective isotropic radiated power actually leaving the antenna. The over-the-air half subtracts path loss between the antenna and the coverage edge. What remains is the edge signal, and the difference between that and your target is the margin — the single number this calculator exists to produce.
The discipline of the exercise is that every term must come from somewhere defensible. This model uses the coax vendor’s own published loss formula, pure mathematics for the splitter division, an explicit user-owned allowance for the miscellaneous losses, textbook physics for free space, and published planning exponents for indoor propagation.[3] Nothing in the cascade is a black box, which is exactly what makes the output arguable in front of an engineer — or an authority having jurisdiction.[1]
Coax loss: the vendor’s own formula
Coaxial cable loss is the best-documented term in the budget, because the manufacturer publishes it. Times Microwave’s public LMR Guide gives, for every cable in the family, an insertion-loss model of the form K1 times the square root of frequency plus K2 times frequency, per hundred feet or per hundred meters — with the explicit instruction to use the K column that matches your length unit.[3] This calculator computes that formula verbatim for LMR-400 and LMR-600, and the vendor’s own free online calculator provides the cross-check for any number this page produces.[4]
The square-root term is why frequency quietly dominates cable planning. A hundred feet of LMR-400 costs about 2.3 dB at 150 MHz, 3.9 dB at 900 MHz, and 5.9 dB at 1900 MHz. Stepping up to LMR-600 buys roughly a third of the loss back at every frequency — the classic trade of copper for decibels — and on long horizontal runs that trade decides whether the branch works at all. Loss scales linearly with length, so a 250-foot run is simply 2.5 times the hundred-foot figure; there is no shortcut and no rounding grace.
Splitters, miscellaneous losses, and EIRP
A splitter divides power, and the division itself is pure arithmetic: ten times the logarithm of the way count — 3.0 dB for a two-way, 6.0 dB for a four-way, 9.0 dB for an eight-way. Real splitters add a few tenths of a decibel of excess loss on top, and every connector pair and jumper in the chain adds its own small tax. This model deliberately does NOT seed vendor-specific excess values; instead the other-losses input carries them, defaulting to a 2 dB allowance you should replace with the sum from your actual bill of materials. That keeps the invented-data count in this calculator at exactly zero.
The sum of the wired half is EIRP, and it earns its place as a visible intermediate result twice over. First, it isolates the distribution design from the propagation guess — if EIRP is wrong, no path-loss model will save the branch. Second, EIRP is a regulated quantity: FCC equipment rules cap radiated power for signal boosters and DAS transmitters by band and service class, so the number this calculator reports beside the result is the one the authorization paperwork will eventually ask about.
Path loss: exact physics, then honest guesswork
Free-space path loss is exact: twenty times the logarithm of frequency plus twenty times the logarithm of distance, less the 27.55 constant when frequency is in megahertz and distance in meters. Every doubling of either frequency or distance costs six decibels. Indoors, the physics stops being exact and becomes statistics: walls, metal studs, glazing, and furniture steepen the distance exponent, and the log-distance model captures that with a single parameter — 2.0 in free space by definition, roughly 2.6 for open interiors, and about 3.3 for dense, partitioned construction, per the published propagation planning literature.
The honest way to read the environment selector is as a bracketing tool, not a measurement. Run the budget at the optimistic and pessimistic exponents and see whether the design survives both; if the margin only exists in the friendly case, the branch needs a closer antenna, a bigger cable, or a shallower split. Measured indoor exponents from below 2 in reflective corridors to above 4 in concrete cores all appear in the literature, which is why no responsible link budget skips the walk test — and why the code does not let ERRCS designs skip the measured acceptance grid regardless of any model.[1]
The ERRCS context — and what this model is not
For emergency responder coverage, the widely adopted floor is a −95 dBm signal at the coverage edge, with area-coverage percentages layered on top: the fire-code family requires the overwhelming majority of each floor to meet it, with the NFPA framework requiring 99 percent in areas the fire authority designates as critical and 90 percent in general areas.[1] The consolidated NFPA emergency-services communications standard carries the survivability, backup-power, monitoring, and testing provisions that turn a coverage system into an emergency responder communication enhancement system.[2] Jurisdictions adopt and amend these codes on their own schedules, so the edition in force — and its reading — belongs to the local authority having jurisdiction.
This calculator is a planning model, and the boundary matters enough to state twice. It will tell you early whether a proposed branch has a realistic chance of meeting the target, which is worth real money before conduit is run. It will not demonstrate compliance — compliance is measured, floor by floor, on the installed system, and accepted by the AHJ.[1] And it deliberately excludes donor-side engineering: rebroadcast agreements, donor antenna design, and per-carrier or per-agency signal data are proprietary to those operators and outside a public-data model’s remit. The cascade math here is the part of the problem that IS public — used inside that boundary, it is the fastest sanity check in the toolbag.