Why RF Engineers Use Decibels
Walk into any RF lab, read any datasheet, or open any link budget spreadsheet and you will immediately encounter decibels — dB, dBm, dBi, dBd, dBW. They are the universal language of RF engineering, and for good reason. A modern wireless system spans power levels from kilowatts at the transmitter to femtowatts at the receiver input — a range of more than 1018. Expressing that range in watts would require unwieldy scientific notation at every step.
The decibel compresses this enormous span into manageable numbers, and — crucially — it turns multiplication into addition. Instead of multiplying gains and dividing losses in a link budget, you simply add and subtract dB values. This single property makes the decibel indispensable.
A decibel is always a ratio. It compares one power level to another using a logarithm. The "reference" that the ratio is taken against determines the suffix — dBm means relative to 1 milliwatt, dBi means relative to an isotropic antenna, and so on. Plain "dB" with no suffix is a dimensionless ratio with no implied reference.
dB — The Pure Ratio
The decibel (dB) on its own expresses the ratio between two power levels on a logarithmic scale. The definition is:
Because we are taking a ratio, the units cancel — a dB value is dimensionless. It simply says "this signal is X times stronger or weaker than that signal." Some key values every RF engineer memorises:
- +3 dB = power doubled (×2)
- +10 dB = power increased tenfold (×10)
- +20 dB = power increased 100× (×100)
- −3 dB = power halved (×0.5)
- −10 dB = power reduced to one-tenth (×0.1)
- 0 dB = no change (ratio of 1:1)
When working with voltage ratios rather than power ratios, the coefficient changes from 10 to 20: dB = 20 · log₁₀(V₂ / V₁). This is because power is proportional to voltage squared (P = V²/R), and log(V²) = 2·log(V). Always clarify whether you are working with power or voltage ratios to avoid a 6 dB error.
dB in Practice — Gains and Losses
The real power of dB is in chain calculations. Consider a signal passing through an amplifier, a cable, and then an attenuator:
In linear terms that same calculation would be: 1 × 100 × 0.5 × 0.25 = 12.5 — achievable, but error-prone with many stages. In dB: 0 + 20 − 3 − 6 = +11 dB. Much faster and far less prone to mistakes.
dBm — Power Relative to 1 Milliwatt
dBm is the most common absolute power unit in RF engineering. It expresses power relative to exactly 1 milliwatt (1 mW):
dBm is an absolute power level — it tells you exactly how many watts you have, not just a ratio. The reference point of 1 mW was chosen because it sits conveniently in the middle of the power range encountered in RF work — strong enough to be measurable but small enough to represent received signals.
dBm Reference Table
Memorising these values lets you instantly sanity-check any power figure in a datasheet or link budget:
- +43 dBm = 20 W (typical mobile base station transmit power)
- +37 dBm = 5 W (handheld radio transmit)
- +30 dBm = 1 W
- +20 dBm = 100 mW
- +10 dBm = 10 mW
- 0 dBm = 1 mW (reference point)
- −10 dBm = 100 µW
- −30 dBm = 1 µW
- −50 dBm = 10 nW (strong cellular received signal)
- −90 dBm = 1 pW (weak but usable signal)
- −100 dBm = 0.1 pW (near sensitivity limit for many receivers)
- −120 dBm = 1 fW (near thermal noise floor at 1 MHz BW)
You can add or subtract dB from dBm to get dBm. But you cannot add two dBm values — that would be adding two absolute powers, which requires converting back to watts first. Common mistake: "+20 dBm + 20 dBm = 40 dBm" is wrong. The correct answer is +23 dBm (doubling the power adds only 3 dB).
dBW — Power Relative to 1 Watt
dBW is identical in concept to dBm but uses 1 watt as the reference instead of 1 milliwatt. It is commonly used in satellite communications and high-power transmitter specifications where milliwatts would produce unwieldy large numbers.
The relationship is fixed: dBm = dBW + 30. Satellite engineers often quote EIRP in dBW; cellular engineers prefer dBm. Know both and converting between them is trivial.
dBi — Antenna Gain Relative to Isotropic
dBi is the standard unit for antenna gain. The suffix "i" stands for isotropic — a theoretical antenna that radiates equally in all directions, like a perfect sphere. No real antenna achieves this, but the isotropic radiator provides a universal, reproducible reference point.
A 0 dBi antenna has the same peak gain as a perfect isotropic radiator. A +10 dBi antenna focuses 10× more power in its peak direction than an isotropic antenna would. A −3 dBi antenna is less directive than isotropic — which is physically impossible for a lossless antenna but can occur with lossy or inefficient designs.
Typical dBi Values by Antenna Type
- Short monopole / whip: −4 to 0 dBi
- Half-wave dipole: 2.15 dBi
- Microstrip patch antenna: 5–8 dBi
- Yagi-Uda (6 element): 10–12 dBi
- Sector antenna (base station): 14–18 dBi
- Parabolic dish (1 m at 10 GHz): 35–40 dBi
- Large phased array: 30–50+ dBi
dBd — Antenna Gain Relative to a Dipole
dBd uses the half-wave dipole as the reference instead of the isotropic radiator. The half-wave dipole is a practical, buildable antenna with a well-characterised radiation pattern, making it a convenient real-world reference for antenna comparisons.
The 2.15 dB offset is fixed and universal. A Yagi quoted at 10 dBd has a gain of 12.15 dBi. Always check which reference a manufacturer uses — quoting in dBd makes an antenna appear to have 2.15 dB less gain than the same antenna quoted in dBi, even though the antenna is identical.
When comparing antennas from different manufacturers, always verify whether gain is quoted in dBi or dBd. A vendor using dBd will appear to have lower-gain products than a competitor using dBi — even if the antennas perform identically. Reputable datasheets always state the reference explicitly.
EIRP and ERP — Combining Power and Gain
Two derived quantities appear constantly in regulatory documents, satellite specs, and cellular planning — EIRP and ERP.
EIRP (Effective Isotropic Radiated Power) is the transmitter power combined with the antenna gain referenced to isotropic:
ERP (Effective Radiated Power) uses the dipole as reference and equals EIRP − 2.15 dB. ERP is used in broadcast regulations (FCC Part 73); EIRP is used in satellite and mobile standards. They describe the same physical quantity with different reference antennas.
Quick Conversion Reference
These conversions and relationships are worth committing to memory:
- dBm → W: P = 10^(dBm/10) / 1000
- W → dBm: dBm = 10·log₁₀(P_watts × 1000)
- dBm → dBW: dBW = dBm − 30
- dBi → dBd: dBd = dBi − 2.15
- dBd → dBi: dBi = dBd + 2.15
- +3 dB: ×2 power
- +10 dB: ×10 power
- −3 dB: ÷2 power (half-power point, used to define antenna beamwidth)
At room temperature (290 K), the thermal noise power in a 1 Hz bandwidth is −174 dBm/Hz. For a 1 MHz receiver bandwidth: noise floor ≈ −174 + 10·log₁₀(10⁶) = −174 + 60 = −114 dBm. For 20 MHz (LTE): −174 + 73 = −101 dBm. This is the absolute minimum signal level any receiver can detect, before adding its own noise figure.
Summary — The dB Family at a Glance
- dB — Dimensionless power ratio. No reference. Used for gains, losses, and differences. Always add/subtract, never add as absolutes.
- dBm — Absolute power, referenced to 1 mW. The everyday unit for signal levels in RF systems. dBm = dBW + 30.
- dBW — Absolute power, referenced to 1 W. Used in high-power and satellite work. dBW = dBm − 30.
- dBi — Antenna gain, referenced to isotropic radiator. The universal standard. dBi = dBd + 2.15.
- dBd — Antenna gain, referenced to half-wave dipole. Used in some datasheets and broadcast regulations. dBd = dBi − 2.15.
- EIRP (dBm) — Transmit power + antenna gain (dBi) − cable losses. The effective power radiated toward a target.
Understanding these units fluently is one of the highest-return investments you can make as an RF engineer. Every calculation — link budgets, noise analysis, amplifier cascades, regulatory compliance — uses them. Once they become second nature, the rest of RF system design clicks into place.
Worked Examples — 5 Real RF Calculations
The best way to cement these conversions is to work through realistic problems. Each example below walks through a complete calculation step-by-step, showing exactly how dB, dBm, dBi, and dBd are used together in practice.
Transmitter Power: Watts to dBm
Problem: A Wi-Fi access point transmits at 100 mW. The datasheet for the downstream LNA requires the input power in dBm. What is 100 mW expressed in dBm?
Sanity check: 0 dBm = 1 mW, and 100 mW is 100× that, so +20 dB above 0 dBm = +20 dBm. ✓
Note: the FCC limits Wi-Fi EIRP to +36 dBm (4 W), so 100 mW Tx leaves 16 dB of headroom for antenna gain.
Received Signal: dBm to Microwatts
Problem: A spectrum analyser shows a received signal at −43 dBm. The ADC datasheet asks for the input power in microwatts (µW). What is −43 dBm in µW?
Quick shortcut: −43 dBm = −40 dBm − 3 dB = 100 nW ÷ 2 = 50 nW. ✓
This level is typical for a GPS receiver input — well above the noise floor but tiny in absolute terms.
Antenna Gain: dBd to dBi
Problem: A Yagi antenna datasheet quotes a gain of 9.85 dBd. Your link budget spreadsheet uses dBi throughout. What is this antenna's gain in dBi? Is it comparable to a 12 dBi patch array from another vendor?
The two antennas are equivalent in gain. The difference is purely how each vendor chose to present the spec — one used dBd, the other dBi. Without this conversion, the Yagi would appear to have 2.15 dB less gain than it actually does.
Full Link Budget — EIRP and Received Power
Problem: A 5.8 GHz drone data link has the following components. Calculate the EIRP and estimate the received signal level at 500 m.
- Transmit power: 27 dBm (500 mW)
- Transmit antenna gain: 8 dBi (patch)
- Cable + connector losses: 1.5 dB
- Free-space path loss at 500 m, 5.8 GHz: 94 dB
- Receive antenna gain: 5 dBi (whip)
A typical 5.8 GHz receiver sensitivity is around −80 dBm, so this link has −55.5 − (−80) = 24.5 dB of link margin — comfortably above the minimum required for reliable operation, even with fading and interference headroom.
Noise Floor and SNR Calculation
Problem: A receiver has a noise figure of 6 dB and a bandwidth of 10 MHz. The received signal is −65 dBm. What is the noise floor and the SNR? Is the signal detectable?
+33 dB SNR is excellent — most modulation schemes need only 10–25 dB SNR for reliable operation (e.g. 64-QAM requires ~24 dB). This receiver will decode the signal cleanly. If the noise figure were 20 dB instead of 6 dB, SNR would drop to +19 dB — still workable but with significantly less margin, illustrating exactly why low-noise front-ends matter.