Big spec number? Check! +10 dB on a slide? Check! Still confused why the link underperforms or fails EMC? Thatโs because dB, dBm, dBi and dBc are not interchangeable. Each has a different reference and swapping them quietly breaks budgets, masks interference and derails compliance.
-
What the dB Family Really Means?
dB is a ratio (unitless, logarithmic) while dBm is absolute power referenced to 1 mW. dBi is antenna directivity gain relative to an isotropic radiator (a pattern metric, not power). dBc is a relative measure to the carrier (spur/noise power down from the carrier). If you donโt track the reference, youโre adding unlike quantities and fooling yourself. -
Why Mixing Them Hurts Real Systems?
You can add dBi to dBm only when forming EIRP and only after subtracting feed/cable/radome losses. Quote spurs in dBc without carrier level and you might pass a relative spec while failing absolute limits. Compare antennas in dBd vs dBi and youโre off by 2.15 dB before the first test. Units drive decisions, wrong units drive re-spins. -
Context Decides the Correct Metric?
Use dBm at a connector plane, dBi for spatial pattern gain, dBc for spectral cleanliness around the carrier and plain dB for ratios like filter rejection or mismatch loss. At mmWave, fixture/probe losses and unintended radiation make โjust dBmโ meaningless without de-embedding and chamber verification. Write the reference every time, your future self will thank you. -
Critical Formulas:
a). Power ratio (dB):
โ dB = 10ยทlogโโ(Pโ/Pโ)
b). Absolute power (dBm):
โ P[dBm] = 10ยทlogโโ(P[W]ยท1000)
c). EIRP from conducted power:
โ EIRP[dBm] = P_tx[dBm] + G_ant[dBi] โ L_cable[dB] โ L_radome[dB]
d). Relative to carrier (dBc):
โ dBc = 10ยทlogโโ(P_spur / P_carrier) -
Real-World Pitfalls:
- An IoT hub โmeetsโ 36 dBm EIRP on paper by adding 9 dBi to 27 dBm but ignores 2.5 dB feed loss, actual EIRP โ 33.5 dBm and coverage shrinks.
- A backhaul team compares 6 dBd to 8 dBi panels and picks the wrong dish, 6 dBd โ 8.15 dBi, costing 0.15 dB of precious margin.
- A transmitter passes โ45 dBc spur spec at low carrier power then fails absolute masks when PA output rises, relative numbers hid an absolute breach.
- A 5G small cell tallies port dBm and panel dBi but skips tilt/radome/coax losses, the โgreenโ map becomes holes in the field.
Get the references right, state the plane and always write the unit. Do that and your link budgets stop lying and your compliance lab stops surprising you.
LinkedIn: 