Reading a BER curve
A BER vs Eb/N₀ curve shows the theoretical bit error rate of a modulation over an additive white Gaussian noise (AWGN) channel as a function of the energy-per-bit to noise-density ratio. Lower and further left is better. To hit a target BER (say 10⁻⁶), read across from the BER axis to a curve, then down to the required Eb/N₀ — that is the number you feed into a link budget as "required Eb/N₀".
Formulas (AWGN, uncoded)
BPSK / QPSK: P_b = Q(√(2·Eb/N0))
M-PSK: P_b ≈ (2/k)·Q(√(2k·Eb/N0)·sin(π/M)), k = log₂M
M-QAM: P_b ≈ (4/k)(1−1/√M)·Q(√(3k/(M−1)·Eb/N0))
Coherent FSK: P_b = Q(√(Eb/N0)) DBPSK: P_b = ½·e^(−Eb/N0)
16/32-APSK * (DVB-S2): no simple closed form — estimated by a nearest-neighbour union bound over the ring constellation (4+12 and 4+12+16), normalised to unit average energy. In linear AWGN, APSK sits just behind the equivalent QAM; its real advantage is robustness to amplifier nonlinearity. Curves marked * are approximate.
where Q(·) is the Gaussian Q-function. Higher-order schemes carry more bits per symbol (better spectral efficiency) but need a higher Eb/N₀ for the same BER. Forward error correction (e.g. LDPC in DVB-S2) shifts these curves sharply to the left. All computation runs locally in your browser.