⚡ Quick Reference
DVB C/N Cheat Sheet
The DVB-S2 thresholds you look up most, straight from ETSI EN 302 307-1 Table 13, plus the three ratios people swap by mistake and the arithmetic that turns a symbol rate into a bitrate.
−2,35
dB, weakest S2 mode
−9,9
dB, S2X VL-SNR floor
10⁻⁷
PER at QEF (S2)
0,2–0,3
dB short-frame penalty
Three ratios, not one
Es/N0
Per symbol. What every MODCOD table gives you.
Eb/N0
Per information bit. Eb/N0 = Es/N0 − 10·log₁₀(η)
C/N
Carrier over noise in a stated bandwidth. Meaningless without that bandwidth.
In symbol-rate bandwidth, C/N = Es/N0. In occupied bandwidth it is lower by 10·log₁₀(1+α) — so "C/N = 7 dB" means Es/N0 = 7 dB in one reading and 8,3 dB in the other at α = 0,35. Establish the reference bandwidth before comparing anything.
QPSK 1/4 has Es/N0 = −2,35 dB and Eb/N0 = +0,75 dB. Same link. The carrier really can sit below the noise floor — anyone who says otherwise is thinking in Eb/N0 while reading an Es/N0 number.
DVB-S2 — the modes you actually use
Ideal Es/N0 at QEF, PER = 10⁻⁷, AWGN, normal FECFRAME. Full 28-mode table in the in-depth version.
| Mode | η bit/sym | Es/N0 dB |
|---|---|---|
| QPSK 1/2 | 0,988858 | 1,00 |
| QPSK 2/3 | 1,322253 | 3,10 |
| QPSK 3/4 | 1,487473 | 4,03 |
| QPSK 5/6 | 1,654663 | 5,18 |
| 8PSK 3/5 | 1,779991 | 5,50 |
| 8PSK 2/3 | 1,980636 | 6,62 |
| 8PSK 3/4 | 2,228124 | 7,91 |
| 8PSK 5/6 | 2,478562 | 9,35 |
| 16APSK 2/3 | 2,637201 | 8,97 |
| 16APSK 3/4 | 2,966728 | 10,21 |
| 16APSK 5/6 | 3,300184 | 11,61 |
| 32APSK 3/4 | 3,703295 | 12,73 |
| 32APSK 5/6 | 4,119540 | 14,28 |
| 32APSK 9/10 | 4,453027 | 16,05 |
Read the overlaps. 8PSK 3/5 delivers 1,780 bit/symbol at 5,50 dB; QPSK 9/10 delivers 1,789 at 6,42 dB — same throughput, 0,92 dB cheaper on 8PSK. And 16APSK 2/3 beats 8PSK 8/9 by 1,72 dB at equal efficiency. "Lower modulation order is more robust" holds only at equal code rate, which is rarely the real comparison.
These are ideal figures
| The table assumes | Reality |
|---|---|
| Computer simulation | Not hardware measurement |
| 50 LDPC iterations | A receiver stopping earlier does worse |
| Perfect carrier & sync recovery | Real demods lose a fraction of a dB |
| No phase noise | Every LNB adds some; high-order APSK is very sensitive |
| AWGN channel | Real links add rain fade, interference, non-linearity |
| Normal FECFRAME, no pilots | Short frames cost 0,2–0,3 dB more |
The standard says it plainly: "For calculating link budgets, specific satellite channel impairments should be taken into account." The table is a floor, not a design threshold.
QEF is not "no errors". It is defined as under one uncorrected error-event per transmission hour for a 5 Mbit/s service — so a higher-bitrate service sees more events at the same PER. And LDPC has a cliff: half a dB above threshold is perfect, half a dB below fails almost completely.
Symbol rate → bitrate → bandwidth
Useful bitrate = Rs × η Occupied BW = Rs × (1 + α)
| MODCOD | Rs Msym/s | α | BW MHz | Mb/s | Es/N0 dB |
|---|---|---|---|---|---|
| 8PSK 2/3 | 35,3 | 0,20 | 42,36 | 69,92 | 6,62 |
| QPSK 3/4 | 27,5 | 0,35 | 37,12 | 40,91 | 4,03 |
| 16APSK 3/4 | 30,0 | 0,15 | 34,50 | 89,00 | 10,21 |
| 32APSK 5/6 | 33,0 | 0,10 | 36,30 | 135,94 | 14,28 |
Roll-off penalties: α = 0,35 → 1,30 dB · 0,25 → 0,97 · 0,20 → 0,79 · 0,15 → 0,61 · 0,10 → 0,41 · 0,05 → 0,21. DVB-S2 offers the first three; S2X adds the rest.
The η column assumes no pilots and is normalised to symbol-rate bandwidth. For efficiency per unit of occupied spectrum, divide by (1+α): 32APSK 5/6 at α = 0,10 is 3,745 bit/s/Hz, not 4,12.
What DVB-S2X adds
| Addition | Why it matters |
|---|---|
| Finer code rates | S2 jumps 8PSK 2/3 (6,62) → 3/4 (7,91), a 1,3 dB gap you waste or fall through. S2X fills it at 6,12 / 7,02 / 7,49 |
| 8APSK, 64/128/256APSK | Up to 5,90 bit/symbol at 19,57 dB |
| VL-SNR π/2 BPSK | Down to −9,9 dB — links DVB-S2 cannot close at all |
| Roll-off 0,15 / 0,10 / 0,05 | More payload in the same allocated bandwidth |
-L MODCOD variants | Constellations optimised for linear channels |
The
-L trade is real: 16APSK 3/5-L needs 0,39 dB less than 3/5 on a linear channel and 0,56 dB more through a hard limiter. Choose on whether your transponder runs near saturation.
Non-linear penalty grows with order: QPSK 2/9 loses 0,40 dB through a hard limiter, 16APSK 1/2-L loses 2,43 dB, 256APSK 3/4 loses 4,45 dB. High-order APSK belongs on a lightly-loaded or linearised chain.
Note the metric change: S2 tables are PER = 10⁻⁷, S2X tables are FER = 10⁻⁵. Different quantities on different objects — close enough to tabulate together, not close enough to compare at the third decimal.
Margin, and why more is not simply safer
| Add for | Notes |
|---|---|
| Implementation margin | From the receiver datasheet, not from any standard |
| Phase noise | Negligible on QPSK, significant at 32APSK and above |
| Short FECFRAME | 0,2–0,3 dB — quantified by the standard itself |
| Transponder non-linearity | Up to 4,45 dB; see the S2X hard-limiter column |
| Interference | The threshold is against C/N+I, not C/N alone |
| Rain fade | Usually the dominant term at Ku and Ka |
Margin is bought with bandwidth. Picking a mode 3 dB more robust than needed means 8PSK 2/3 instead of 16APSK 3/4 — roughly a third of the capacity gone. The question is not "how much margin can I add" but "what availability do I need, and what is the cheapest mode that delivers it".
The structural alternative: receive the same transponder at more than one site and fail over. Rain fade is local, so geographic diversity addresses it in a way no amount of coding gain at one site can.
Not on this sheet: DVB-T2 C/N
EN 302 755, the DVB-T2 standard, specifies the system but contains no required-C/N table — those figures are in the implementation guidelines. And unlike satellite, one number per mode is not very meaningful: the same modulation and code rate can differ by several decibels between fixed-rooftop, portable-outdoor and portable-indoor channel models.
So: ask which channel model any quoted DVB-T2 C/N assumes. Without that, the figure is not comparable to anything — including the satellite numbers above, which are AWGN.