News
Every optical transceiver datasheet lists a "FEC type" — RS-FEC, oFEC, C-FEC — but most network engineers treat it as a checkbox rather than a decision. Forward Error Correction is not just a protocol detail. The FEC you choose — or the one built into your transceiver — directly determines how far your signal can travel, how much margin you have against fiber degradation, and whether your link stays up during a micro-bend event. Here is what each FEC type actually gives you.
FEC works by adding redundant bits to the transmitted data stream. The receiver uses those redundant bits to detect and correct errors without requesting a retransmission. The benefit is measured as net coding gain — the effective improvement in signal-to-noise ratio, in dB. A stronger FEC gives you more dB of headroom, which translates to longer reach, more margin against aging, or the ability to use lower-cost optics on marginal links.
The tradeoff: stronger FEC adds more overhead (redundant bits), consuming a slice of your line rate. And FEC encoding/decoding adds latency — negligible for most applications, but material for high-frequency trading and some HPC workloads.
| FEC Type | Net Coding Gain | Overhead | Latency | Where Used |
|---|---|---|---|---|
| RS-FEC (528,514) | ~6.2 dB | ~2.7% | ~100 ns | 100G/200G/400G PAM4 (intra-DC) |
| RS-FEC (544,514) | ~6.5 dB | ~5.8% | ~120 ns | 800G PAM4 (SR8/DR8/FR4) |
| oFEC (OpenZR+) | ~10.8 dB | ~15% | ~1 µs | 400G/800G coherent ZR+ |
| C-FEC (proprietary) | ~11.5–12.0 dB | ~15–25% | ~1–2 µs | Proprietary coherent (long-haul) |
| SDFEC (soft-decision) | ~11.0–11.5 dB | ~20% | ~2 µs | Ultra-long-haul, subsea |
Before OpenZR+ standardized oFEC, coherent transceivers used vendor-specific FEC algorithms. A module from vendor A could not interoperate with a module from vendor B — the FEC implementations were incompatible at the bit level, even if the modulation and wavelength matched. This was a primary mechanism for vendor lock-in in coherent optical networks.
oFEC changed this. It provides 10.8 dB of net coding gain — strong enough for 500+ km metro/regional reaches — while being fully interoperable across any OpenZR+-compliant transceiver, regardless of manufacturer. For operators building multi-vendor DCI networks, oFEC is the single most important interoperability standard after the optical modulation format itself.
Here is a concrete example. An 800G ZR+ transceiver operating at 120 km with DP-16QAM modulation needs a pre-FEC BER below approximately 2.0 × 10⁻². The oFEC decoder can correct errors down to this threshold and deliver a post-FEC BER below 10⁻¹⁵ — effectively error-free.
If you tried to run the same link with only RS-FEC (6.5 dB gain), the pre-FEC BER threshold would be approximately 5.0 × 10⁻⁴ — two orders of magnitude tighter. You would need 4–5 dB more signal power to hit that threshold, which means either shorter reach or more amplification.
This is why PAM4 transceivers (SR8/DR8/FR4) use RS-FEC — they operate at short distances where the optical signal-to-noise ratio is high enough. Coherent ZR+ transceivers use oFEC because they push longer distances where every dB counts.
Practical takeaway: When comparing transceivers from different vendors, check the FEC type and net coding gain — not just the Tx power and Rx sensitivity. Two modules with identical optical specs but different FEC implementations can have a 4+ dB difference in effective reach. The FEC is not a footnote on the datasheet — it is part of the link budget.
APEX Group 400G CFP2-DCO and 800G QSFP-DD ZR+ coherent transceivers use OpenZR+-compliant oFEC with 10.8 dB net coding gain, tested for multi-vendor interoperability. For intra-data-center links, the 800G SR8, DR8, and FR4 PAM4 modules use IEEE-standard RS-FEC (544,514) with full compliance across switch platforms.
APEX GROUP — www.apexallinone.com
