Today, data center optics are rapidly transitioning to 200GE/400GE rates. In fact, the adoption of 400GE pluggable optics in the data center is in full swing, with the deployments anticipated to spark in 2022. To get to this point, new silicon and circuits were required, such as linear broadband amplifiers and drivers, and adaptive digital equalizers to handle the PAM4-based 400G interfaces. As with any new optics technology, standards were developed and ratified and new test solutions were developed to validate the new standards.

With 400GE optics mass deployment around the corner, the data center interconnect (DCI) ecosystem now has its eye on 800GE rates, as bandwidth and performance demands on the network continue to accelerate.

In this blog, we will cover the test challenges we faced with 400GE optics, and how those challenges will carry over and still be quite relevant for the new 800GE rates.

Key tests for 400GE optics to avoid dangerous testing shortcuts

Optics vendors exert great effort to differentiate their products by minimizing/optimizing power budgets, reducing noise, and operating within IEEE specs. Even with that careful tuning, 400G links can have substantial bit error rates (BER), so forward error correction (FEC) is now a requirement for 400G rates.

We saw testing gaps emerge with 400G optical transceiver modules that will certainly carry over to 800G unless addressed now. Ideally, a test solution must provide clear insights into both component and system design tradeoffs while FEC is “turned on”. This is the only way the systems engineer can predict the FEC margin under real conditions (i.e., stress on the link).

Today, we often see test engineers running their BER test (BERT) and then calling it a day without employing FEC, let alone analyzing FEC margin. To fully test the system, it is critical to test the optical transceiver module itself and the associated host application-specific integrated circuit (ASIC).

To avoid validation pitfalls, 400GE and 800GE FEC-aware receiver testing should include frame loss ratio (FLR) testing in 400/800GE links with one lane of stressed FEC encoded test pattern (using a BERT) while the other 7 lanes are supplied with non-stressed data that is aligned and properly encoded with the FEC test pattern (using a PRBS and FEC codeword generator).

This testbed enables other critical validation results such as:

Per-lane Raw BER (unstressed)
Pre-FEC BER
Error distribution across FEC code work
Error density across all lanes
Common Management Interface Specification (QSFP-DD MSA)

Conclusion

The new and unique problems in testing 400G and now 800G optical transceiver modules requires purpose-built test solutions that combine BERT and FEC testing.

Keysight’s N4891A is the world’s first to address the test challenges described in this blog. A flexible and complete Layer 1 FEC-aware compliance test solution, it addresses a critical gap in the testing of 400GE-class optical transceiver modules and associated host ASIC.

The Keysight N4891A solution combines the M8040A 64 Gbaud high-performance BERT to generate the correct striped FEC test pattern on the single lane under test, along with a A400GE-QDD 400GE Layer 1 BERT and KP4 FEC multiport test system to provide the proper aligned striped FEC traffic on the other seven lanes to measure the FLR of the tested interface under stress conditions.

Combined with Keysight compliance test solutions such as the optical receiver stress test and the IEEE 802.3bs receiver test pre-compliance application, the 400GABASE FEC-aware solution provides fast, accurate, and repeatable optical and electrical stress signal calibration to test interoperability requirements.

Michelle Havard

Improving 400/800GE optics: Is your Layer 1 FEC compliance testing enough?