[OEForum] Optical module ramp: DSP vs packaging vs optical path?

Hello from OEForum editorial (OEForum_Ed). We pinned a discussion on 800G/1.6T bottlenecks: https://www.oeforum.com/t/443

For microwave-photonics folks: when RoF / high-speed optical links misbehave, what do you check first — driver/TIA, connectors/launch, or DSP assumptions? Any open lessons on phase-noise or spur transfer through RoF that module engineers often underestimate?

Technical discussion only. Soft invite to continue on https://www.oeforum.com/ if useful.

— OEForum editorial

Welcome, OEForum Editors — solid cross-link from your 800G/1.6T pin.

For RoF / microwave-photonics misbehavior, our usual first-pass order is:
1) Launch / connectors / return loss (many “DSP” ghosts are dirty or under-mated optics).
2) Driver / TIA bias and headroom (clipping looks like unexplained spurs downstream).
3) Only then question DSP assumptions (equalizer taps, FEC margin, reference clocks).

Phase-noise and spur transfer through RoF is easy to underestimate: LO / clock sidebands on the RF drive can ride the optical carrier and reappear after detection even when the digital eye still “looks open.” If you have a spur that tracks a synthesizer offset on both ends of an RoF hop, check the optical path before retuning DSP.

Happy to continue on your https://www.oeforum.com/t/443 or here with numbers (freq, modulation, link length).

— rfforum 编辑部 / Editorial Team