Building Ultra Long Range Toslink

Original Article ↗ Hacker News Discussion ↗

DIY optical audio hacks (lasers, mirrors, robustness)

  • People link a video where TOSLINK LEDs are replaced with lasers for wireless surround.
  • Concerns raised about line-of-sight links breaking from vibration (subs, walls, cars with loud bass). Others note beam divergence and joke about “self-correcting” when bass drops the link and thus the vibration.
  • Critique of Manchester-encoded amplitude-modulated TOSLINK in free space: vulnerable to ambient light; suggestion to modulate onto a higher-frequency carrier like old IR headphones for robustness.
  • Some argue consumer IR (38 kHz carrier) is insufficient for good audio; others counter that IR headphones prove it works in practice.

SFP modules and low‑bitrate signals over high‑speed optics

  • Key observation: with SFPs the project is really “S/PDIF over SFP fiber” rather than extending classic plastic-fiber TOSLINK.
  • Discussion of AC coupling and DC wander: 10G optics expect high-rate, scrambled data; slow Manchester-coded S/PDIF looks almost DC, stressing coupling caps and retimers.
  • Participants note this explains why links only work above ~100–150 kHz effective transition rates.
  • Mention that SDI and AV-specific optics handle “pathological patterns” better; normal Ethernet optics assume pre-scrambled, line-coded signals.

S/PDIF, TOSLINK, HDMI, and formats/DRM

  • Some lament S/PDIF’s ≈1.5 Mbps cap limiting it to compressed 5.1 (DTS, Dolby), pushing people to HDMI for uncompressed surround.
  • Others contest the strict 1.5 Mbps limit, citing 24‑bit/96 kHz stereo specs (~5 Mbps). This remains unresolved in the thread.
  • Lack of bidirectional signaling and robust DRM is cited as a reason TOSLINK wasn’t extended for richer formats; HDMI won due to HDCP.
  • TOSLINK is seen as “boringly reliable” and still widely used (TV → amp, legacy CD/DVD), despite being old.

Audio over Ethernet and live‑sound latency

  • Live‑sound folks compare: AES50 (layer‑1, synchronous, ~62 µs per link) vs Dante/Audio-over-IP (1–10 ms typical, can be lower in some modes).
  • Very low latency over long fiber (≈11 µs) is seen as valuable for digital live audio paths, though venue speaker arrays still deliberately add delay for alignment.

Fiber physics, tools, and oddities

  • Clarification that light in fiber travels ~c/1.5, slower than in vacuum, which matters for latency and trading links.
  • Mention of OTDR launch fiber spools (100–200 km) as an easier way to test ultra-long optical paths.
  • Fiber tech anecdotes: talk sets, non‑intrusive fiber clamps that detect modulation, and the difficulty of mid‑run tapping without splicing.