Apocalypse 0: Fibre and wavelengths
Author: Erik van der Bij / Peter Jansweijer
Date: 1 July 2010
Last Modified: 1 July 2010
Physical layer
White Rabbit's physical layer bases itself as much as possible on standard Ethernet. In particular IEEE802.3-2008 section 5, chapter 59 1000BASE-BX10 [2].
For optimal results of interoperability and precision of timing, it should use:
- One Single-Mode fibre operated at dual wavelength, one for each
direction
- The use of one fibre guarantees an equal physical length in both directions (not equal delay though)
- LC connector [2]
- Fibre type according to "ITU-T G.652 a,b,c or d" / "IEC 60793-2-50 B1.1 or B1.3" [1] (i.e optimised for 1310 nm)
- 1490nm and 1310nm transmission over a single fibre (1000BASE-BX10)
- More specifically, the Switch ports transmitting downstream (to endpoints) should use 1490nm on the transmitter and 1310nm on the receiver.
Pay attention to the optical power spectral width of the lasers in combination width the fibre dispersion [3,4]. Usually two types of lasers are used:
- Laser Diode (LD) at 1310 nm that has a wide optical spectral width which is not problematic since the selected fibre type has a low chromatic dispersion at 1310 nm
- Distributed Feedback Laser (DFB) at 1490 nm that has a small optical spectral width so more chromatic dispersion in the fibre can be tolerated at 1490 nm
References
-
[1] An overview of singlemode optical fibre specifications The Fibreoptic Industry Association FAI
-
[2] SFP Wavelengths for White Rabbit - from mail archive
- "White Rabbit should follow the 1000BASE-BX10 standard and use 1310/1490 pairs with a single LC connector."
-
[3] Measuring propagation delay over a 1.25 Gbps bidirectional data link - Appendix A
- note: images for 1550 nm instead of 1490 nm, results will be similar.
-
[4] RE: SFP in the MCH v2 - from mail archive
- "the cable type is important that kind of has to be the best at the wavelength of the worst laser."
Note: The mailing archive messages should be clicked on twice. The first time to tell you're not a spammer. The second time it will take you to the exact message.
Erik van der Bij, Peter Jansweijer - 1 July 2010