question about fiber optics

[VirtualLarry]

Do long-haul links use single-mode or multi-mode fiber? Ditto for campus-wide links? Is the signal loss from single-mode fiber less than with multi-mode? Can you fit more data on a multi-mode fiber? Just curious, thanks.

 

[drebo]

It depends on the length. Single-mode fiber has a longer viable length than multimode because the light inside follows one path. Multimode is made in such a way that the light takes multiple paths through the fiber. Multimode adapters are cheaper than single mode. I'd wager that most campus networks are not using fiber for their building-to-building links at all yet. Most are probably still on CAT5. The ones that are on fiber are more than likely using multimode, but it would vary based on the people who are building the network and their preferences.

As far as speed, neither is explicitly faster than the other, and the speed has more to do with the equipment on either end than the fiber itself. For instance, if you have equipment that can understand multiple wavelengths, it's going to be much faster bandwidth-wise than equipment that doesn't. Likewise, whether or not you're using LEDs or lasers as your light source can impact the potential bandwidth (as well as the potential length). That said, you may be confusing multimode with multiple wavelength. They're not the same thing.

Edit: to clarify, the mode is the angle at which the light enters the fiber (or the angle at which it strikes the wall of the fiber, depending on who you're talking to). Single mode fiber enters at one angle and follows one path through the fiber. Multimode enters at multiple angles and follows multiple paths through the fiber. To compensate for the fact that a wider angle of entry would normally take longer to reach the other end, multimode fiber has a variable density core, which is densest at the center decreasing as it goes out (light travels slower in denser glass/plastic). So, the light that spends more time in the outer part of the core, due to its wider angle of entry, will travel faster, reaching the other end at the same time as the light which travels more slowly through the center of the core due to its narrower angle of entry.

 

[VirtualLarry]

Yes, I was confusing multi-mode with multiple wavelength. Thanks for the clarification.

With multi-mode fiber, if the same light source actually travels in multiple paths through the fiber, isn't there a problem with "echo" at the recieving end? Eg. the light takes multiple paths, and one path takes longer than the other, so you end up with two overlapped signals at the end? I can understand how the core density variations can mitigate that, but isn't there still a problem?

 

[cmetz]

VirtualLarry, long-haul links are SMF, generally campus-wide are too. MMF doesn't go all that much farther than cat5.

The issue isn't loss, it's dispersion. You have signal, you just have it scatterred so much your receiver can't reliably detect each symbol correctly at high data rates anymore. SMF is way better about this than MMF, and also, SMF optics are all around better quality / designed for low dispersion and higher sensitivity, while MMF optics are optimized for cost.

In theory, you can multiplex different transmission modes (angles of incidence) on a MMF. In practice, I've never seen it done. Folks do multiplex in wavelength, but that's generally done on SMF (wave division multiplexing, abbreviated WDM). At the distances covered by MMF, the multiplexing gear is much more expensive than just running more fibers. At the distances covered by SMF, it's worthwhile.

 

[cmetz]

VirtualLarry, in MMF, different angles of incidence go in, different angles of exit come out. The idea is that, thanks to the thicker fiber, the signals can bounce around each other and never quite interfere with each other. Draw it on paper and it will become more clear, it's hard to explain in words.

 

[skyking]

MM is good for 220 meters, with an enhanced spec out to 300. That will answer your campus question, since most locations are farther than that.

 

[imagoon]

Multimode fiber:

Typical transmission speeds/distances limits are 100 Mbit/s up to 2 km (100BASE-FX), 1 Gbit/s for distances up to 500?600 meters (1000BASE-SX), and 10 Gbit/s for distances up to 300 meters (10GBASE-SR). *per wikipedia

In most of the campuses my company has done data installs for, you always run MM between buildings. The requirements to meet code on copper data tends to exceed the cost of fiber in a fiber duct. It is far easier to pull a 6 or 12 strand once which is about the same size to slightly larger than a cat6e cable. The size of the cores also have an effect. (same as the differing gauges of copper in the various categories of twisted pair.) Larger cores mated with the proper transponder will go farther and operate better near the limits than a smaller stand cable. Most bad fiber issues are from improperly installed strands. Someone touched the core before termination or cut it incorrectly etc. The gear required to install fiber is very expensive and is not something that most "Joe's data shack" can install properly.

I have 1000BASE-SX running more than the 100 meter limit on copper (between buildings), all the strands tested well with in tolerances on the fiber tester for that speed. (about 450 meters including the tails in this case)

~Chris

*edit messed up meters <-> feet

 

[spidey07]

Also, keep in mind MM fiber is kind of on the way out. Even at 10 Gig Ethernet it doesn't go very far. It's role has largely been put to the data center or horizontal riser cables where you don't need longer than a few hundred meters. Outside of that the smart choice is single mode and what any smart campus will put in for new installations. The fiber is cheap so may as well run both then decide on what money you want to spend on optics and how.