[PLUG-TALK] Fiber specs

Keith Lofstrom keithl at kl-ic.com
Sat Jul 23 04:44:20 UTC 2016


On Thu, Jul 21, 2016 at 06:14:37PM -0700, John Jason Jordan wrote:
> for me, I just got a piece of paper and a pen and wrote down everything
> on the side of the spool (about 3 meters in diameter). There was a paper
> like an invoice under a clear plastic envelope that said:
> 
> 	Customer KGP Logistics
> 	Fiber Specs	Fiber Type
> 				SMF
> 	Wave Length	Max Attn
> 	1550		0.25
> 	1310		0.35
> 
> None of that means anything to me; I just report it in the hopes that
> someone else here can decipher it and help fuel our wild speculation. :)

There's a lot to learn from those few numbers.

A 3 meter spool can be a lot of fiber optic cable; 
if the cable was a centimeter in diameter, and the
spool channel was a meter wide and deep, it could
hold 64 kilometers (40 miles) of cable.

SMF is "single mode fiber".  That means the light is
constrained to a narrow optical channel down the middle
of a wider (but still hair-thickness) glass fiber.  The
light is forced to travel absolutely straight - it cannot
bounce side to side and take a slightly longer time
getting to the far end.  And that means the fiber can
carry a LOT of bits per second.  More than 1 terabit
per second, 1,000 times the bandwidth of a gigabit
internet connection.  Not for one customer - they use
"wavelength division multiplexing" to cram hundreds of
signals through the fiber, at ever so slightly
different frequencies.

The first number is wavelength in nanometers, and my
guess is the second number is the loss (in decibels)
per kilometer. 

Optical light ranges from 400 nanometers (violet) to
750 nanometers (red).  The two wavelengths listed are
near-infrared.  1550 nanometers is a magic wavelength,
because they can use "erbium doped fiber amplifiers",
a magic bit of laser wizardry, to amplify the optical
signal with additional energy provided by a bright
optical source (like LEDs).  

I'm only guessing, but I presume a similar amplifier
works at 1310 nanometers.  The amplifiers amplify
signals going in one direction, so I will FURTHER
guess that the customer-directed signal (download
direction) is at 1550 nanometers, and the lower
speed upload signal is at 1310 nanometers.

So, the loss for 64 kilometers of fiber would be 16 db
(a factor of 40) in the download direction and 22 db
(a factor of 170) in the upload direction.  That
seems like a lot, but radio and electrical cables
lose a lot more signal over much shorter distances.

1550 nanometers is 190 terahertz light, oscillating
almost 80,000 times faster than a 2.45 gigahertz
wireless access point signal.

-----

Two final bits of optical fiber trivia.  Before the
invention of single mode fiber, erbium amplifiers,
wavelength division multiplexing, and very high speed
electronics, they could only move about one gigabit
per second through a trans-ocean fiber cable.  The
internet would have needed vast numbers of cables
between Bandon, Oregon (furthest west in the 48
states) and Japan. 

The cables route from Bandon through Corvallis.  Oregon
State University hosts the amplifiers between the
transpacific fiber and the routes south to California
and east through the Columbia gorge.  OSU gets some of
that bandwidth, and uses it to host the main servers for
the Linux kernel, Firefox, Gentoo, and other open source
projects.  Because of the confluence of fiber bandwidth
and cheap electric power at the Dalles, Google built
their largest data center there.

Alcatel built a huge optical fiber manufacturing plant
in north Portland to make these fiber cables, splice them
with old-style electronic amplifiers, and load the 8000
kilometer lengths of cable into huge cable creels in the
holds of fiber laying ships.  They built 13 huge creels
at the factory to store the cables as they were made,
and test them end to end.

And then technology improved - a LOT.  Now one supercable
can do the work of 1000 old ones.  Alcatel shuttered the
brand new but obsolete plant.  There are 13 huge round
rooms waiting for some new, clever use.  I've never seen
them, but if they were designed to hold 8000 kilometers
of 3 centimeter diameter transpacific cable, they must
be 10 meters deep and 30 meters around.  Sports?  Gaming?  
360 degree surround movies?   Really frigging huge low
tech hard disks?

Keith

-- 
Keith Lofstrom          keithl at keithl.com



More information about the PLUG-talk mailing list