Last time I asked the list for recommendations on how to solve a short reach
fiber problem I got some amazing answers and suggestions -- for which, I am
still appreciative of. I summarized for everyone who asked and will gladly
do so if anyone else would like to know.
I've got a new pickle I am trying to work through and have been wholly
unexcited by the solutions various vendors are talking about (read: $=x^n).
I am trying to signal over a dark fiber (SMF) loop of about 200 miles.
(150mi on one leg and 50mi on the other). I would _like_ to find a 10GE
solution (say a XENPAK module that will do 150miles), but GE would do if it
supports some kind of xWDM.
I am afraid I am going to have to go Lucent LamdaXtreme or Cisco ONS 15800,
but am hoping that is just overkill. Heck, a source for used LH/ELH SONET
OADMs would be fanastic if a router-based solution isn't obvious.
I am certain many on this list have tackled this kind of problem in
particularly ingenious ways. Your pointers are very much appreciated.
Thanks in advance,
DJ
As soon as you want to go more than 60mi/100km in one
unrepeated/unregenerated hop, things start to cost large amounts of money.
Are you sure you cannot put some kind of repeater 75mi into the 150mi
leg? Then you could do it costeffectively with a GigE/OC48 solution which
would probably also support CWDM.
A 1550nm GigE repeater with 24dB dampening budget will only set you back
$5-6k or so, with a OC48/STM16 equivalent costing not much more.
I have successfully done 150km in one hop with the Finisar 31dB FTR1619
GBIC (specced at approx 31-33dB link budget), but that was only for test
purposes, didn't have enough margin for production environment I would
say.
I'm afraid that there isn't a good commercially-available way to send any high speed optical signal over a single continuous fiber for 200 miles. The physics simply prohibit it.
The source of the problem is a limit on the maximum launch power into the fiber before you start to see interesting material interaction effects of high laser power on glass fibers. At last guess, that limit was about +17dBm into standard SMF-28. Couple this with a typical receive sensitivity of -12dBm for 10gpbs signals and you've for about 29dB of link margin to work with. Assuming of course that there aren't any patches or other attenuators in the path, and at a typical attenuation of .25db/km, that gets you about 116km. That's about 70 miles. Of course, I might be wrong about the exact figures, but you get the idea.
Using an optical amplifier is the best bet but like all amplifiers it will introduce noise into the signal. You'll need something to measure OSNR (Optical SNR) to make sure that your signal will even be readable at the far end. Amps generate more noise at high gain levels. You COULD go out and buy an optical gain block to do this but it would be a kludge and COULD damage your optics on the far end.
In short, there aren't many solutions that are cheap that will let you do this without putting some equipment along the way to amplify, regenerate, or otherwise enhance the signal.
FWIW, most terrestrial fiber networks are designed around a 25db span loss, or roughly 70-80 km. If there was an easy, cheap way available today to reduce the number of expensive AMP/REGEN huts your typical long-haul provider would require, it would have been found by now.
Dont believe anyone who tells you they can send a signal 4000km without regeneration. They are talking about electrical regen, not optical amplification.
Deepak Jain wrote:
Interesting. I have GigE stuff that have a receive sensitivity of -29dB,
if we could launch with +17dB somehow (this equipment only launches with
typical +4dB), things would get interesting.
With a .2 dB per km (I have seen this on newer fiber) that'll actually put
the length of fiber into the 200+km territory (230km according to my
math). Still, it's not 200 miles, and the amp to amplify +5 into +17 would
not be cheap, probably at least 10 times more expensive than a regen/amp
to put in the middle to go from -29 into +5 again.