Real-world 1000baseZX distance
Bookmark this page:
 Yahoo!
 Google
 Windows Live
 del.icio.us
 digg
 Netscape
|
|
||||||||||||||||
|
Posted by Daniel J McDonald on March 31, 2007, 5:32 pm
Please log in for more thread options I work for a power company that runs fiber optic commincations on transmission lines. The fiber is contained in the "static" ground wire on the top of the transmission line (above the three conductors). We use this for both ethernet (1000Base[LZ]X) and Sonet (OC-12 and OC-48). We are planning a 56 mile run. The fiber team expects to have at least 55 splices, and expects a best-case 22dB loss along the entire span. Will a normal "ZX" GBIC reach that far? What can we do to improve our chances? For example, are there "better" types of singlemode fiber that we can run instead of the 9nm core we normally specify; or, perhaps, would I be better off using a CWDM gbic of a particular color instead of 1510nm? -- Daniel J McDonald CCIE # 2495, CNX Visit my website: http://www.austinnetworkdesign.com | ||||||||||||||||
|
Posted by on April 4, 2007, 8:26 am
Please log in for more thread options On Mar 31, 4:32 pm, djmcd...@fnord.io.com (Daniel J McDonald) wrote: > I work for a power company that runs fiber optic commincations on transmission
lines. The
> fiber is contained in the "static" ground wire on the top of the transmission
line (above the
> three conductors). We use this for both ethernet (1000Base[LZ]X) and Sonet
(OC-12 and OC-48).
Howdy Daniel, Sounds like a very interesting installation! > We are planning a 56 mile run. The fiber team expects to have at least 55
splices, and
> expects a best-case 22dB loss along the entire span.
chances? For example,
> > Will a normal "ZX" GBIC reach that far? What can we do to improve our > are there "better" types of singlemode fiber that we can run instead of the
9nm core we
> normally specify; or, perhaps, would I be better off using a CWDM gbic of a
particular color
> instead of 1510nm?
The short answer is that some GBIC's might reach that far, others may not. And yes, there are better (low-dispersion and/or low-loss) fiber to be had. The longer answer is that as you indicated, there are other factors in how long a span can be besides the "reach" that the optic is "rated" for. Not only do you have the particular model of optical module (with its power-output level, extinction ratio, receiver sensitivity and dispersion tolerance), but you have the fiber affecting the signal as well. I'm not an expert in that area, so for an application like this, I'd recommend doing more research... but for installation that would be as difficult as this one to work on, it might be worth researching something in like non zero-dispersion-shifted fibers (NZ- DSF).You have an advantage that you aren't trying to do 10G (at least right now)... so that helps the dispersion side of things a great deal. On the module front, you might have better luck with SFP's. Since that is where all the growth is, optical vendors have put more R&D into further reach parts (you can find 120 km optics if you look hard enough). I'd be very tempted to set up the exact fiber, with splices and connectors, if at all possible. I would also ask the fiber vendor and optical equipment vendor for advice. I'd hope that they would provide such advice for free - I expect some do and some don't. Just don't let them tell you that power level is all you need to worry about. Good luck, Marc --- From address is fake - please reply to the newsgroup | ||||||||||||||||
|
Posted by Dmitri(Cabling-Design.com on April 4, 2007, 1:14 pm
Please log in for more thread options
Daniel J McDonald wrote: > We are planning a 56 mile run. The fiber team expects to have at least
> 55 > splices, and > expects a best-case 22dB loss along the entire span. > Will a normal "ZX" GBIC reach that far? What can we do to
> improve our chances? > For example, > are there "better" types of singlemode fiber that we can run > instead of the 9nm > core we > normally specify; or, perhaps, would I be better off using a CWDM gbic > of a > particular color > instead of 1510nm? You may get better results using non-zero dispersion shift SM fiber (alternating positive and negative shifts). BUT Considering the cost and the efforts involved in the project: I would NOT based my design on an advice I picked up on Usenet, no matter how valuable. Pick two vendors you'd go forward with - one for the fiber equipment and one for the active Ethernet equipment and have their engineers provide you with a design. In fact, you might already be using someone's equipment preferentially - so why don't you just give those guys a call and make them come up with a design they will stand behind in case you run into any unexpected issues there? -- Best Regards, Dmitri Abaimov, RCDD http://www.cabling-design.com/ Home Cabling Guide, Cabling Forum, color codes, pinouts and other useful resources for premises cabling users and pros -- +----------------------------------------------------------+ | http://forums.cabling-design.com/ | | *** a better way to USENET *** | | no-spam Web and RSS interface to your favorite newsgroup | | comp.dcom.lans.ethernet - 5567 messages and counting! | +----------------------------------------------------------+ | ||||||||||||||||
|
Posted by stephen on April 5, 2007, 4:41 pm
Please log in for more thread options
> Daniel J McDonald wrote:
> > > > We are planning a 56 mile run. The fiber team expects to have at least
> > 55 > > splices, and > > expects a best-case 22dB loss along the entire span. >
> > Will a normal "ZX" GBIC reach that far? What can we do to
> > improve our chances? > > For example, > > are there "better" types of singlemode fiber that we can run > > instead of the 9nm > > core we > > normally specify; or, perhaps, would I be better off using a CWDM gbic > > of a > > particular color > > instead of 1510nm? >
> You may get better results using non-zero dispersion shift SM fiber > (alternating positive and negative shifts). > > BUT > > Considering the cost and the efforts involved in the project: I would NOT > based my design on an advice I picked up on Usenet, no matter how > valuable. Pick two vendors you'd go forward with - one for the fiber > equipment and one for the active Ethernet equipment and have their > engineers provide you with a design. In fact, you might already be using > someone's equipment preferentially - so why don't you just give those guys > a call and make them come up with a design they will stand behind in case > you run into any unexpected issues there? > agreed. also - see if you can get the number of field splices reduced - splice losses may well dominate the link budget with so many. I have seen similar o/head fibre with 4 Km cable sections. >
Regards
> -- > > Best Regards, > Dmitri Abaimov, RCDD > http://www.cabling-design.com/ > Home Cabling Guide, Cabling Forum, color codes, pinouts and other useful > resources for premises cabling users and pros > > > > -- > +----------------------------------------------------------+ > | http://forums.cabling-design.com/ | > | *** a better way to USENET *** | > | no-spam Web and RSS interface to your favorite newsgroup | > | comp.dcom.lans.ethernet - 5567 messages and counting! | > +----------------------------------------------------------+ >-- stephen_hope@xyzworld.com - replace xyz with ntl | ||||||||||||||||
| Similar Threads | Posted |
| Real-world 1000baseZX distance | March 31, 2007, 5:32 pm |
| Ethernet requirement for Data rate:100 Mbps, Distance:2-3 km | September 12, 2005, 5:54 am |
| The best distance covered by a transmit power booster Linksys WRT54g | October 9, 2004, 4:14 pm |