Hmm. This group appears to be it for fiber on usenet? Hope I won't offend with content that may belong more in comp.lan.fiber (but it does not exist) or sci.fiber.basics (ditto - but I really think I've got the basics rammed in - still what I'm actually doing is pretty basic.)
Oh, and - Hi Phil. Hope the rack is working out.
Dragging a school LAN into the late 20th century. Balkanized (3 separate) copper networks. Trying to get to a gigabit (for now) fiber backbone all over campus. Starting from scratch, fiberwise. Looks possible that the powers that be are going to vote for in-house and less cash out of pocket, given the state of the economy and in-house labor that's not afraid to go there (I'm with H. Ford on that one - I think I can, so I'm probably right.)
Started out thinking I was going multimode and OM3 so I'd have 10Gig in hand for the future, then found $49 list price SM SFPs and the savings on fiber (a couple of kilometers, total) overwhelmed the "prohibitive price of electronics" (lighting at 1Gb in either case - will await lower prices before heading to 10 Gb) that was the assumption leading to MM in the first place.
Having no installed base to conform to and being on a LOW budget I have the following general ideas in place. Thinking to take my indoor/outdoor (no splices, no junctions) distribution cables to locked cabinets that will be large enough to house the electronics, rather than the typical "fiber patch panel wall box" or equivalent rack unit. Patching is not really an issue here - might happen once in 5 years. Thinking to put something like a CCS spider breakout on the end of the cable, along with
20 feet or so of service loop, fastened to the inside of the box, and skip the whole patch panel, patch cords, etc. expense - and the associated optical losses. Terminate to LC connectors, stuff into SFPs, turn distribution cable into long bundle of patch cords between my switches. SFP Xmit - LC - distribution cable - LC - SFP RcvI've had claims that if the fiber is 900um tight buffer I might not even need the breakout kit, but if it makes it more robust, it's not too much of an expense - on the other hand, the whole point of the locked box is to keep the ignorant out. I don't have a feel for how subject 900um is to expire when handled by non-idiots, and rarely. I'm now also looking at a 250um product, and there most of the available fan-out products (not all) are only getting it up to 900um - so is it worth the expense and extra shopping hassle of hunting down 250um to 2 or 3mm product, or is 900 um and not handling it like a doofus likely sufficient?
The benefits (if any) of a patch panel and patch cords come at a large cost for my application - more than the whole 2km of fiber, actually. The only time I see unplugging anything would be when 10Gb SFPs become affordable, or in the probably rare instance that one fiber goes bad and we need to switch to a backup fiber (where I think the more typical problem is all fibers got broken by a mad backhoe and I rent a splicer as fast as possible.) Other than that, it would be to replace things (SFPs or switches) gone bad. That's it.
After quite a bit of reading, and between the lines reading, and considering that I used to polish the ends of laser rods with a puck 20 odd years ago, I'm inclined to epoxy polish connectors. The labor is already paid for, the connectors are much cheaper, I have a good starting kit of abrasives courtesy 3M at the NFOC and there's nothing to go wrong later as there is with the various mechanical splice connectors everyplace is trying to sell as quick and easy. Nor do I have anything like the budget for a fusion splicer (even the rentals are terrifying), and that type of splice on connector is still far more expensive. One thing I do find confusing is that I can find mention of a clever Corning spring-loaded LC polish puck, and a part number, but I'm danged if I can find anyplace actually selling it, including places selling Corning connectors...
I need to do about 100 connectors - I figure I get 150 and plan to throw the first 10-15 away on learning curve issues, plus a loss percentage on the rest.
LC connectors appear to make the most sense with no installed base to conform to - it's what all the SFPs I want to plug directly into take, and they have generally lower loss numbers than the bigger connectors.
Thought - tips - tricks - advice - donations of piles of LC connectors or other goodies all accepted (private non-profit high school with tiny budget.)
Final observation from me - it's amazing how much you can charge for a $5 laser pointer when you call it a visual fault locator...