I installed fiber to connect two pv inverters at home, over a distance of 120m (to avoid equipotential bonding, through an empty conduit underground). On the one hand, fiber is a nightmare compared to Cat 7a - you have to handle it carefully, there's a pulling grip attachment that's too thick to get through most wall holes, and you can't pull too hard or the fiber will stretch beyond a certain limit and degrade. On the other hand, our cable withstood a lot of force, was bent at 120° angles, and pulled through all kinds of dirt/water without damage. I just wish that I could crimp fiber ends myself to avoid all the hassle, but the equipment is just too expensive.
It's also interesting that 25m of pre-crimped cable costs almost the same as 125m, because the crimping seems to be the expensive part, not the cable material.
You can buy some very strong fiber. My home fiber internet setup was done with some seriously good stuff. The installer left an unneeded chunk and you can't cut that with scissors without destroying them. It's some sort of outdoor rated, Kevlar reinforced thing.
For LC connectors, the two cables are held with a plastic clip, you can remove that to make it fit through smaller holes. After that it's much thinner and nicer to work with than Cat7.
The most common last-mile fiber assembly I have seen in use is called flat-drop. It is flat, and it is optimized for use on the drop to the customer premises. There are two (fiberglass?) rods and between them runs a loose tube fiber bundle with typically 6 fibers in it. Good picture here: https://www.novalight.com/Drop-Cable-Fiber-Optic-Flat-Toneab...
The fiberglass rods provide excellent crush and cut resistance, as well as bend radius control.
Our ISP installed fiber too late in the year to bury the cable, so it just laid on the ground for 7 months until thaw, right across our walkway and yard. They weren't even worried about it getting damaged.
We had fiber installed at our house and our derpy dog chewed up the line twice before they were able to bury it. Was a good time indeed and thankfully AT&T didn't charge us anything to come out and fix it each time.
They only ended up burying it a few inches deep, i was surprised and thought it would be deeper.
What surprised me the most is one of the techs telling me that ants are one of the most common threats to fiber lines outside of them accidently being severed by construction crews digging.
And my ISP had to replace mine multiple times due to lawn maintenance guys running it over, neighbors (where the line was pulled from) cutting it, etc.
I don't think I'll ever do a fiber install again. If it has already been run to the house, great, but otherwise, I'd rather not.
> For LC connectors, the two cables are held with a plastic clip...
Be careful doing that. There are LC connectors that are not capable of being split. (I know this because I attempted to split one of this type and damaged a fiber assembly at.)
The vast majority of LC connectors I've worked with can be split but some cannot.
They have special jackets on the fiber with 2 metal wires on each side. You would pull the metal wires not the fiber directly. Then you would polish the fiber and connect to the ont and turn it into an electrical signal (cat6 Ethernet). They say the max bend radius is a little larger than a coke can.
> It's also interesting that 25m of pre-crimped cable costs almost the same as 125m, because the crimping seems to be the expensive part, not the cable material.
When I was doing fiber quoting you would see this with some of the bundles, e.x. 72 vs 144. Cost wasn't doubled per foot like you'd think, it was a little less.
For the last ~decade, NANOG [1] has regularly had a (two hour) presentation/tutorial called "Everything You Always Wanted to Know About Optical Networking"; most recent one:
Related question I had for a looong time: how do submarine cable layer ships test the cable when they attach a new section? Like, these cables need many kilovolts of voltage to supply the inline repeaters so you can only power them when the cable is fully laid, but how do they verify if the fibers aren't broken somewhere along the path?
But after a repair they are probably tested on deck by re-energizing the cable from the landing stations. Although, they might behave differently when on deck than on the bottom due to the different dielectric constant of water and air.
Naive question - are these fibers pure glass, or some sort of composite? Are they flexible because of how thin they are, or is it an inherent property of some sort of composite?
The article mentions that these are just silicon dioxide, but my brain has trouble connecting the rigidity of glass as I know it with the flexibility of fiber.
"Regular" pure glass fibers would be nearly as flexible. As an analogy - the same way that a solid steel rod would be strong, but if you make steel fibers that are very thin they can easily be shaped and bent. The material properties did not not change, just the difference between having a "lot" of something vs very little.
In this case, dopants are added to the glass to improve its properties as a fiber optic cable and the processing itself is important to attain certain properties. But even without those added dopants regular silicon dioxide glass would make flexible glass fibers.
Composite fiberglass materials aren't flexible because of their composite structure, the layers and resin are there to increase strength and stiffness of an otherwise extremely pliable material. Think about fiberglass cloth or pink fiberglass insulation, glass strands get very flexible as the diameter decreases.
Let’s hope this multi billion dollar initiative isn’t just absorbed by private companies.
I can’t find the previous initiative but there was something like this in the past where “broadband” internet was to be accessible. Miles and miles of wires laid down but most of it wasn’t used. Benefited the contractors though.
Fiber ISP came to my town years ago. I live within a few miles of town on a major road yet I still don't have fiber. I call every few months but they just say its not available at my place.
I'm building a cluster of 128 GPUs. 16 chassis. Each chassis has 8 NIC's in it to do 400G into a single switch. This has ballooned into 384 Singlemode Fiber cables and 64 Multimode Fiber cables to connect it all together.
Does someone make, well, low speed fiber? Something to carry Fast Ethernet but that can't be susceptible to EMI and doesn't require the expensive and power hungry stuff that the real deal does?
You're not looking for different fiber really. You're looking for different transceivers.
There are fiber ethernet transceivers for all the ethernet speeds, I've seem them for 10M, 100M, 1G, 10G, all the Gs.
I feel like SFP (1G) is the sweet spot for cost and applicability right now, you can pick up a media coverter with one port rj45 and one port sfp for $20 from a reputable (consumer) brand, and SFP transcievers aren't too expensive. Switched with a bunch of 1g rj45 and one or two sfp are also inexpensive new and lots of inexpensive old enterprise gear at those speeds too.
For 100M (fast ethernet), I don't know if you'll have many options other than old old used enterprise equipment.
The biggest cost is in the opto-electric transition parts. The mechanical alignment needs to be precise, it needs to have a laser, a laser driver (most likely VCSEL), and a photodiode with a transimpedance amplifier. It's expensive because these parts cannot be integrated with the electronic chips easily. They are also bulky.
There are initatives to use crappy alignment optics and plastic optical fiber (POF) for short reach Ethernet [1] though these aren't commercially available. They don't aim for data center or consumer market but automotive and aerospace .
Probably useful if one end is 100m and not simple to upgrade. Optical transceivers don't usually negotiate on the fiber side, so you'd need a 100m transceiver on the other end.
I use a pair to isolate my cable-modem from the rest of my network, to the extent possible. I once lost a modem and a Unifi USG (thankfully only that) to a lightning strike. Now with the media converters, hopefully it's only a modem and a converter.
I worked on this for a while, years ago at Corning. We wrote control software for the vertical furnace that consolidated the glass before it was pulled into fiber.
I once got to do some QA lab systems work at a fiberglass plant. Fascinating place.
The molten glass in giant hoppers is gravity fed through very expensive rhodium-plated platinum blocks, then spun into a plethora of different products and spooled or chopped. Their downstream customers would use that as the raw material for end products.
I can't read the article because of the paywall though.. not sure the nuances between fiber optics and general fiberglass products.
If I remember correctly, the first doped fibre manufacturing, or maybe the first fibre manufacturing, used an old peening drop tower (https://en.wikipedia.org/wiki/Shot_tower) for the gravity feed. I guess they are called "drawing towers" now.
It's also interesting that 25m of pre-crimped cable costs almost the same as 125m, because the crimping seems to be the expensive part, not the cable material.