Slashdot Mirror
Bent Fibers Put Networks At Risk
opticsorg writes "The combination of moderate optical powers and tight bends can prove catastrophic for optical fibers, according to research carried out by BT Exact in the UK. Although the effect is unlikely to cause problems in current networks, it means that designers may need to think carefully before scaling up the power in their systems or deploying Raman amplifiers with pump powers of several hundred milliwatts or more. In the July 10th issue of Electronics Letters, Ed Sikora and his colleagues report that powers as low as 500 mW can induce permanent damage in singlemode fiber that is bent (13 mm bend diameter or less). 'These bends could be found in exchange racks or splice trays, for example, especially if a fiber is tugged or pulled,' Sikora told Optics.org. The BT researchers carried out tests on four types of fiber subjected to a range of bend diameters (5 to 15 mm) and optical powers of up to a few watts. In all cases the fibers fail within 53 hours. 'What was unexpected was that the catastrophic failure can occur in 90 bends at fairly low powers of less than 1 W or so,' said Sikora. 'It's important to understand that we're not saying that networks are going to fall over tomorrow, but as powers go up you have to aware this effect could occur under certain circumstances.'"
Im' no expert in this technology, but is it possible that somebody could remotely set up the power of the transmitter in such a way to produce this kind of damage on purpose.???
I don't know how the average fibre installer works today, but I know the few times I played with it, we always installed with corners were gentle enough that a full loop would be about 30 cm. This included ensuring no significant load on the fibre at the attachment points, so no 90 degree bends at the switch or server.
I'm only talking about the last few feet, not the 'last mile' of course, but if I upped the power and had a fibre failure, I'd be saying very rude things to the rep of the company that did the installation (if they survived the .bomb, of course).
... the "new" thing being reported is the microbends fail by going opaque when higher optical powers are being transmitted in the fibers. For modern systems in most inter-city networks, the number of channels (40, 80, ...) is going up, as is the power per channel. This is a combination not seen earlier in installations where most fibers (bent or not) carried fairly low power signals.
Interestingly enough, microscopic dust particles are equally hazardous to the system's health at these high power levels. Dust particles caught in unclean connectors has been known to scatter enough power to fuse/weld (its those friggin laser beams) together the connector parts together.
yawn.
yes, 42.
Optical fibres can be used as couplers between two lasers (the second laser amplifies the signal from the first). We use such a system in our lab, where average powers of 40-100W can be sent down a single fibre (multimode in our case). If the surface of the end of the fibre gets scratched, or if dust lands on it, the tip can explode. With each pulse (it's a 25kHz pulsed laser) another piece of fibre is destoyed, and it acts like a fuse. If you don't turn the laser off quickly you can soon lose kilometres of fibre. All that's left is a ringing in your ears and a few bits of scorched plastic.
Does anyone know of supplemental cladding (preferably something more sophisticated than a thick layer of duct tape) that can be added to the cable at critical points to prevent excessive bending while still allowing a reasonable degree of flexibility?
Please donate your spare CPU cycles to help fight cancer and other diseases
In grad student (chemistry) days I ran an apparatus for stimulated, mass-selective Raman spectroscopy of molecular clusters. The Raman pump beam was two colors, generally tuned for power (a watt or two each, depending on tuning). The molecular clusters were formed in a vacuum chamber and we had a quartz window to let the laser light in. If there was a speck of dust on the window at the point the beam entered, the absorption was sufficient to start drilling a hole in the window. The noise was our cue to cut the laser beam before the window was breached (there were dedicated electronic circuits to protect the vacuum chamber's diffusion pumps, but we didn't want to take the risk of failure).
Of course, we spent a lot of time cleaning that window, and for that matter all the optics.
The clearance system sounds logical. It is not. It is completely arbitrary. -- John Bolton
Nah. The maximum radius doesn't depend whatsoever on the intensity of light passing through the fiber. If the radius is too tight, light will leak out, regardless of how weak. Light is already leaking out of these over-bent fibers. The problem is when the power gets too high, and the amount of leaked light becomes so great that it actually starts heating up the cladding.
BTW, the maximum curvature radius you can use depends on both the material the fiber itself is made of, and the material the cladding is made of. You want the two materials to have dissimilar indices of refraction -- the more dissimilar, the tighter you can bend the fiber without light leaking out of it. To some degree it also depends on the frequency of light you are using. But it does not depend on the intensity of the light.
Fortunately for them, there are tools that not only tell you how much signal loss you're experiencing in a given cable (so you can replace a lossy cable) but also tell you where a significant problem exists along the cable. Very useful for underwater and other long links - you have a good idea, to within a few cm, where the problem is. Go to it and splice - but, of course, only if the splice is going to be significantly less lossy than the problem itself.
-Adam
you need higher power to get a higher data rate (this is, after all, why the powers keep increasing)
In the applications for which I design fiber optic transceivers, the data is always DC-balanced. The duty cycle is not rate dependent--it's always approximately 50%. Thus, the average power also does not change with data rate.
You don't need higher power for higher data rates, per se, either. You need higher power if you are trying to go the same distance at a higher data rate.
I believe what they're saying is 'although we always knew that bending your fibre would result in less efficient connections, we're now finding out that it will ALSO actually damage the fibre itself over time.'
Much like, kink a CAT-5 enough, and it won't pass traffic at full speed, but it's not going eventually burn the cable.
Vintage computer games and RPG books available. Email me if you're interested.
Today, normal enterprise/campus networks don't work with any amount of power, quite honestly. Horizontal building cable is almost exclusively copper or multi-mode fiber, and riser/inter-building cable is single-mode, but relatively cheap stuff. Where this kind of thing comes into play is in the long-haul networks of companies like AT&T, Level 3, Sprint, etc., where you have 100Km+ between OA (Optical Amplifier) sites.
:-)
Many people are working to extend the OA interval to 600Km through doped and Raman amplifiers, which are giving you launch powers in the 30db+ range, and are starting to approach the powers that can do this. However, as someone pointed out, none of this happens with normal correct fiber installation. I know my company, which runs a large (tens of thousands of miles) network has reams of paper describing exact splice tray designs, stress on cables, bend angles. It goes down to how you support things going in and out of a OA, etc., and addresses the radius, which I believe we try and keep around 15-20cm minimum.
If you follow smart rules, these don't matter. If you don't, well, it probably won't affect anyone who is working outside the large telco space. The cost of an EDFA (Erbium doped fiber amplifier) is tens of thousands of dollars.
No story, move along.
It's the schmucks and management that are cutting corners that are going to get bitten. -- What this world needs is some geeks with the backbone to stand up for what they believe in.
If this is a sig it's great, but particularly appropriate for this post. In every data centre I've been in, there have been giant cable infrastructure nightmares - everyone knows about the monsters under those raised floors, but have you ever seen a rats nest of cables so large that it has to be supported or the weight will take down the racks attached to it? Rats nests you have to cut through to clean up? Yuck. These are the circumstances which lead to bent fibre. In most situations, management is hard pressed to cough up cash to clean it up unless it poses an immediate risk to the business. Most of us don't make the time to put together the proper business case to prove to the overpaid overlings that these are important issues that need to be addressed!! Hopefully the report opens a few blind eyes and helps to clean up a few data centres!
The article states that 4 different types of fiber were tested, but doesn't state what those types are, or related too. There are different materials fiber optic cabling can be made out of, as well as different diameters. Were they talking about different diameters, or the materials?
This was obviously written for a somewhat technical audience, given the subject matter and source that published it. By omitting the facts of which 4 types they tested, it really doesn't do justice to the subject. For all we know, they tested low end cabling made from plastics.
A slip of the foot you may soon recover, but a slip of the tongue you may never get over. -Benjamin Franklin
I design and build fiber-coupled semiconductor lasers as a day job, and some of the stuff in our R&D lab has a significantly higher power than what is currently used in most systems out there. A fiber bend radius that leaks/absobs x% of the power at 10mW with no difficulty becomes dangerous when you put a 5W laser in the system.
I used to do that job. Now I don't have a job and the site where I worked is for sale.
One of our packaging designs for fibre coupled semicondutor lasers was quite old but worked OK. As we ramped up the power of the laser chips we found that the light which wasn't coupled into the core would cause physical distortion of the packaging and move the fibre tip* due to the ammount of heat absorbed by the fibre-cladding to package interface. We fixed it after a fashion for one generation and redesigned the whole package for the next one which was due to see nearly 1W of 980nm ex-facet.
* You can guess what happens to power vs drive current linearity when the coupling ratio changes with facet output power.