How Fiber Optics Work

Add Your Comments

Anyone who reads my articles knows that sooner or later I would write an article on this topic. This is a quick primer on how different optical fiber connections work. Optical fiber connections are going to end up in your home far sooner than holographic or first surface optical storage. I am not going to give a review of individual standards in this article but instead an overview of the functioning of the major technologies used today.

The first and most simple type of optical connection uses simple LED light.

By switching the LED on and off, you can produce the 1s and 0s of machine language. The problem with this mechanism is it takes a long time to switch the state of the light (at least, a long time by computer standards). This is still used in the optical connection for audio systems. The S/PDIF / TOSLINK on the back of audio equipment sends information by pulsing an LED on and off and sending the light via a plastic optical fiber to the receiver.

It is possible to get a glass optical fiber for this purpose but this is somewhat akin to using single malt scotch to run your car. It will work but it’s an unnecessary expense. The bandwidth on this type of system has long since been surpassed by copper wire. The reason it is still used in audio equipment is the fact fiber does not receive electromagnetic interference.

Next up is single channel un-amplified coherent light fiber optics.

This is the most common type of system in use today. A laser is used to transfer information over glass fiber. Rather than pulsing the laser electrically, these systems use an optical switch.

Lasers use coherent light which holds some advantages for switching. As you may remember from my holographic storage article, coherent light waves can be used to interfere with one another. This phenomenon can be used to make an optical switch.

First, split the light beam into two paths using a beam splitter of your choice. Recombine the two such that the two beams are in phase. This will be an ‘on’ signal – a digital ‘1’. Now extend or shorten one of the two paths by half of one wave length. The two paths will destructively interfere, effectively turning the laser off.

Using a MEMS (micro electro mechanical system), this can be done faster than switching the laser electronically. Furthermore, it is possible to place multiple switches in a single path. By staggering the timing at which they operate, it is possible to send more information than a single copper circuit can handle.

Thanks to the wave nature of light, it is possible to multiplex many signals to the same fiber. For instance, start with one single channel signal at 1500 nm. There is nothing stopping you from adding another signal – 1000nm for instance. Still more channels can be added at 656nm and 502nm etc.

At the other end, you can divide the signal out into its component channels with a diffraction grating. The number of channels you can add to a single fiber is a function of the quality diffraction grating and the bandpass window of your fiber (see first surface optical storage for a definition of ‘bandpass window’).

The quality of the grating determines how narrow a range of frequencies you can differentiate one from the other. It is simply a matter of dividing the bandpass window by the number of frequencies you can differentiate. With good fiber and a good grating, the theoretical limit can range into the thousands.

My favorite technology is by no means the fastest but it is by far the most elegant. An un-amplified signal is fine over a few kilometers, but if you want to cross an ocean, the glass will absorb your signal. The solution is to include a number of amplifier-repeaters. The punch line is you can’t use electricity. Copper absorbs electricity faster than glass absorbs light. You could use electrical amplifiers but most of your energy would go into boiling water of the coast of Labrador.

The solution is to use an all-light system.

Two lasers, one is an amplifier beam the other is the information beam. There are a number of amplifiers along the line. The amplifier beam powers these. The amplifiers use the energy from the amplifier beam to amplify the information beam. This raises the obvious question: Why not just use the amp beam to carry the information?

The simple answer is the amplifier beam is more than powerful enough to cut steel plate. It takes little imagination to know what will happen to the precision micro electronics of a switch.

The amplifiers themselves are light pumped solid-state lasers (not to be confused with semiconductor lasers). For the sake of non-physicists like myself, I will back up a little. Imagine a single neon atom in a neon sign. An electron will hit it and it will become energized. Something will then cause it to release its energy in the form of a photon. The quantum mechanics of neon atoms determine the exact frequency or wavelength of this photon. The cause of this could be another photon. This is called stimulated emission, the “SE” of laser.

If you have a population of atoms in an energized state, all it takes is one photon to cause a chain reaction releasing the energy of the whole population. One photon hits one atom and the atom releases the photon and each continues on to stimulate another atom and so on. The beauty of this is all the atoms release exactly the same frequency of light perfect lock step (in ‘Phase’ as physicists like to say). This light amplification is the LA of Laser. For sake of those interested, the R is for radiation.

In this case, that just means the stimulated emission produces light rather than something else.
This is important to all-light amplifiers because it is possible to use light of any frequency to energize the population. Once the population is energized, it takes a certain specific frequency to cause it to release its energy. When this happens, it releases its energy at the same frequency.

Thus, one frequency can be used to power the amplifier to the energized state without causing it to release its charge. When the carrier signal hits the amplifier with an ‘on’ signal, the amplifier releases an ‘on’ signal of many times that strength. Thus, we can send information across oceans with a readable signal strength at the other end.

Ian Anderson

Leave a Reply

Your email address will not be published. Required fields are marked *

Discussion
  1. just to give you an example, high end fiber can do up to 40 gigabits per second and perhaps maybe even more by now.

    DSL broadband uses a telephone wire connection that has the potential to broadcast at different frequencies at the same time, multiple channels will exist from anywhere from 25khz to 1100khz, this becomes divided up into 256 4000hz channels. If you can imagine that with a 56k modem, they were able to transmit 56kbs out of the first 4000hz, having 1100khz (or possibly more, depending upon the line) will allow much much higher speeds; however, theres absolutely no way you could get 40gb/s out of it without using 20 thousand lines or so.

    As far as cable broadband is concerned, the central office probably uses fiber optics to get out to your block, and then they use a regular old coaxial cable to get from there to your cable modem. Coaxial cables have a much higher potential speed than phone lines using broadband technologies. I can't say exactly, but i'm pretty sure they can operate at much higher frequencies than phone lines due to the higher quality line; however, usually your cable bandwidth is limited drastically because you are sharing the trunk with everybody on your block, and with DSL you have your own dedicated line, which may not be bandwidth limited, but is maxing out about where cable is limited, perhaps by software on the WAN devices or on the routers.

    Fiber optics is faster and much more stable than the other technologies because light isn't disturbed by the same kinds of forces that cause signal degradation in electric signals. You have to remember that home WAN connections we use run at like 256kb/s up and 4 megabit down, for cable, and roughly the same for DSL. Fiber connection going right up to your house that would give you theoretically up to 100 mb/s is a huge difference, placing your WAN connection theoretical speed in the same ballpark as fast ethernet.
    As always Ian has given us a very informative article. I'd just like to add to it from a different angle, that of the cable installer. The technical part of fiber I have little to no knowledge of, that's why I liked Ian's article. My little blurb is more about the actual systems that are being put in place. One note that I do have for Ian..although the theoretical is in the thousands, the actual is usually limited to around the equivalent of 100 twisted pair copper/fiber. I don't know why this is, but I'd suspect it's the engineers playing it safe.

    I work for an underground construction company. For the last few years we have been putting in fiber to the home systems in small rural towns in Minnesota and North Dakota. We have been installing 2 types of systems, pure fiber and hybrid.

    Pure fiber systems are just that, fiber right to the back of the house. The term is a bit misleading, as at the back of the house there is still a small fiber to copper converter. It's usually housed in a small (aprox. 1 foot wide by 1.5 foot tall by 3 inches deep) grey plastic enclosure which is mounted to the house. it draws it's power from the house electrical(a small price to pay for the speed).

    Hybrid systems are what's in use in my area. We plow in a mainline fiber to a junction box. These juction boxes are located to keep as many of the drop cable coming from them to less than 2 miles. They aren't as capable as the pure fiber systems, with a maximum DSL speed of around 3 Mbits/sec. They are a lot cheaper than the pure fiber systems, as all the electronics are in one box and can be more easily maintained. Also the copper drops to the house can be spliced in less than 10 minutes in the event of a cut, where as a fiber pretty much has to be totally replaced.

    One day I'll post up a pictorial showing the types of equipment that we use and the different types of cables as well.
    What is the location of the article?
    I've designed a low noise EDFA as an undergrad. Mass usage of Optical repeaters/amplifiers is by no means a small investment. The sources and detectors are cheap L,C and S band lasers, but other devices such as test & diagnosis equipment cost a pretty penny. All such costs add up.
    Interesting article and great post. I was particularly interested because I have FiOS, and it was intriguing to see how all of it worked out. For those intereseted, I saw a very informative documentary on Fiber Optics on Nova awhile back, I think it was at the end of summer when I saw it. I'm sure if you do a quick google search you could find out what the name was and probably find a TV stream on the internet.
    I learned a few things from the article that I didn't know before (long-haul amplification methods mainly) I mainly have experience with fiber in the LAN and a little bit of MAN, where I work we run fiber to everything in the lan, all computers, servers...everything, I like running it better than copper...easier to route, don't have to worry about EMI etc. I might be a bit biased but I actually like running and terminating fiber more than copper (especially shielded cat6, that stuff is pure evil)
    by necessity...:( . Was extremely stupid and not wearing safety glasses when I should have been....Down to 20/50 in the right eye. Oh well, that's how the cookie crumbles.
    sorry it's taken me so long to post back on this(damn eye surgery). In answer to some of the posts above...

    Fiber optic cable is actually cheaper per foot than copper. The labor to install it is the same as well(except where the telco spec demand deeper than 36" cover). The cost isn't in the cable but the conversion equipment. One DSLAM converter can run upwards of $100k. The boxes we are now installing at the houses are aroud 5 bills each. By comparison, a 100 pair copper cable is around 70 cents a foot and a 4 fiber(about the same as 100 pairs in reality) is about 67 cents a foot. The labor cost is somewhere in the neighborhood of $1 a foot, if I'm remembering correctly.

    We did a town called Carrington, ND recently. Something like 70 miles worth of fiber in all, and a total cost of aprox 3 million. Most of that was in the fiber to copper converters at each house(roughly 1300 of them if memory serves).

    To give an idea of time frames and such....

    We ran with a 16 man crew 5 days a week(4 1/3 counting travel to and from the job). The job was 70 miles/aprox 1300 residences. Total time...16 weeks. That was balls out fast as we could go. the town had relatively few utilities underground to contend with(power in areas, gas in areas, and cable TV). In a larger town(oh let's say 15000 residences/business) there would be gas, phone,power, cable TV, and possibly lots of storm drains. Each of these is an obstacle to getting from point A to point B. Even with the best equipment and men possible...I'd estimate that to be a 5 year job. Estimated cost to the telco would be around 15 million...easy.

    On the political front....well that is ruled by a little thing called R.U.S.. these are the guys in charge of getting supplemental funds out to the rural carriers. They are also the ones who decide what the minimum specs are for those systems(hint hint...these are what keeps the prices high).
    You got that almost right.

    Read Deadbots post (post #1) he works laying fiber or hybrid in RURAL Minnesota and N. Dakota..

    It is hard if not impossible to get more rural in the US then the two states he listed.

    I live just south of DC and used to lay Fiber for Comcast and Cox. The problem isnt that it would be too expensive to lay the cable, the companies already did that. They arent willing to spend the $$ to enable everyone to have business grade (in the US) speeds because they know that not everyone will pay for it. The US is falling further and further behind the Internet connected world due to the monopolizing power of the cable/telco's. I have to stop here or it will get political, the problem isnt distance, it is lobbyists paying and paying well to make it sound much more complicated.


    I read all the post. However my point was that to lay the fiber from the main channel to our homes is a big cost. Now that might not be the reason we don't have it, and I agree that the damn telecos monopoly is not helping. November 4th the repubs lost the house\senate power so we will see what happens.
    The reason the entire US is not laid with fiber to our doorsteps is because a majority of our land is rural and it would cost a fortune, fiber optic cable is expensive.. In NYC, LA, all the big cities where the money\commerce is at, is laid totally with fiber. The rest of us sheep are at the mercy of our ISPs. The only way to lay the entire US with fiber in the immediate future would be to make it a government job, and we don't want that. We are speaking on a computer enthusiast forum so of course we all want 50mbps internet, but realistically, most Americans are content with 3mbps DSL\cable. MAke a big enough demand for it and companies will bite and give us fiber.


    You got that almost right.

    Read Deadbots post (post #1) he works laying fiber or hybrid in RURAL Minnesota and N. Dakota..

    It is hard if not impossible to get more rural in the US then the two states he listed.

    I live just south of DC and used to lay Fiber for Comcast and Cox. The problem isnt that it would be too expensive to lay the cable, the companies already did that. They arent willing to spend the $$ to enable everyone to have business grade (in the US) speeds because they know that not everyone will pay for it. The US is falling further and further behind the Internet connected world due to the monopolizing power of the cable/telco's. I have to stop here or it will get political, the problem isnt distance, it is lobbyists paying and paying well to make it sound much more complicated.
    The reason the entire US is not laid with fiber to our doorsteps is because a majority of our land is rural and it would cost a fortune, fiber optic cable is expensive.. In NYC, LA, all the big cities where the money\commerce is at, is laid totally with fiber. The rest of us sheep are at the mercy of our ISPs. The only way to lay the entire US with fiber in the immediate future would be to make it a government job, and we don't want that. We are speaking on a computer enthusiast forum so of course we all want 50mbps internet, but realistically, most Americans are content with 3mbps DSL\cable. MAke a big enough demand for it and companies will bite and give us fiber.
    Good article. Except for a few of the technicals, it did not really tell me anything I did not already know however.

    You know ... what frustrates me is how the telcos and cable providers were promising ultra fast fiber optic connections to our homes 8-10 years ago. They even mentioned on demand streaming video capabilities. To this day none of that has come to fruition.

    If you go back to news reports in the mid to late 90's you'll see how the telcos were promising fibre service with live streaming video capability to residential users "within a few years". That is the main reason Congress approved a complete restructure of the way telcos charge their customers in the early 90's. Used to be they were limited in what they could charge. The premise was the extra profits were to go into updating to a ultra broadband infrastructure. Well, where is it? When the telcos & cable providers start rolling out single mode fiber at 2,400 Mb/sec to our homes, let me know. :shrug:

    What I find particularly disturbing is how the USA has fallen to 12th place worldwide far as average internet connection speed goes. And we are only going down on that list.

    There was a very interesting program on public television about this very subject a few weeks ago. Wish I captured it.

    In S. Korea, the country with fastest average speeds, they enjoy over 45Mb/sec. And >200Mb service is readily available in many S.E Asian countries. On demand streaming video is almost commonplace in major cities. At the rate we are going that won't be available in the US for many many years. This is directly converse to what we typically like to think - that the US is on top of the technology game. Really makes one pause to think what is going on in this country.