Affordable, Homebrewed Optical Networking?

Graham Wheeler asks: "Lately I've been obscessed with grassroots community network projects, and the hardware that enables them. Most sites I have seen focus on wireless RF networking, but I have noticed a few projects revolving around free space optical transcievers. Twibright Labs' RONJA is a good example of what I'm talking about. Not being an electronics hobbyist, however, makes the various plans for building a comm laser from scratch look rather daunting. It seems to me that it would be easier to just make a lens and housing system into which would go one of the many cheaply available copper-to-fiber media converters. Then you could simply modify it so that the laser ports were optically connected to the TX and RX lens assemblies instead of the standard fiber interface. So, what factor(s) am I overlooking that would explain why nobody seems to be doing this?"

Comments

interference..

[martin]

Birds, snow, rain fog etc...

Can fun to work around...

Things to ponder.

[Glonoinha]

At first I thought you were interested in a free range optical system, similar to ... say the IR connectivity between laptops or the newer HP printers - but with the second part of your question that changes towards 'using fiber optics in any fashion possible.' When I did something along the lines of free range (line of sight) transmission of data over a beam of light the problem I ran into was focusing the beam of infrared light. Tried using an IR LED and receiver pair but couldn't see the light to debug it or focus it. But hey, that was just my first attempt. If you are trying to convert the signal coming out of your Ethernet card for use on fiber optics that is easy. Get an Ethernet card with an AUI port out, and buy a AUI -> Fiber dongle. Cost about $50 to $100 at blackbox last time I checked (which was three years ago.) The real problem will be finding a network card with an AUI port on the back (looks sort of like a game port) - try looking at 5-8 year old hardware from SMC or 3COM. Odds are it will be ISA, and not plug and play so it helps if you know how to assign IRQ's, memory addresses and DMA settings manually (tip : IRQ 10, Memory Address 300, DMA 1 - this generally works if you are not using a SCSI card in your computer.) Glonoinha

Re:Things to ponder.

[cdrudge]

UTP to fiber converters are also available so old outdated NICs with AUI ports are not required. They are usually a little more expensive (around 140 from Black Box).

Power, interference, coupling.

[Yarn]

Telecomms operates around 1550nm, because that's where fused silica fibre is the most transparent, and has low dispersion. We can see up to about 740nm.

Note that in ideal conditions (perfect lens etc) light diverges at an angle related to its wavelength and the minimum point, the longer the wavelength, the faster it diverges.

The light coming out of a typical FC connector diverges at about 60 degrees and the beam is pretty crap*. Even coupling between two properly cleaved fibres with a sub micron air gap loses more than half the signal.

To get any signal out of the noise out there you'd *have* to use a lock-in amplifier. These are not trivial to make, unless you have training and/or experience. These 'lock-in' to a regular signal, you'd have to modulate the laser to this signal, and the data on top of that modulation.

What you want is a laser which outputs a TEM00** beam, a collimator with a large lens, a good external light modulator and a lock-in amp. At each end. Then you have to modulate the laser with carrier and data, from some kind of device. (Start with a serial or parallel port, nice and easy)

* Technical term. Take too long to explain :P
** Transverse Electromagnetic Mode 0,0: the most basic gaussian mode of a normal cavity laser. The easiest to focus, as it is a self-fourier-transform. Unfortunately the more power you want out the more likely it is you'll get power in the other, less clean modes.

If you can wait.

[Anonymous Coward]

I used to work for Honeywell opto-electronics devision and I'm sure I've still got a load of crap at home.

Expect an attached post with a reasonable explanation of the problems/costs involved, performance and a few hints and tips about doing it on the cheap sometime soon (over the weekend i exprect).

My company does this

[MrResistor]

...for agricultural telemetry and irrigation control. What we use is lower bandwidth than what you want (RS485 over ifrared), but the basic principals are similar.

The biggest problem we have is reflection. A unit sending data gets back a reflection of that data off the receiving lense. If you're just using 2 stations, it's not that big of a problem to scan the buffers for reflections, but we've found that with several units (usually all communicating with a central base-station) collision becomes a serious issue, and detecting/masking reflections is extremely difficult.

Hope this helps. Good luck on your project.

why bother?

[Anonymous Coward]

everyone knows SNEAKERNET is more than anyone will ever really need in a HOME environment!

The answer to his question.

[cybermace5]

No one is actually answering this guys's question, which is "why can't we modify a commercial fiber transceiver to work in freespace."

I'd say that it isn't impossible, but the work/yield ratio would be too high. Fiber tranceivers are designed to output into the end of a fiber, and everything in them is designed to do that as efficiently as possible. The optics are so tiny, and to get a beam in or out might require some microscope work.

Even if you did this, the odds are it would only work for a short distance. Fiber lasers only have to project onto the end of an optical fiber, which is naturally a very small area. In order to achieve a high enough power density across the end of the fiber, the lasers don't have be very powerful at all. You'd end up with a beam so tiny it would be nearly impossible to aim. And if you tried to use collimating lenses, the laser isn't powerful enough to begin with, and most of the power would probably be lost in the optics.

Most freespace lasercomm projects involve something a bit more powerful, which allows the laser to punch through smoke and fog a little better. I guess if you made the laser powerful enough, it could punch through interference such as pigeons too.... :-)

I think that off-the-shelf optical transceivers could have a use in freespace comm, but I think the most useful work would involve somehow using the tranceivers as modulators for a more powerful laser. Perhaps you could have an optical fiber running to the big laser and receiver optics, like a repeater?

Re:The answer to his question.

[Anonymous Coward]

I assume the goal here would be 10mbit ethernet, which requires, in theory, 10MHz signalling.

If you were to take the output on one of these fiber transceivers, hook it up to a laser pointer (through a circuit that could handle the current), and a suitable photodetector on the other end, would it work? Can a diode laser pulse at 10MHz?

Re:The answer to his question.

[cybermace5]

Diode lasers can easily pulse at 10MHz. The problems start when you want to use laser pointers.

Laser diodes are very finicky components. It only takes a tiny bit of overcurrent to ruin one; this is because laser diodes are operated close to the physical limits of the material. Some of the first LEDS would lase if you ran just enough overcurrent through them, but it was next to impossible to control the current precisely enough to avoid frying them. Laser diodes have a detector integrated directly into the laser, which allows control circuitry to lock in the diode to a current level that allows it to lase but doesn't fry it.

Laser pointers became cheap and popular once the method for integrating a detector was perfected, and laser current controllers became widely available. The problem with modulating one of these is the controller; they were designed to hold the laser at a continuous level. It is possible that whipping the controllers with a 10MHz signal will keep them from being able to handle the current properly, and either fry the diode or not allow it to light at all.

This is why tranceivers have their own current controllers embedded in the modulation circuitry. Some diode lasers (not laser pointers) have modulation inputs for communication purposes. I'm not aware of what the most popular laser is for these experiments.

Re:The answer to his question.

[Anonymous Coward]

Thanks for all the information. So I guess the short version of it is that a laser diode can definitely work, but not the variety typically found in a $20 laser pointer.

Can one obtain special diodes that don't suffer from this current control limitation, and for a reasonable cost? And do they have favorable divergence characteristics?

Since you seem to know a bit about lasers, could this work with any of the larger lasers typically found in undergrad physics labs?

Needs to be as cheap as RF, not just affordable

[hamjudo]

The 802.11 stuff is pretty darn cheap, and available in rather high bandwidth flavors. Optical won't be cheaper, so its only niches are where 802.11 doesn't work, but optical does, or really high bandwidth, or just because it is cool.

I'll assume you want optical because it is cool.

The lens assemblies are specialized telescopes. You need to focus down to the tip of your fiber. This is a test of your optical skills instead of your electronic skills.

There are a gazillion tradeoffs. Each step creates some optical loss. You can buy bigger lasers, more sensitive receivers, better optics and/or optical amplifiers until the system works.

• Tight focus gives long range, but the telescopes will require active tracking.
• Looser focus means the telescopes don't have to be precisely aimed.
• big mirrors or lenses on your telescopes can produce beams larger than most birds. ... and expenses larger than most budgets.
• Use "Wavelength Division Multiplexing", WDM, so you can put send and receive on the same fiber and telescopes. If the wavelengths are too far apart, then you can't focus both at the same time. Closer wavelength spacing is pricy. This will eliminate any problems with stray reflections.
• Don't use WDM, make sure a receiver doesn't get blinded by its own transmitter. Use four telescopes instead of two. but the pieces are cheaper.
• Use carrier grade fiber hardware and get carrier grade bandwidth.
• Use stuff you can afford.
• Use 802.11, its cheaper and more reliable, just not as cool.

Today's ask slashdot brought to you by: ignorance.

[Anonymous Coward]

Hi, slashdot! I've got a question for you.

I really don't know much about optical principles, especially in the field of lasers, but I know that they're used in cool shit like OC3s and Fiber Channel disk arrays.

Since it's such obvious technology (it's just light, right?) it should be really easy to build one of these things from scratch, at home, using spare parts I found in my garage, and a laser pointer or two. If I can figure out how to duplex the laser pointers, I'll only need two, right? That should be easy, too.

If anyone knows of any webpages where I can get more info about building this Really Good Thing, let me know!