Hacker News Reader

Comment by dthul | Hacker News Reader

I wonder if you could send power at one wavelength, and data at another wavelength, over the same fibre.

Yes. This is already done. It's how almost all submarine communication cables currently work. Most long-distance fibre links do not use electronics to regenerate their signals.

They use optical amplifiers, which take light at one wavelength and use it to intensify light at another wavelength. They're much like lasers (technically I think they count as optically-pumped lasers?), and they turn on from a very small input signal, effectively reenforcing it.

This can happen across multiple signals, on different wavelengths, in parallel. Like a broadband radio amplifier, it boosts everything across a large working bandwidth. There are even optical compressors (also powered by light), which speed up the baud rate of signals. That way a slow electronic system can produce the original pulses, and then they can be compressed to faster than electronics can work with, and then multiplexed with many other signals at different wavelengths, and this whole composite thing is sent down the line, amplified without decoding along the way, and then finally the whole thing is reversed at the other end.

This is the trick behind how fibre links are so fast, considering there are no electronics that can handle data serially at those speeds.

You're right about submarine fibers, but you seem to suggest that the pump light for the laser amplifiers is transmitted through the fiber from the cable landing point - like the technology discussed in the OP.

That is certainly not the case, the pump light is generated from electricity right where the laser amplifier sits in the fiber. No real amounts of energy are sent optically down the fiber. To power the amplifier, a high voltage DC line is designed right into the submarine fiber cable. And those things carry a lot of power, a long fiber cable will draw tens of kilowatts of DC for all the optical repeaters.

The reason is, of course, that thousands of miles of cable has a pretty insane optical attenuation, no matter what you do, because optical attenuation rises exponentially with length. The electrical resistance of a high voltage DC power line only rises linearily, on the other hand.

Just to prove I never took physics, where are the photons actually going in a long distance undersea cable that makes it impossible to just flash a signal across an ocean sized length of fiber? (As I had assumed was the case.) Is it more a loss of clarity/resolution in terms of wavelength rather than the photons going astray?

Think of it as the same as the internal resistance of a conductor. Some of the photons are effectively absorbed by the fiber and turned into heat.

Thanks! I am embarrassingly so much a 'software' kind of person that this didn't occur to me, but I get the idea a bit better now.

You're right about submarine cables running DC along the shielding/armor to power the optical amplifiers. However it's worth pointing out that there are so-called "repeater-less" systems that do use optical delivery of pump power to the amplifiers (typically they combine this with Raman amplification). Those systems can deliver high capacity communication (not sure where the record stands at but 100s of Gb/s to low Tb/s) over >500 km without any electrical connection (you still need power at the receiver though).

These are typically used for short submarine connections to e.g. connect an island. As it is much cheaper than running a full repeatered system.

> because optical attenuation rises exponentially with length

I believe the attenuation is stated in dB/km, therefore rises linearly (or even logarithmically if you look at it from the uncommon energy-wise point of view) with distance. Why should exponential be the case?

Imagine a length of fibre that transmits 90% of the light put in. Take the output, and pass it through another equal length. At the output of the second fibre length, we have 90% of 90%.

Why not both via the same wavelength?

I was thinking that the "power extraction" might attenuate the signal too much, and it would probably lower the power output since you need to modulate the light to transmit data, instead of having it on full brightness all the time. But maybe it would work for certain applications!

Efficiency of laser diodes goes down quite a bit with bandwidth. More importantly you typically want the data going the other way (from the powered sensor). If you would use the same fibre for both directions (might be done for space constraints) the issue of using the same wavelength is that there are scattering processes (some fundamental to how fibres work) in the fibre that will cause some light to be back scattered and act like noise essentially. Your sensor would transmit with only very little power so the SNR might be completely destroyed by the back scatter of the high power piwer delivery light. If they are at different wavelengths they can be easily separated.

Oh, you are right - I didn't think about the other direction at all. Thanks for the explanation!

Could they use separate phases of the same frequency, or do things like DWDM preclude that?

There's PSK and the likes which mess with phase but I don't think that's the same as what you're asking as you would typically then use something else to actually get separate streams with it. The problem with pure and plain phase division multiplexing is how do you separate it back out on the other side? You can make up as many pairs of wave values to create the resulting wave without doing something else beyond plain multiple phases.

Time division multiplexing and frequency or amplitude division multiplexing preclude most things because they are so cheap and simple these days. Polarity is also another knob to mess with when you need to squeeze more.

Reply on news.ycombinator.com