Color Multiplexing Through Fiber Optics

If you want to go high bandwidth, fiber optics is the way to go. From trans-oceanic cables to the yet-unseen ‘fiber to every home,’ fiber optics allows a lot more bandwidth than a copper cable. In low-bandwidth applications, fiber optic cable transmits data using one color of light. There’s a way to get more bandwidth out of a fiber optic cable, as [Shahriar] found out while experimenting with an RGB LED.

For his experiment, [Shahriar] used a BlinkM programmable RGB LED and a Sparkfun color sensor. In fiber optic lines with one light, it is possible to send many simultaneously using PWM, but noise becomes a problem at high data rates. Using an RGB LED, [Shahriar] sends three levels of Red, Green, and Blue to transmit 9 bits at a time – perfect for sending a byte with a parity check in one quick light burst.

[Shahriar]’s technique is exactly how the pros pump massive amounts of data through a single fiber optic cable. All the tools, code, and MATLAB functions are available on [Shahriar]’s site, ready to be used by anyone wanting to experiment for themselves.

In the video after the break, [Shahriar] breaks everything down, including the tools, theory, and actual circuits. It’s an amazing video demo, so thorough we’re wondering if [Shahriar] has any teaching ambitions.

24 thoughts on “Color Multiplexing Through Fiber Optics

  1. “fiber optics allows a lot more bandwidth than a copper cable”.

    Says who ?!! A copper cable can transmit signals well over tens of Gigahertz (with hundreds of carriers in the signal). Look at the back of your satellite dish, no optic fiber there.

    The main difference is the signal strength over distance. Optic fibers are way more efficient to carry information this way.

    1. The same enter my mind as well, but when we really get down to it, getting the maximum amount of data data from one end of a a cable to the other end the statement, “fiber optics allows a lot more bandwidth than a copper cable” is valid.

    2. The bandwidth of Fiber exceeds that of Copper by orders of magnitude. :) The absolute best Fiber cable one can purchase is a 110GHz 1mm dielectric cable at hundreds of dollar per inch. Needless to say, such a cable has very poor frequency response for long lengths.

  2. Wavelength-division multiplexing (WDM) is there even since optical fibers are in use, I’m even subscripted to a magazine called “WDM” which is specifically about this (the whole magazine!).

    Just as with RF, the data transmitted over fiber is uses as much bandwith as its available. “Ultra dense WDM” uses GHz-order channel spacing, fitting more than a hundred channels (colors) in the same fiber strand. Keep in mind that 10, or even 100 GHz is a very small quantity when speaking about light, the typical 532nm green light is about 564THz!

  3. On the event Shahriar has teaching ambitions, he’s already full filling them online, and my guess is that he has a teaching position at University of Toronto. the @alvieboy I suppose perception is one of those things that is relative to the individual, the only thing I got from TFA is that is was about an experiment that teaches. Ohm’s Law and Kirchhoff’s certainly aren’t not new, but practically every EE or electronics tech that went through formal education performed experiments to help them understand why the laws are valid, along with constructing “old” circuits as a learning experience. That 2+2= 4 is well known but we still teach it. I pretty much had a valid clue why fiber has the bandwidth it does, rather dismissing Shahriar lesson I viewed, and learned more than I knew before. More great work from Shahriar, thank you.

  4. I’ve red LEDs with at least a couple of different wavelengths in a component catalogue. Would be nice to see exactly how many colours you can get just using LEDs. The problem might be getting monochromatic filters in all the colours necessary, ideally with a narrow optical bandgap (ie a *specific* wavelength).

    AFAIK, which is not much, they use lasers for real-life WDM. It’s an interesting field, how to get specific and narrow colours.

    I don’t suppose the AM modulation he’s doing with the brightness would be good for long-distance, but there’s plenty of bandwidth available with time-multiplexing.

  5. This is Shahriar, the host of The Signal Path.

    Thank you all for your kind remarks and interesting discussions. Just a few remarks:

    The bandwidth of Fiber and Copper are not even in the same league. A good single-mode-fiber (SMF) which is dispersion compensated, can transmit many Tb/s of data over many kilometers using WDM. Most Fiber systems are limited in bandwidth due to the Optical-to-Electrical, or Electrical-to-Optical interfaces.

    As for my teaching interests, aside from my day job, I hold an adjust professor position at Columbia University in NYC. If you ever decide to visit while I teach there, drop me a line!

  6. If I may interrupt the copper/fiber debate, using 3 levels of each color now goes into analog, light level over fiber gets a bit weaker at distance, this is why digital, is it on or is it off, perhaps transmit 3 bit streams in parallel?

  7. It seems to me that the use of PWM by the blinkm sort of negates the point of this experiment. The PWM produces what appears to humans to be a range of colors due to persistence of vision, but to a fast receiver, it is 3 channels of data with the information encoded as duty cycle. It seems that greater bandwidth could be achieved more simply by sending binary data at the maximum data rate on the three light channels rather than relying on the (usually rather slow) PWM clock and adding the complexity of the RC filters to provide some electronic “persistence of vision” for the receiver.

    1. Not that it isn’t an interesting proof of concept, but I’d be curious how a practical application would be implemented that could outperform a traditional system. Perhaps a fast, low-resolution D/A converter could be used rather than PWM for driving the LEDs. Also, how would practical bandwidth of that system compare to a simple system using the 3 LEDs to send 3 bits of data?

      1. You are right that the PWM -> Analog defeats the purpose. However, I did want to use the original components without modifying them. I could also simply measure the duty-cycle at the receiver to decode the bits.

    1. I was just about to comment about FTTH. I work for a telecom provider and commercially, Fibre-to-the-Home TV, internet and phone has been growing for the last several years in more and more areas ( I’m in the Canadian Maritimes ) !

  8. Guys i am trying to send atleast 3 types of signal via coaxial cable over a single fiber optic cable. One Dc signal of 5v and one DC signal of 15v and one analog of atleast 1.5Mhz. I cant use a power supply more than 24v dc. Is there anyway i can do this guys please help

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.