I am confused because I read the LEDs/LASERS used in optical fiber have finite levels. If for instance ASK is used to transmit digital baseband data then will the LED or LASER be able follow the ASK continuously and will detection be possible?

  • 1
    \$\begingroup\$ You've never seen an LED get brighter and dimmer before? \$\endgroup\$
    – user253751
    Sep 14, 2020 at 13:10
  • \$\begingroup\$ I have but just seeing is different from transmitting with ability to detect and then regenerate the amplitude requires precision and the what I am asking is it practically done. \$\endgroup\$ Sep 14, 2020 at 13:16
  • \$\begingroup\$ No, it's not different. \$\endgroup\$
    – user253751
    Sep 14, 2020 at 13:53
  • 1
    \$\begingroup\$ What exactly do you mean by "analog signals"? \$\endgroup\$
    – Andy aka
    Sep 14, 2020 at 14:01
  • \$\begingroup\$ There is a great optics demo for school children that uses a HeNe laser and a photodiode to send audio to a speaker across the room by modulating the laser power. Audio quality is very high, similar to a wire. \$\endgroup\$ Sep 14, 2020 at 14:26

2 Answers 2


Indeed, optical comunication can transmit analog information. In fact, this is used quite a lot in, for example, Radio over fiber. Even in raw optical comunication, multi-level signaling such as PAM4, and more complex modulation such as IQ signaling is used.

How is this done? Well, you hinted at one method: modulate the light source. You give the example of an LED, however most high-performance optical communication systems actually use laser diodes such as VCSELs, instead of non-coherent (and less spectrally pure) sources like LEDs.

But the key is that you don't have to modulate the light source. You can turn the lightsource on all the way, and then modulate the light instead. Because lasers give you a single coherent wave of light (instead of lost of non-coherent waves at the same time), you can use wave-properties to modulate. This is the idea behind high-performance optical modulation using Mach-Zehnder modulators. In a Mach-Zehnder modulator (or interferometer), we start by splitting the signal up into two paths. We can then change the phase delay of one (or both) of the paths somehow (how we do this electronically is outside of the scope of this response). We then add up the two paths again. If both paths have the same phase delay, they add up constructively, and we get the same output power. However, if we have 180 degrees phase shift in one of the paths, the two paths cancel out and we get no output signal. By modulating somewhere in the middle, we get a lower output power that we started with.

  • \$\begingroup\$ The MZ modulator typically has a sinusoidal transfer function. It's only going to be useful for analog systems with some additional circuits (pre-distortion or feedback control) to linearize its response. \$\endgroup\$
    – The Photon
    Sep 14, 2020 at 18:03

LEDs can certainly be run at any light output level, and that level controlled very quickly to modulate it. However, the problem, like with any analogue system, is gain, noise and linearity. Optical systems are non-linear, and have uncertain gain in the extreme.

LEDs are somewhat non-linear in current to light conversion, and are temperature dependent. Photodiodes are also non-linear. The fibre couplings can change coupling efficiency with disconnect and remaking, and with temperature, strain, vibration, probably ageing. Lasers are extremely non-linear, having a threshhold. Systems are often intended to be run over any length of fibre up to some maximum. Signal to noise ratio is not good.

All of these point to running the system digitally for preference.

That's not to stop any hobbyist connecting a LED system up and calibrating the linearity, or using over a small enough swing, to run it in an analogue mode. Note that there are opto-couplers designed to be used for transmitting analogue data. These usually have a second photodiode monitoring the output of the LED, used as feedback to linearise the response of the LED+diode combination. The solid coupling between source and receiver mitigates the several problems noted with the fibre part of the connection.

  • \$\begingroup\$ Neil, your linearity standards seem high. I would regard photodiodes among the most linear sensors around, with a handful of decades of dynamic range. LEDs are linear with current to a first approximation. LASERS too, when run above threshold knee. \$\endgroup\$
    – glen_geek
    Sep 14, 2020 at 14:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.