1
\$\begingroup\$

I know this must be trivial for many... Which are the main reasons why line data is not modulated/demodulated in normal enterprise wired networks (in the sense that discrete voltage levels are used)? I'm not talking about DSL of course. I'm thinking of Ethernet BASE-T (e.g. UTP cable) but also older versions (coaxial cable) enterprise computer networks. I undestand it is the best compromise, but I can't find details. For example, a sinusoidal carrier wave could help to reach longer distances on a single cable. Also, without a sinusoidal carrier, line data must be scrambled with a non-trivial techniqhe to ensure a constant variation of the signal. Some of the reasons I may guess: higher costs, data rate limitations, less reliable components... Thanks to anyone who will be kind to provide real world details.

\$\endgroup\$
  • 1
    \$\begingroup\$ When you use cable as an adjective it almost always refers to "cable Television" networks. Eliminating cable throughout your write up would get the meaning you want without confusion. If you want to refer to the actual electrical wires you refer to it as cable plant - strange but true. \$\endgroup\$ – placeholder Sep 16 '14 at 13:06
  • 2
    \$\begingroup\$ @placeholder perhaps.. and maybe the OP is just confused about digital "time domain" modulation and signal transmission compared with analog "frequency domain" modulation and transmission, since he uses sinusoidal carrier waves as his comparison point \$\endgroup\$ – KyranF Sep 16 '14 at 13:11
  • 1
    \$\begingroup\$ @matpop well, a sine wave immediately makes it analogue - imagine all the analogue components and filters and quality control issues involved with the receivers? Digital makes it much simpler for the interface - modern DSP could be used though to decode an analogue signal. With digital signals though you can do time multiplexing and synchronous magic to squeeze much more data out of a "slow" main clock frequency, sort of like a "carrier". \$\endgroup\$ – KyranF Sep 16 '14 at 13:23
  • 1
    \$\begingroup\$ @matpop you may like to read the answer (and maybe also the question) to this EE.SE question:electronics.stackexchange.com/questions/32755/… \$\endgroup\$ – KyranF Sep 16 '14 at 13:30
  • 1
    \$\begingroup\$ @venny REALLY ? do you understand the spectrum of a balanced signal and the analogous 8b/10b? Modulation and encoding are not so easily separable in discreet domains in modern signalling. \$\endgroup\$ – placeholder Sep 16 '14 at 14:57
1
\$\begingroup\$

Let's consider the following classification:

  • wireless Ethernet
  • fiber optic Ethernet
  • wired Ethernet, except fiber optic

and begin to work with it.

1. Wireless Ethernet

Because Wireless Ethernet is out of scope of the OP's question, here i am only propose for the rest of my answer to use the term "single harmonic carrier" instead of the term "sinusoidal carrier" the OP inseparably relates with "analog modulation" as i understand from the comments to the question.

AM and FM are simple examples of modulation based on a single harmonic carrier. In AM, the modulating signal changes (up and down) the amplitude of that carrier. In FM, the modulating signal changes the frequency of that carrier.

In the case of FM with a binary signal as the modulating one, we obtain the simplest FM called "frequency manipulation" where the recessive state of the modulating signal causes the carrier's frequency steps down (or up) and next up (or down) again to the initial frequency after the modulating signal is changed to or kept in its dominant state.

Frequency manipulation with a signal of 3+ discrete states, works similarly.

2. Fiber optic Ethernet

At least up to 10 Gbps fiber optic Ethernet uses typical AM. As we know from physics, a laser (laser diode in our case) generates coherent light signal, i.e. strongly directed electromagnetic waves all are of (presumably) single length (frequency) and in (presumably too) single phase. The modulating signal is also binary here and affects the power of the light being emitted that speculatively equals to affecting the amplitude of the imaginary "superwave" resulting from summation over the set of all source waves in the coherent light signal.

The main rule of AM (and FM) to be possible is that the frequency of the carrier (single harmonic carrier for us) must be several times higher than the frequency of the modulating signal the carrier needs to deliver. For wavelengths, this rule is simply "rotated" upside down.

Infrared light used in fiber optic based telecommunications, including fiber optic Ethernet, has frequencies of about 366 THz (corresponds to 820 nm center wavelength in vacuum), 229 THz (1310 nm), 193 THz (1550 nm) that satisfied the main AM rule for fiber optic Ethernet variants those modulating signal is only up to tens of GHz.

Yes, fiber optic Ethernet is not classified as wired Ethernet in my answer, but it employs "analog modulation" and here we should remember the main rule mentioned above thanks to which such the Ethernet became possible.

3. Wired Ethernet

Which frequencies are available for us in the wired world? Comparing to RF in wireless and to light in fiber optic Ethernet, not so good: Cat 6/6e twisted-pair with 250 MHz, Cat 5/5e twisted-pair with 100 MHz, Cat 4 twisted-pair with 20 MHz, etc. And which frequencies of the modulating signal are available in that case per AM/FM and its main rule? Also not so good (high). This is the reason why "analog modulation" (AM/FM)---single harmonic carrier in our terms---is not employed in wired Ethernet.

But i am concretely not agreed with the claim that wired (at least BASE-T by the interest of the OP) Ethernet does not use "analog modulation" at all. It uses. But oppositely to AM/FM single harmonic carrier variants, it employs multi harmonic carrier approaches. After you know this, the claim the there is no carrier in wired Ethernet is also false.

Consider the 100BASE-TX phy transmitting its idle signal free of scrambling (it's assumed only to make the explanation clearer). There is no data transmitted during the idle therefore the corresponding signal on the medium is the sole carrier. This carrier corresponds to the repetition of the composite symbol JJJJ (four J's) in the "linguistic" representation and to a multi harmonic carrier with its first harmonic of 31.25 MHz (=125/4) in the mathematical (DTF) representation.

(As a common note relating to the carrier: facing a PCS based technology, it is simpler to analyze its carrier in a symbolized, "linguistic" representation first, and then, if necessary, to ground down to the physical medium, finding out an image to the symbol.)

The modulating signal here is also binary: J and K, the dominant/default and recessive states, respectively. The K state is applied only during data transmission and when it's happen at least the first harmonic of the multi harmonic carrier steps down and up again after the state changed. In common case, all the harmonics, including the first (fundamental), its supers (f*n's, responsible for jitter) and subs (f/m's, responsible for wander), in the multi carrier are changed complexly but interdependetly, corresponding to the changes in the modulating signal (data to be propagated over the medium).

enter image description here

I also want to say that in many 100BASE-TX and 1000BASE-T phy chips manufactured today, the signal on the medium is generated and detected by (fast) DAC and ADC techniques correspondingly (as per theirs datasheets) therefore it could be noticed that the signal is analog by itself initially and all the time it presents on the medium.

To end the text, please refresh in mind that "CD" in CDMA/CD is for "Carrier Detect" (joke :-).

Conclusions

Returning to the OP's initial question (in the context of wired Ethernet):

Which are the main reasons why line data is not modulated/demodulated in normal enterprise wired networks?

Well... To be short, there is only one main reason and it is frequency (spectra) limitations of available mediums preventing the use of single harmonic carrier AM/FM.

\$\endgroup\$
  • \$\begingroup\$ Wow! Thank you very much for taking the time to provide such a detailed answer! It's late night here right now but this is very interesting for me and I'm going to read it all carefully - I think I'll have to accept this soon as the best one... Thanks again alex \$\endgroup\$ – matpop Mar 28 '15 at 22:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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