Talk:Line code

Unnamed section

We should have pictures showing everything, including:

• raised cosine pulses, nyquist pulses (same thing?), triangle pulses? instead of just rectangular
• unipolar and bipolar (pseudo-trinary or alternate mark inversion) instead of just polar
• and whatever else there is - Omegatron 02:43, May 2, 2005 (UTC)

Modulation

I think It should be explained the relation with modulation.

I added a brief mention of modulation -- anything else I need to add? --68.0.120.35 20:27, 9 December 2006 (UTC)

clock sync

Maybe I'm just missing it, but it doesn't seem like any of the immediately related pages mention *why* clock sync is important. Explaining for example why constant synchronization is required, rather than just syncing at the start and sending data for the rest (ie. cheap clocks are not nearly accurate enough). Where does this sort of thing belong? Line code is a pretty weak page in general, and I'm not sure this applies only to line codes. It's pretty relevant for any sort of digital communication regardless of scope. On small scales it's usually out-of-band, but perhaps this is worth mentioning as well?

Thoughts?

Please, someone take the time to describe why this type of code is called a "line code." Is it because the code was used to signal directly upon a physical "line", such as a telephone pair of twisted copper wire? If so, please add this; if not please add the correct information to the page. The jump directly into detailed information, without fulfilling the most basic requirement of complete definition of the term itself, makes the article much less useful than it should be. —Preceding unsigned comment added by 68.15.57.109 (talk) 20:36, 18 July 2008 (UTC)

Some communication systems (such as the Local Interconnect Network) *do* transmit a "sync header" at the start to synchronize clocks, then just data for the rest of the packet/sector. Is there a good reason some systems sync at the beginning, while other systems use a self-clocking signal ?

Yes, I would like a paragraph explaining why clock sync is important. It certainly does apply to all line codes, so perhaps we could stick it in the "line code" article for now, until we find a better (more general) place to put the explaination. Perhaps telecommunication or synchronization or modulation or digital communication or self-clocking signal or intersymbol interference ? The Run Length Limited article has a one-sentence explanation ... --68.0.120.35 19:15, 9 December 2006 (UTC)

The kind of "clock sync" that all line codes deal with is finding the beginning and ending of each symbol. This kind of sync is sometimes called Phase synchronization. With accumulate-and-dump systems such as PRML, slight errors in starting and stopping accumulation lead to increased ISI which leads to increased errors.

Line codes that transmit 8 data bits at a time (such as 8B10B) can also help find the beginning and ending of each byte of 8 data bits (a different kind of clock sync), reducing framing error.

--68.0.120.35 07:23, 13 December 2006 (UTC)

The clock recovery article might be a good place to discuss this. --75.37.227.177 16:24, 7 August 2007 (UTC)

"code" vs. "symbol"

The text seems use the word "code" both for "symbol" and for "code" (a set of symbols). That makes some text quite confusing, e.g., "design each transmitted code such that every code that ..."

The text should use "symbol," or at least disambiguate the references (e.g., by using multiword phrases for one or both meanings of "code").

12.109.151.100 (talk) 20:15, 27 November 2007 (UTC)

Please help me make this page easier to understand.

I agree that this article ought to use a different word for a specific single pattern -- such as the pattern "1010", used to represent the data sequence "101" in 3b/4b encoding -- and the full list of replacement patterns -- such as the entire "3b/4b encoding".

Alas, I must protest using "symbol" for either of these meanings.

To me, symbol (data) is closely tied to the analog bandwidth.

Using "3b/4b encoding", transmitting 3 data bits requires sending 4 symbols.

Using 2B1Q encoding, transmitting 4 data bits requires sending 2 symbols.

(As you can see, some encodings use more than 1 symbol per data bit; others use less than 1 symbol per data bit).

Would using "code word" for the replacement pattern, and "code" for a particular full set of code words, be better?

--75.19.73.101 05:39, 3 December 2007 (UTC)

That would be better indeed. In some literature they use "line coding scheme" for "code", but it should be clear that within this article "code" refers to one particular line code. (not necessarily just a full set of code words, not all line codes are mappings) MrZap (talk) 21:24, 1 April 2009 (UTC)

Line coding vs. Line encoding

I just reworked the introduction a bit. The current state looks like the conflation of at least two edits, with the introductory sentences of both still present. Under this theory, line encoding is just a synonym for line coding and is (redundantly) redefined in the second paragraph. However, I don't have enough domain-specific knowledge. Maybe there is a distinct usage of the second term. Is there? -- DRE (talk) 20:51, 1 September 2008 (UTC)

I dont think there is a difference. MrZap (talk) 21:12, 1 April 2009 (UTC)