Talk:T-carrier

Old talk

It may be necessary to link Plesiochronous Digital Hierarchy somewhere in here... --Joy [shallot] 10:31, 4 Nov 2004 (UTC)

Huwr, yes, a see-also section is easy, but it would be better to have it properly integrated :) --Joy [shallot]

I've looked at this page several times, and thought - damn, this sucks. I wish I had the time to add some data. Then a google search for something unrelated ran across DS1. A DS-1 is basically a T-1, but a search on the Wiki for "T-1" drops you here (where there's very little useful info), while the DS-1 page is much more useful, I think. Looking around, the DS0/DS1/T1/T3/E1/T-carrier section needs work. Inconsistently named, information is fragmented, redirects go to odd places, etc. As a stopgap, I added a link to the DS-1 page; maybe next month (after finals) I'll see about helping straighten this page out some more, if it still needs it. Codayus 08:03, 18 Nov 2004 (UTC)

I'd like to see the E1 page separated out from T1, since E1 is by far the most common worldwide standard (T1 really only seen in US/Canada these days). E1 signalling should refer to R2-MFC channel associated (CAS) tone methods as well as ISDN and SS7 methods. I don't know how to change the page redirection and setup a new page though. Paidup

Agree, but same kinds of out of band signalling (SS7 and ISDN (PRI)) are carried across a T1 carrier here in the US. Both the T1 and E1 articles should refer to signalling methods. Kgrr 15:15, 9 August 2006 (UTC)

T-1 and DS-1 not the same thing

Paidup (and all), There is a key thing missing in this article. T1 is a physical layer specification only. It describes the four-wire point to point circuit that carries the DS1/24*DS0 signals. T1 transmits DS-1 formatted data. On the other hand, the DS-1 specification describes the Digital Signal - synchronization, framing, clocking, timing, encoding... DS-1s can be carried across other media such as fiber.

Also, in this article, I expected more information about the physical interface

• smartjack / NIU / demarcation
• CSU/DSU
• NIU loopback (bellcore) vs * CSU/DSU loopback (ANSI)
• extending the demarc
• V.35
• RJ-45 vs RJ-48
• BERT testing
• "Dry" versus 'wet" T1 circuits

etc.

I will try to contribute more to this article when I find the time. Kgrr 01:24, 12 August 2005 (UTC)

It's a year later, and I have not re-written the article. Clearly, the DS1 information needs to be moved out of this article and this article should concentrate on the physical copper T1 medium, not the DS-1 particulars. They should be in the separate DS-1 article. Kgrr 15:15, 9 August 2006 (UTC)

Kggr I couldn't agree more. I was a Member of Technical Staff at Bell Laboratories 1976-1984 in the T1 Transmission System Engineering Group. [My last name was my 'Kaup' at the time, my step-father's. On his death I changed it back to my birth name].

I just went down to the crawl space and dug out a handful of references on T1 and DS-1 that I've managed to hang on to over the years, and would really like to make a contribution here, citing references whenever possible. But I don't want to step on any toes and wonder first of all how multiple contributors manage to collaborate. In particular how do we socialize, review, and finalize content amongst ourselves before committing to a page, as opposed to individually making edits that overwrite previous edits and wind up de-moralizing previous contributors?

Regardless, I strongly share your opinion that the topics of T-carrier and DS-1 need to be completely separated. The only thing that should remain in one section is a brief reference to the other citing their relationship. Where to start? — Preceding unsigned comment added by Tballister (talk • contribs) 01:48, 19 January 2013 (UTC)

Other changes

• There are three NIUs (NIUs are the cards inside of the two slot mounting cases, not the case themselves) in the first picture, not two. The single card in the left case looks to be either a HDSL or HDSL4 NIU. The left card in the left case is a HDSL2 NIU. More accurately, those are HTU-R (or H2TU-R or H4TU-R) cards, that terminates HDSL data and outputs a DS1 signal, functionally similar to older T1 NIU, but are not compatible with a T1 system.
• Explanatory information on variants (T1C (two concantenated T1s), etc.
• From a service provider to a customer, it would look something like

[DSX-1}--{ORB]--3kft--[RPTR]--6kt--[RPTR]--...--[NIU]--[CSU]--[DSU]--[CPE]

showing the spacing between the office repeater in the office repeater bay (ORB) and the first line repeater as 3000 feet (3Kft) and spacing between subsequent repeaters as 6Kft.

• Emphasis (at least orginally) on T1 as being a span system.
• Designed as a short haul interoffice transport for use inside the telephone companies, and as part of T Carrier System linking two D1 channel banks:

[D1 bank]--[DSX-1]--X--[DSX-1]--[ORB]--[RPTR]--....--[RPTR]--[ORB]--DSX-1]--X--[DSX-1]--[D1 bank]

• Links to other (analog) carrier systems, showing T1 development in context.

Historical Note on the 193-bit T1 Frame

I was a junior engineer at Bell Labs on the team that developed the T1 system. Many people who heard my story thought it interesting to record this for prosperity.LoopTel 07:55, 29 November 2006 (UTC)

Too technical

Editor Felix is right. An outsider who looks at this will never understand why T-carrier exists. I thought of writing a "Background" section to explain why multiplexing is good and digital multiplexing is better, but then saw there's already a link to carrier system. That's merely a terse government definition, little more revealing to outsiders than this one, but that's the place to put a small history of telco carrier systems, unless someone can recommend a better location for it. Jim.henderson 14:07, 18 September 2007 (UTC)

Bandwidth confusion

I disambiguated one instance of the word bandwidth to Bandwidth (computing) because it was clearly referring to a bit rate. However, the following sentence is confusing me:

"Initially, T1 used Alternate Mark Inversion (AMI) to reduce bandwidth and eliminate the DC component of the signal."

I put the there because I was guessing this was probably the other sense of bandwidth, namely Bandwidth (signal processing). However, the context does not make this clear. It does not make sense that someone would want to lower a bit rate on a data line, but maybe they would want to reduce the range of frequencies used to carry the data. This is circumstantially corroborated by the mention of reducing the DC (direct current) component of the signal. However, I think perhaps this should be reworded, because it is confusing to use the word bandwidth to mean different things in the same article. If my guess is correct, maybe a phrase like range of frequencies would be clearer. Can an expert confirm and/or reword this? Thanks! CosineKitty (talk) 02:20, 28 June 2008 (UTC)

This being an art in which both literal and figurative bandwidth are important, I recommend using "bandwidth" where it is intended literally, and "bit rate" where bit rate is intended. Jim.henderson (talk) 05:17, 28 June 2008 (UTC)

That makes sense. I changed bandwidth to frequency bandwidth, and linked to Bandwidth (signal processing), to emphasize that they were trying to limit the range of frequencies consumed, not trying to limit the bit rate. While the context may have been clear to an electrical engineer, someone familiar only with computing concepts reading an article about a digital technology like T-carrier could easily have been confused. I did some other minor rewording for style. CosineKitty (talk) 13:03, 28 June 2008 (UTC)

Yeah, the problem stems from the use of "bandwidth" as synecdoche for either bit rate or channel capacity which arises in turn from computer people misunderstanding communications people twenty years ago. Literal bandwidth is an important factor in determining either of these digital questions, so its word is used to represent both of them. Fixing the situation by specifying bandwidth (signal processing) works well enough in signal processing contexts, which this is not, since alternate mark inversion is a way to increase bit rate, decrease bandwidth, and avoid the complexities of signal processing. Grumble grumble; the whole magillah could be avoided by not mixing the metaphorical use of words with their technical use but the habit seems entrenched, and piercing the resulting symantic fog will require much rework. Jim.henderson (talk) 16:43, 28 June 2008 (UTC)

highest voice frequency which the telephone system transmits is 4,000 Hz,

--except that it's not. Or at least it didn't use to be. It was 3400 Hz: http://portal.acm.org/citation.cfm?id=1142855 —Preceding unsigned comment added by 218.214.18.240 (talk) 00:27, 6 February 2010 (UTC)

Reason for rate choice

The article currently says "To accommodate loading coils, cable vault manholes were physically 2000 meter (6,600 ft) apart, and so the optimum bit rate was chosen empirically[.]" What does the distance being 2000 meters have to do with the optimum bit rate being chosen empirically? These seem to be completely unrelated issues. 130.179.29.59 (talk) 20:34, 31 August 2011 (UTC)

The very short answer is that, as the distance of a wire gets longer, high frequencies degrade much faster than low frequencies. That is the key to that sentence of the article. The T1 signal is 1,544,000 bits each second. Each of those bits can be either a 1 or a 0. You "set" those bits to 1 or to 0 by mixing small amounts of higher frequencies with the T1 signal. Since those higher frequencies degrade (disappear) as the wire gets longer, you lose bits as the line gets longer. Lines longer than 2,000 meters lost too many bits. TheWasScriptGuy (talk) 16:01, 30 March 2012 (UTC)

That really doesn't address the question. Long lines require lower bit rates, but why does that choice have to be "empirical"? 206.45.170.157 (talk) 16:58, 25 November 2012 (UTC)

I think there are a several discussions missing from this article that would help to clarify.

First should be a discussion presenting the bipolar encoding of 1's and 0's employed by the T1 transmission system. That encoding results in not just 1's and 0's, but rather +1's, 0's, and -1's on the tranmission line. E.g., the definition of Bipolar encoding is that while each 0 is represented by a 0, each successive 1 is represented at the opposite polarity as the previous 1. For example the 8-bit binary sequence 11011101 (taking the 1st 1 to be positive) is: +1, -1, 0, +1, -1, +1, 0, -1. There is no "setting of the bits by mixing small amounts of higher frequencies". Rather, the bits are exactly (ignoring B8ZS for the moment) the result of the Analog to Digital (A/D) conversion of the voice audio. As a result of the encoding there is a power spectrum which centers at 772kHz, exactly 1/2 the bit rate. In other words the shape of the energy spectrum of low to high frequency components is a function of the 1's and 0' in the data, not the other way around.

Now, with that in mind I'll try to address the question: "What does the distance being 2000 meters have to do with the optimum bit rate being chosen empirically?". Answer: the starting point is something like, "Ok, we've got these manholes already dug into the ground at 6000ft spacings, and conduit in between them. And we've figured out we can get more conversations onto a single pair of copper with a digital encoding scheme. But how can we figure out exactly how many more conversations?" The answer to that is a function of additional aspects that should be discussed in the article.

One of these is the natural line loss of the signal. Those nicely sharp edged and flat topped pulses that emit from the transmitter disappear into a waveform more than 30 dB lower in amplitude and more representive of a sine wave than pulses. So, we've got loss to deal with, and that loss has a "shape" to it; more loss at higher frequencies that at lower frequencies. We don't "lose bits as the wire gets longer". Rather we lose energy; and more at higher frequencies than lower. [If you studied the amplitude versus frequency (spectrum) of the signal immediately at the transmitter output on a spectrum analyzer, you'd see its dominant 772kHz peak, but you'd also see a shape with significant energy above this, tapering off relatively slowly from the 772kHz level and extending all the way out to 20 or 30 MHz or more(the rise and fall times of the pulse edges represent the highest frequency components). In contrast a spectrum analyzer view of the signal at 6000 ft of cable will show that energy drops off sharply above the 772kHz level to almost nothing beyond 3-4MHz or so.]

Now we get to a third discussion that should be part of the article, mainly that there is energy coupled into individual copper pairs from other sources such as rotary phone dial pulses (at that time), electrical motors and other electromechanical spikes, and signals coupled in from other adjacent pairs in binder groups. This of couse is referred to as noise. At 6000 ft along the wire then, we have the attenutated signal plus the added noise from the environment, and therefore a resulting "Signal to Noise Ratio" to deal with.

So the question that had to be answered was "What level of signal will be necessary to overcome the inherent noise? This is the question that got answered empirically. For example, you start out with a signalling rate of 1Mbps and discover "Wow, the noise is almost negilgible relative to the level of the signal". Then you crank things up to a 2Mbps signalling rate, and discover "Phoey. We're getting bit errors every second now". It's because at this higher frequecy there is too much attenuation of the signal to overcome the noise. So you find a compromise (the art of engineering) where the available electronics (of the time) can reliably recover the +1, 0, -1 levels. That is, where the added in noise does not cause inaccurate decisions about whether the mixed signal+noise on the line was a +1, 0, or -1. Through trial and eror a signalling rate emerged that could accommodate 24 voice channels with adequate error margin, and 24 channels was sufficient to justify the investments in engineering and deployment.Tballister (talk) 07:53, 19 January 2013 (UTC)

Content Re-Organization

I would like to submit that most of the content on this page is specific to the T1 Transmission System which is just one of the systems defined by T-carrier. It feels to me like a better organization would one of two choices. [1] The content of this "T-carrier" article should be reduced to discussing the copper based transmission system heirarchy of T1, T1C, and T2, without going deeply into the specifics of any of them. Then separate articles generated and cited for each of the "T1" or "T1 Transmission System", "T1C" or "T1C Transmission System", and "T2" or "T2 Transmission Systems" topics. Or [2] rename this article to "T-carrier Transmission Systems", and then use 2nd level headings for each of the systems that go into their specific details. In either case the content should stay specific to the physical transmission systems. All topics related to "Digital Signal Heirarchy", DS-0's, DS-1, framing, and signalling should be merged to the DS-1 article or to the Digital Signal Heirarchy article.Tballister (talk) 08:12, 19 January 2013 (UTC)

Move?

T-carrier → T1 Transmission System –

• Most of this article is very specific to the T1 Transmission System, which is just one of several different telephone transmision systems known as a group as T-carrier. I propose to re-organize the content here to be specific to that one system, and then create a new T-carrier page that discusses the full set of systems. Tballister (talk) 07:52, 20 January 2013 (UTC)

Moved from speedy

• Object this clearly needs discussion, since you want to reorganize the article and split off a new article for the same article name "T-carrier" -- 76.65.128.43 (talk) 08:12, 20 January 2013 (UTC)

Yes. I propose to move most of the existing content to be organized under the Article titled "T1 Transmission System" because it is specific to that T1 system, and let the small amount of content that is applicable to the set of transmission systems defined as T-carrier systems remain organinzed under "T-carrier". Further, I propose to expand the "T-carrier" content to encompass all of the systems that were defined as T-carrier systems (e.g., T1, T1C, T2, T3, T4, ...). TCBallister (talk) 05:27, 21 January 2013 (UTC)

Survey

• Oppose – First, for the unnecessary capitalization; should be T1 transmission system per MOS:CAPS. Second, it would be better to go ahead and add the new material that's more general about the T-carrier, and then see if a split is needed. Dicklyon (talk) 18:54, 20 January 2013 (UTC)

Dicklyon: I respectfully disagree. Point 1: As a member of the T1 Transmission Systems Engineering group at Bell Telephone Laboratories I can tell you that the formal name is exactly "T1 Transmission System" (with capitalization). Not being argumentative, but I've always thought that proper names are capitalized. I main, "New york city" just doesn't feel right. Point 2: Honestly I don't feel the need to wait because the distinctions are clear already, or I wouldn't have suggested the split. The definitions of "T-carrier" as a category is well established; in simple terms, its a group of copper based transmission systems. In contrast, the "T1 Transmission System" is just one of those systems, all of which were engineered by Bell Telephone Laboratories. Each has quite a number of system-specific characteristics that in my opinion warrant dedicated articles. There will no doublt be more content in the "T1 Transmission System" article because it is the most ubiquitous in the United State. But for the record the other systems are real, and deserve separate descriptive articles. TCBallister (talk) 05:21, 21 January 2013 (UTC)