Talk:Ethernet physical layer

Ethernet over twisted-pair cable

Isn't this drawing incorrectly labelled TIA-B when its TIA-A? What does 'z' mean in the table? —Preceding unsigned comment added by 204.244.209.138 (talk) 11:45, 24 May 2008 (UTC)

Additionally, in the table Early implementations (10 Mbit/s and 1 Mbit/s) says: Runs over four wires (two twisted pairs) on a Category 3 or Category 5 cable. An active hub or switch sits in the middle and has a port for each node. This is also the configuration used for 100BASE-T and gigabit Ethernet. Which seems a bit incorrect, as far as I know Gigabit Ethernet requires 8 wires (four twisted pairs). ibarrera — Preceding unsigned comment added by Ibarrera (talk • contribs) 20:00, 13 January 2011 (UTC)

Does this page just duplicate other pages?

There's already a gigabit Ethernet page giving details of gigabit Ethernet physical layers, and the Fast Ethernet page should probably swallow the pages for the individual 100Mbit/second standards; this page doesn't even mention 10 gigabit Ethernet. Guy Harris 17:19, 14 June 2006 (UTC)

Probably it is, though User:Oakad clearly did a nice job with it. It would probably be better to merge the respective wording and tables onto the Fast Ethernet and gigabit Ethernet and the not-yet existant early/ancient/original/10mbs ethernet pages. -- KelleyCook 18:58, 14 June 2006 (UTC)

I've merged stuff from here into Fast Ethernet, and turned the "Fast Ethernet" section of this page into a summary that points to the Fast Ethernet page. I'll look at doing the same with the Gigabit Ethernet stuff.

The Varieties of Ethernet page also has a summary of various Ethernet physical layers; should the two pages be merged? Guy Harris 21:06, 15 June 2006 (UTC)

Yes. I've done that merge. Guy Harris 08:08, 3 October 2006 (UTC)

Does this overlap with Ethernet over twisted pair. --Boscobiscotti 04:57, 8 May 2007 (UTC) I also added a link to [[autonegotiation]--Boscobiscotti 05:00, 8 May 2007 (UTC)

Echo cancelation

This sentence: "Telephones also use echo cancellation to simultaneously transmit in both direction over a single cable pair" is probably wrong. Analog telephone does not use echo cancelation. This would require sophisticated signal processing equipment which was not available in the age of analog telephony. Telephone does use echo cancelation if transmitted over a digitial medium such as ATM or IP with higher delay as pure analog switching, because the echo would be disturbing. The need for echo cancelation was one of the main critic points in ATM: 53 byte cells are too long to use without echo cancelation and thus made this expensive equipment nescesarry when ATM was introduced in the phone network, while still beeing way too short for sensible data transfers. citation: "Interconnections: Bridges, Routers, Switches, and Internetworking Protocols" by Radia Perlman. 85.178.193.38 17:29, 30 November 2006 (UTC)

The sentence is correct, actually. See the article on echo cancellation, with particular attention to its use in modems. (I've also added the appropriate links to the article.) Peppergrower 18:45, 13 November 2007 (UTC)

The sentence is incorrect . Analog Telephones use a "Hybrid" (eg a Directional Coupler) to separate Transmit and Receive directions. This has nothing to do with echo cancellation. Gutta Percha (talk) 10:07, 18 June 2017 (UTC)

Long distance telephone systems, I believe even modern systems, use an echo cancellation system, but not like is described here. Simplest explanation is that it attenuates the lower level signal. When one person is talking, the echo will be lower, and is attenuated, but commonly described as cancelled. Maybe it should be called passive echo cancellation. You can hear this as the background noise gets quieter when the other person is talking. Full duplex modems require this turned off, which specific frequencies are designed to do. Gah4 (talk) 19:30, 14 September 2017 (UTC)

Back to Ethernet. The sentence from 2006 is no longer in the article. What we now have is, "...use all four cable pairs for simultaneous transmission in both directions through the use of echo cancellation." I don't think echo cancellation is the best description. The physical layer technology in 1000BASE-T is a combination of digital telephone hybrid and adaptive equalization. ~Kvng (talk) 15:14, 17 September 2017 (UTC)

Minimum cable lengths

The section on minimum cable lengths is a mess. It has random links strewn throughout and a number of gramatical errors. I'm hesitant to change anything though, because I know very little on the subject. Perhaps someone else more knowledgable should have a look? Yanroy 21:13, 25 April 2007 (UTC)

As far as I can tell it is mostly wrong. There is no minimum cable length for xxxBase-T. The 802.3 spec clearly says so for 10Base-T, and for the others I see no reference to minimum length. I can't see any way that collision detection depends on a minimum cable length. As for 100Bse-FX, again IEEE says nothing about minimum length. Cisco mentions one, but doesn't give a reason. It can't be signal level; the attentuation on 2.5 meters of fiber is too small for that theory. So the only part that appears correct is the one about long haul transceivers used with short fiber (because the transmitter output power is higher than the max permitted receiver input power). Paul Koning (talk) 18:55, 25 February 2008 (UTC)

It appears that minimum cable length only comes into play when ethernet is used in a bus topology because collisions need to be detected in order to avoid data corruption. The explanation that I heard is that there is a race condition when more than 2 devices (or 2 half-duplex devices) are connected via a shared bus. Each transmitter needs a minimum amount of time to detect collisions along the shared transmit bus. Collisions are detected by noticing that the voltage level the transmitter wishes to put on the bus (to indicate a 1 or a 0) is not what it would expect due to constructive or destructive interference. Stated more explicitly, a collision is detected when the measured voltage of the bus is double the expected voltage for a bit or 0 volts when it is expected to be non-zero for the currently transmitting bit. If the cable is too short, there is a danger that this condition will pass undetected and data will be sent and/or received without any connected device realizing that it is corrupted. Most modern network configurations completely avoid deploying ethernet using the bus topology or use only full-duplex communication. Consequently, each transmitter has its own dedicated signal transmission lines and need not be concerned with collisions at all. This means that minimum cable lengths can be done away with in the star / switched context. I am reluctant to re-write this section without further citations to support this exposition of the electrical issues involved. If anyone cares to re-write the section, this link may prove to be a useful source: http://www.ctrlink.com/2006_07_01_archive.html —Preceding unsigned comment added by 164.67.97.133 (talk) 20:43, 1 June 2009 (UTC)

Agreed; the section is wrong: The main problem is that it confuses "collisions" with "standing waves" and "reflections" - a collision is when two nodes try to transmit at the same time and is handled in the media protocol with "collision detect" (on media with no full duplex operation). Reflections and standing waves are caused by impedance mismatch and are completely different. It might not be easy to tidy up as there's further confusion over media types; coax, twisted pair and fibre. I'll collate correct citations and edit soon. (shocked that something so wrong could be on the wikipedia for so long!)

--Qfissler (talk) 11:57, 30 August 2012 (UTC)

I've cleaned up the section. In general it is difficult to find reliable sources to say that something does not exist. I found many unreliable sources that said there both is and is not a limit. I believe it would be considered original research for WP editors to point out that there are no minimums specified in IEEE 802.3. My personal understanding is that, aside from the 10BASE2 minimum I've added with citation, there is no minimum. I did find one authoritative statement supporting this but since it is secondhand through an unreliable source, I'm hesitant to cite it. --Kvng (talk) 15:09, 30 August 2012 (UTC)

The following is from Adrian Young over at fluke Networks - Fluke Networks is the company that makes the ungodly expensive test gear for structured cable plants; ie the guys that do stuff like large installs.

I've talked to Adrian before concerning non-standard eth connectorization - his response was how (Big Name Aeospace company) has issues on the ISS using mil type with standard pins instead of impedance matched inserts that should be used.

Anyway here's a post from him on the BICSI forum; BICSI is the group that certifies structured cable plant installers:

"For Category 5e and 6, there is no minimum length requirement. ANSI/TIA/EIA-568-B.2-1 in Annex K does give a warning about reflected FEXT on shorter links with minimally compliant components. The obvious solution is not to purchase minimally compliant components. In the early days of Cat 6 when vendors were struggling to do better than marginally compliant, short links were an issue. Today, this is not an issue if you stay with a main stream vendor.

Within this same standard, there is also advice on distance when using a consolidation point. It advises a minimum distance of 5 m between the CP and TO. In ISO/IEC they are a little more clearer is specifying 15 m between the DP and CP. This is all for Category 6/Class E.

With regards to Category 6A, there is a minimum length requirement - kind of. In Annex J of ANSI/TIA-568-B.2-10 is describes worst case modeling using a 10 m link. The suggestion therefore is that you should not go less than 10 m. But again, that is with minimally compliant components. As with Category 6 stated above, there are now components available that will give you passing field tests below 10 m. HOWEVER, even vendors with good components may still have a minimum length requirement in their design specifications. The only way to know where you stand is to talk to the vendor AND test it to see.

If you are talking specifically about patch cords, then 0.5 m is the implied minimum length in ANSI/TIA/EIA-568-B.2-1 for a certified patch cord. That's because the math for the limit lines really does not work below this. Infact, getting a certified patch cord of 0.5 is going to be tricky. Many vendors only offer a certified patch cord of 1.0 m or longer. I suspect that this may be the most useful information with regards to your question.

Kind regards

Adrian Young Sr. Customer Support Engineer

Fluke Networks Technical Assistance Center 6920 Seaway Blvd, Everett, WA 98203"

Wamnet (talk) 21:09, 19 July 2017 (UTC)

Informal tone

The minimum cable length section has stuff like "This is normally 10-100 Mb stuff" and "So here's where it gets interesting". Doesn't sound like something I'd read in an encyclopedia.

Tagging for informal tone. 59.154.26.124 23:25, 1 May 2007 (UTC)

10x steps between revisions

I think this article would be the right place to answer an old question of mine: what motivates the exponential speed increase of 10x between each revision? Why can't we have 25 Mbps or 450 Mbps Ethernet? Exxos77 18:27, 2 May 2007 (UTC)

Economics and a pleasing sense of symmetry, I guess. For installed base, there's not much point in junking all your installed equipment for a 2.5x speedup. Generally speaking, the IEEE Task Forces have been given the job of creating the next Ethernet standard at 10x the speed but only 3x the cost. Having said that, IEEE are currently working on a 40Gbps Ethernet standard, so there goes the progression. It's not really encyclopaedic data, though, I think. Gareth8118 (talk) 13:00, 6 March 2008 (UTC)

IPoAC

IP_over_Avian_Carriers

Where does it fit? Humorous as it is, it is an actual data transfer protocol.--Marhawkman (talk) 22:56, 15 December 2007 (UTC)

It's a physical layer and data link layer protocol (not part of the Ethernet physical layer; the physical layer involves, as per the RFC, birds and sheets of paper). Guy Harris (talk) 23:19, 15 December 2007 (UTC)

Fair use rationale for Image:Ethernet73.gif

Image:Ethernet73.gif is being used on this article. I notice the image page specifies that the image is being used under fair use but there is no explanation or rationale as to why its use in this Wikipedia article constitutes fair use. In addition to the boilerplate fair use template, you must also write out on the image description page a specific explanation or rationale for why using this image in each article is consistent with fair use.

Please go to the image description page and edit it to include a fair use rationale. Using one of the templates at Wikipedia:Fair use rationale guideline is an easy way to insure that your image is in compliance with Wikipedia policy, but remember that you must complete the template. Do not simply insert a blank template on an image page.

If there is other fair use media, consider checking that you have specified the fair use rationale on the other images used on this page. Note that any fair use images lacking such an explanation can be deleted one week after being tagged, as described on criteria for speedy deletion. If you have any questions please ask them at the Media copyright questions page. Thank you.

BetacommandBot (talk) 19:42, 13 February 2008 (UTC)

Advertisements in Links

Links are filled with advertisements (Collection of product information from various vendors, related to ETH PHY only in business sense) and useless information (some guys are hugging and their bios...? I understand it's after a short description of a transciever product, but then again it's a product of a company). I am going to remove them. —Preceding unsigned comment added by Slidersv (talk • contribs) 02:27, 7 April 2008 (UTC)

Could physical layer discussion include an outline of the electrical problems being addressed?

Remarks removed by the original author. My comment is not good enough. Sorry, Lee_Mck

Actually there is a point here. Mostly the issues for each PHY choice are discussed in each one's article, since they are different issues for each PHY. (we need to explain that jargon, and mention MAC vs. PHY etc.) I suppose some overall issues might be discussed, and the Experimental Ethernet issues, which does not have its own. Probably does not need its own, but I made a few edits. The idea of 48-bit addresses, and CRC size, for example, might need to be discussed. I removed the link to Xerox Ethernet since that links back to the Ethernet article, but maybe the historic info in there might be relevant. It did get deployed outside of Xerox: the SUN workstation at Stanford had an interface, as well as a Unibus one for PDP-11s, and CMU and MIT had some. W Nowicki (talk) 23:06, 10 April 2011 (UTC)

The Living Computer Museum has an Alto (or two) on a 3Mb ethernet cable, with a bridge to 10Mb ethernet, running. They might also have a PDP-11 Unibus interface running. Gah4 (talk) 23:53, 29 January 2018 (UTC)

Citation needed on non-existant products?

In 100BASE‑T2 and 1000BASE-TX it is stated that no products have implemented those standards - the statement is also decorated with a citation needed. I'm wondering how anybody could possibly cite a source that would verify such a statement? The statement is true if no products exist and thus no sources to cite - unless someone wrote a paper on the lack of implementation. Bluecmd (talk) 16:52, 6 May 2011 (UTC)

You can interpret the decoration to indicate that the statement has been challenged. It can be resolved by an editor finding evidence of a product claimed to be nonexistent. There may well be secondary source out there that summarizes the fated of these technologies. I have not found it though. --Kvng (talk) 13:51, 8 May 2011 (UTC)

The generic "proving the negative" problem. For example, to "prove" elephants do not fly, one would need to examine every elephant over its entire lifetime for the existence of the species. One "solution" is to just not say it. The vast majority of what comes out of standards committees never results in products (although IEEE might have a better record that other bodies in this case; I do not know). But it might make sense to just drop these assertions. Claiming it is "unimplemented" is also stronger than "no announced products". Someone might have implemented it in a lab or some internal product, or a product that was just never picked up by the press. W Nowicki (talk) 17:18, 9 May 2011 (UTC)

Products usually means commercial products, though. Test setups in a lab don't count as products. I have some 100baseT4 transceivers, which actually made it as a commercial products, though not for long. Gah4 (talk) 23:50, 29 January 2018 (UTC)

10BaseS

This is listed without a source saying it is over Very high data rate Digital Subscriber Line (VDSL). A quick search turns up some fairly old products that used this moniker, but it seems a misnomer. VDSL is passband, not baseband I would think. So perhaps this was a commercial name used before 10PASS-TS was standardized in IEEE 802.3ah-2004? W Nowicki (talk) 20:30, 27 August 2011 (UTC)

Found it, evidently a trademark of Infineon, deevloped in 1997 and sold at least until 2002. I added some to Ethernet in the first mile article. Those PHYs need to be added here too. Also not sure how it relates to Long Reach Ethernet marketed by Cisco about that time. W Nowicki (talk) 23:28, 27 August 2011 (UTC)

Speculation

I removed the following paragraph from the introduction as it contains more speculation than fact. An encyclopaedia article should document what is, not what might be.

"10 Gigabit Ethernet is becoming more popular in both enterprise and carrier networks, with 40 Gbit/s and 100 Gbit/s Ethernet ratified. Higher speeds are under development. Metcalfe, one of the co-inventors of ethernet, now believes commercial applications using terabit Ethernet may occur by 2015 though he says existing Ethernet standards may have to be overthrown to reach terabit Ethernet."

109.154.64.107 (talk) 21:13, 15 October 2011 (UTC)

Firstly, can you please explain in the edit history when you remove information of this kind so people can see your reasons easily when considering the edit? I didn't see this until after I reverted. Having said that, I would have done so even if I had: there is no speculation here at all. 10GbE is being deployed quite widely now and 40/100GbE are ratified as the article states. Similarly, we are not speculating by reporting Metcalf's views and he is a reliable source for our purposes. Crispmuncher (talk) 21:58, 15 October 2011 (UTC).

References

HDMI Ethernet Channel?

What is the physical layer when Ethernet is transported over the HDMI Ethernet Channel? Can the the signal be split or separated from the HDMI cable? Apparently not, as the Ethernet signal is only transported over pins 14 and 19. — Preceding unsigned comment added by Petri Krohn (talk • contribs) 01:44, 14 November 2013‎ (UTC)

Its own 100Mb/s variant, if U.S. Patent 8.472,351 is to be believed. Guy Harris (talk) 02:44, 14 November 2013 (UTC)

This patent describes 100BASE-TX adapted through use of a hybrid circuit to convert from 4 to 2 wires. ~KvnG 16:50, 21 November 2013 (UTC)

HDMI_Ethernet_Channel#ARC_and_HEC says it is 100BASE-TX. ~KvnG 21:24, 17 November 2013 (UTC)

It's just using two pins/one twisted pair, that can't be 100BASE-TX... -- Zac67 (talk) 07:37, 18 November 2013 (UTC)

What HDMI_Ethernet_Channel#ARC_and_HEC says is "The physical layer of the Ethernet implementation uses attenuated 100BASE-TX type signals on a single twisted pair for both transmit and receive.", emphasis mine; it doesn't say "it's 100BASE-TX", it says it's "100BASE-TX type". Guy Harris (talk) 07:46, 18 November 2013 (UTC)

Someone should actually find out what that means before we proceed. It might be what's described in the patent cited above. It might be BroadR-Reach. It might be something else. ~KvnG 16:44, 21 November 2013 (UTC)

It is now almost four years later, and this still hasn't been added to the tables on this page. As well as I know, it is as stated above: 100baseTX signals on a single twisted-pair. For the distances allowed by HDMI, there should be enough signal to make this fairly easy. It is much harder to detect a small incoming signal in the presence of a large outgoing signal, as 1000baseT does. Gah4 (talk) 19:42, 14 September 2017 (UTC)

There is reasonable coverage at HDMI_Ethernet_Channel. It is part of HDMI, not stand-alone Ethernet so it is not clear it should be included in the Fast Ethernet table. Also the references at HDMI_Ethernet_Channel don't fully support the statements made in the article there. I have marked this. ~Kvng (talk) 15:27, 17 September 2017 (UTC)

2.5Gb Ethernet

2.5Gb Ethernet should probably be included here somewhere. While it's not an 802.2x standard, a number of vendors are shipping product (Broadcom and Intel, for example). It has the advantage of being compatible with a fair chunk of the physical infrastructure out there (unlike 10Gb Ethernet, it'll actually run on most Cat-5e wire), and appears much easier to implement in silicon. Rwessel (talk) 18:37, 29 January 2014 (UTC)

 Done Ethernet_physical_layer#2.5_and_5_Gbit.2Fs ~Kvng (talk) 15:45, 21 September 2016 (UTC)

"Ethernet cable" and similar redirects

Ethernet cable redirects here, as do Ethernet Cable and Ethernet network cabling.

Ethernet cables used to point to Ethernet over twisted pair, but I retargeted it here for consistency. Is there consensus that this makes sense?

I am starting to think that I should have done the opposite, that is, retargeting them all to Ethernet over twisted pair, as that is the most specific article that covers what we colloquially call an "Ethernet cable." (Perhaps the situation may be different many years in the future, when fiber-optic cables become prevalent in consumer networking equipment.)

Alternatively, perhaps Ethernet network cabling should point to this article, as that particular wording distances it from the colloquial usage, while all the others should point to Ethernet over twisted pair. Any suggestions? --SoledadKabocha (talk) 05:04, 18 August 2016 (UTC)

I think it makes sense for Ethernet cable, Ethernet Cable, and Ethernet cables to redirect here - it's all cable, even the stuff that's not twisted pair. Given that, I'm fine with Ethernet network cabling redirecting here as well. That's not only past-proof but future-proof. Guy Harris (talk) 06:20, 18 August 2016 (UTC)

Agreed. We should consider that WP is an encylopedia, not a consumer Q&A. As long as the latter is no problem it's fine but the former is more important. Alternatively, we could use a small page to direct the reader to the page he's looking for. --Zac67 (talk) 09:31, 18 August 2016 (UTC)

Ok. I should have realized, based on my need to use the word "colloquial," that my doubts were unfounded. --SoledadKabocha (talk) 23:19, 18 August 2016 (UTC)

I was really thinking about an disambig page for the various Ethernet cabling, but Networking cables already exists. How about redirecting all the above redirects there instead of here? I think it's more like what the reader expects. --Zac67 (talk) 15:51, 19 August 2016 (UTC)

Neither this article or Ethernet over twisted pair are really about the cable; They are about what the cable connects to. Also cable can mean copper or fiber-optic cable. I think it best to redirect all of these to Networking cables as Zac67 has suggested. ~Kvng (talk) 13:48, 22 August 2016 (UTC)

That's even better. Guy Harris (talk) 17:05, 22 August 2016 (UTC)

 Done ~Kvng (talk) 20:39, 24 August 2016 (UTC)

So where is the "Ethernet Physical Layer" information?

I came here to learn about Ethernet Physical layer. What did I find? Nothing.

No description of Transmit or Receiver circuits. No voltage levels, no nothing.

As usual for Wikipedia, it launches into arcane networking gobbledygook, but with absolutely no description of the underlying electronics.

Can I build an Ethernet tap with the information provided? Of course not.

Gutta Percha (talk) 10:02, 18 June 2017 (UTC)

Please add new sections on bottom (the New section does that for you).

Most readers and editors are interested in practical information, ie. what you need to set up a link and what the differences between the PHYs are. Building your own Ethernet hardware from scratch isn't too practical nowadays. Non-ancient Ethernet copper variants use differential signaling, so there are only maximum voltage levels which aren't really that interesting. --Zac67 (talk) 10:42, 18 June 2017 (UTC)

Terabit Ethernet may occur by 2015?

Zac67, I tagged a sentence for updating, but you reverted saying "historical cite doesn't need update". I disagree, it needs update. See:

"In 2008, Robert Metcalfe, one of the co-inventors of Ethernet, said he believed commercial applications using Terabit Ethernet may occur by 2015, though it might require new Ethernet standards."

As a reader, the question that arises from this sentence is: did it occurred or not? We are already in 2017, 2 years later. Did it occurred?--200.223.199.146 (talk) 14:13, 21 July 2017 (UTC)

It needs an addition, but the statement is from the past, and clearly so – it's about where the term "Terabit Ethernet" comes from as it has yet to be developed. You're right that it needs an update, I was mentally more at the Terabit Ethernet article where someone made a similar but inappropriate change. Feel free to improve my change and thanks for the heads-up. --Zac67 (talk) 17:21, 21 July 2017 (UTC)

Skinny columns

Zac67's new skinny columns are upsetting my aesthetic sense. I don't mind adding a new column to indicate media type and I beleive there's room for it. Let's please not do it like this
c
o
a
x
i
a
l
And let's have a column heading and capital letters and no cryptic abbreviations. ~Kvng (talk) 13:47, 21 September 2017 (UTC)

I'm not too happy with it either but couldn't find another way. {{vert header}} didn't work here and horizontal text wastes so much space. Maybe colored cells? --Zac67 (talk) 17:50, 21 September 2017 (UTC)

Separate tables, with the table caption indicating the medium? They're not currently sortable, so that won't break sortability.

Or could the tables have subheads? If there's nothing in the table syntax to support that, perhaps there could be spanned rows, in the same shade of gray as the table heading. Guy Harris (talk) 05:18, 24 September 2017 (UTC)

I had something more compact in mind but subheads are a good idea. --Zac67 (talk) 07:38, 24 September 2017 (UTC)

Thanks Zac67. Looks much better now. ~Kvng (talk) 15:28, 27 September 2017 (UTC)

External links modified

Hello fellow Wikipedians,

I have just modified one external link on Ethernet physical layer. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

• Added archive https://web.archive.org/web/20131219024707/http://www.virtual-strategy.com/2012/06/11/l-com-introduces-commercial-grade-thinnet-10base-2-and-thicknet-10base-5-converters-legac to http://www.virtual-strategy.com/2012/06/11/l-com-introduces-commercial-grade-thinnet-10base-2-and-thicknet-10base-5-converters-legac

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Y An editor has reviewed this edit and fixed any errors that were found.

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 03:52, 24 September 2017 (UTC)

Amperage

Nothing in this article discusses the actual typical current or amps used on ethernet physical layers. This information is difficult to find, and it would be nice if it were explained on this page.

~Davemc50 21:50 29 January 2018 (NZDT)

Are you referring to Ethernet over twisted pair? EoTP uses differential voltages for signalling. The currents on the wires are somewhat negligible. --Zac67 (talk) 11:50, 29 January 2018 (UTC)

UTP uses a few volts for signalling, and the cable impedance is about 120 ohms. So, maybe 20mA. Gah4 (talk) 17:55, 18 May 2019 (UTC)

Grouping by speed or annex?

The 100 and 200 Gbit/s PHYs are currently spread over multiple sections due to being developed in groups (based on the lane types), ie. for historical reasons. I'd like to regroup them solely based on the speed. Any objections, thoughts? --Zac67 (talk) 08:30, 20 May 2018 (UTC)

Most of these tables are repeated in some form in the associated {{main}} articles so a lot of it can be gutted. Once that is done, any proposed reorganization will be easier. ~Kvng (talk) 18:35, 23 May 2018 (UTC)

New "Overview" table

@Nightwalker-87:, your non-TP table shows considerable effort, provides some interesting additions, and I was fighting myself for what to do with it. I finally reverted for multiple reasons:

• There are many errors ("FC" column header for non-fiber; 10BASE5: BNC; fiber grade compatibility; fiber connectors are not usually part of the standard; 10GBASE-CX4: X2,XENPAK; Media count vs lanes; ...)
• various missing information from the already present tables
• While some columns are useful, the table itself is mostly redundant with the already present tables.
• The Symbol coding... column requires significant changes (esp: PCS coding vs PMA coding)
• A single table makes sense but is too large, IMHO. It also misses the chance for short class summaries.

Essentially, there were three options with their respective problems:

• remove the old tables – too much information is lost, too many errors in new table
• integrate both schemes – a lot of work I wasn't fond of
• remove the new table (and possibly use a revised version later)

We need to focus on which information is actually needed in this article. Many PHYs have complex coding and operational parameters – I'd like to leave that to the more specific articles and only summarize here. OTOH, it makes sense to provide a comprehensive list of all PHYs here which doesn't exist elsewhere.

Maybe it's better to reduce the table to provide a more specific summary of fiber PHYs, similar to the Twisted-pair section? Then again, we've got multiple tables with much redundant information. In summary, I think we should move over interesting details to the present tables and leave it at that.

--Zac67 (talk) 08:06, 29 July 2018 (UTC)

@Zac67:, thank you for your positive feedback. Indeed no information should be lost. It was also not the intention to simply delete the old tables instead. The aim is/was to introduce a common layout with unified formatting. I locally maintained such a table-sheet for quite a while out of personal interest and collected all info I could find. (Though I'm not a professional or expert on this topic.) I recently decided to share this info on wikipedia for the benefit of all. I have started and formatted the table an a personal user page first: User:Nightwalker/IEEE 802.3. In this context I'd prefer to adress any issues of yours (some are very clear to me already) and further include currently missing content from the old tables (if relevant) at this place, while you monitor and give me further feedback. If style and content is considered to be sufficient, one could start a new attempt. This would avoid redundant information in the article while having it improved. How do you think about that? I'd clearly like to do the right thing and help contributing. Nightwalker-87 (talk) 10:39, 29 July 2018 (UTC)

Excellent. Where do we start? --Zac67 (talk) 12:52, 29 July 2018 (UTC)

I'll edit the list @User:Nightwalker/IEEE 802.3 first by taking into account some points that are clear to me. If you bookmark it, you can keep an eye on it and comment on the related talk page. Then we can decide how to proceed or when to move/replace content in the article. If you prefer, you can roll back the edit in the main article in the meanwhile. Would that be a reasonable approach for you? Nightwalker-87 (talk) 19:31, 29 July 2018 (UTC)

I think it's better to prep the draft on your page for the time being, it'll take some time. --Zac67 (talk) 21:01, 29 July 2018 (UTC)

T2 and T4

As well as I know, T4 got actual products to market, and I presume some were installed. I have some actual T4 transceivers. I don't know that any T2 device was marketed, but then again, I didn't look very hard. I think this means that they should be described differently in the article. Gah4 (talk) 17:52, 18 May 2019 (UTC)

Groups and standards

Ethernet physical layer#100 Gbit/s says 10 and 25 Gbit/s lanes and 40 and 100 Gbit/s Ethernet are in IEEE 802.3ba and 50 Gbit/s lanes and 50 and 200 Gbit/s Ethernet are in IEEE 802.3cd. This isn't meshing well with what's said in 25 Gigabit Ethernet and Terabit Ethernet. ~Kvng (talk) 21:06, 4 October 2021 (UTC)