Talk:56 kbit/s modem

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Computing (Rated Start-class, High-importance)
WikiProject icon This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
 High  This article has been rated as High-importance on the project's importance scale.

Is 56 kbit/s really the fastest?[edit]

Have there been any laboratory experiments with faster rates over phone lines? - Theaveng 12:22, 27 September 2007 (UTC)

It depends what you consider to be a phone line. Copper phone pairs can carry far more than 56K and indeed to in many parts of the world (see DSL), the 56K limit comes from the structure of the modern digital phone systems. Plugwash 13:21, 29 September 2007 (UTC)
The 56Kbps limit comes from the infrastructure. It is common for the core telco system to use Robbed Bit Signaling (RBS) - so a 64Kbps DS0 circuit 'loses' 8Kbps to signaling, leaving 56Kbps for the voice content - or the modem. I used to work for Livingston Enterprises, and later Lucent, maker of the PortMaster line of access servers. I was there for the introduction of PCM modems - Livingston supported K56flex with Lucent chips, which is a factor in why Lucent acquired us. (We tried Rockwell's chips - but they were too hot and too unstable at the time.) I've personally seen K56flex modems hit speeds over 56Kbps, and theoretically a PCM modem could hit 64Kbps, if the line is a clear 64Kbps circuit - such as an ISDN B-channel which doesn't have RBS. But when you're talking a POTS line 64Kbps is the theoretical max raw rate, that's what the DS0 carries. The copper itself can be used for other protocols - like DSL - but that's a different story entirely. (Lucent's original V.flex2 proposal was the most advanced of the three, with things that finally made it into v.92 - like PCM uplink. But for strategic reasons they partnered with Rockwell to take on USR, and Rockwell's K56Plus was the weakest - so the K56flex compromise sacrificed some features for the partnership.) MegaZone (talk) 06:59, 28 July 2009 (UTC)
I don't think the original question (thrust) was explicitly answered. I believe the thrust was, "are faster speeds for dial-up really technically impossible?" All replies here and elsewhere seem to say 56k is the max on a 56k system. True, but doesn't that dodge the thrust? That 56k system includes technology economically motivated to never exceed 56k, AND US federal regulations capping it at 53-56k. Before the regulation, modem speeds were doubling every year or two. Everybody assumed this would continue. At the time, software settings for max port speed settings went much higher than 56k, based on that assumption, in fact even my current PCI SoftV92 dialup modem still has a 115,200k max port speed setting (which is where I uselessly have it set, (since no ISP would dare attempt to exceed a 56k dialup connection with me)). That FCC regulation killed off the multitudes of local mom and pop ISPs by forcing users to subscribe to the high speed phone companies. How many coincidences does that make? And what was the telco lobby doing regarding this regulation? I'm a big believer in simple economics. I'm pretty sure simple economic forces are often more powerful than government, technology, etc, and might even create a conventional wisdom now and then.

quotes: ...1971 article, “The Theory of Economic Regulation,” ... “capture theory.” Stigler argued that governments ...regulate at the behest of producers who “capture” the regulatory agency and use regulation to prevent competition. ...(for his) work on the effects and causes of regulation, Stigler was awarded the 1982 Nobel Prize for economics.

Under "Speed" the article says:

Each conversion adds noise, and there will be too much noise from the second conversion for 56k to work. The modem's negotiation processes will fall back to the less demanding 33.6 kbit/s mode. Other local loop conditions, such as certain antiquated pair gain systems, may have similar results.

Isn't all that only true due the system only attempting a software-limited 56k max, and the 33.6k is SOFTWARE selected? And similar "failures" would occur over the same lines if only 33k was the self-limiting max limit...33k could NEVER be achieved due to the above lack of perfection? (But this is NOT just here, this type of questionable reasoning or vague explanation seems universal across the internet on this topic. There are no crisp, bullet proof explanations that exclude economic forces (nor FCC regulation) for the limits...which I think was the "lab" (ie; science or physical limits) question the original poster was asking.)
And didn't the FCC regulation economically kill all attempts at developing faster dialup modems AND LINES? The universal explanations all seem circular: "We reached the theoretical limit approaching 53K on systems that never attempted to exceed a theoretical (or regulatory) limit of 53K." I think in the real world, "never attempted to exceed..." is equiv to: designed to never exceed....
I don't mean to imply that a PR funded Winston Smith has been patrolling the Internet with his memory hole as simple economics would demand if my insinuations were true. However, I do notice that the article does not even contain the words "fcc" "f.c.c." nor "regulation." After all, Winston Smiths can be motivated by such things as self-image and employer or team loyalty. Yes, I do NOT believe in coincidence.
-- (talk) 20:02, 28 June 2012 (UTC)Doug Bashford
Surely the Federal Communications Commission article's talk page is more appropriate to discussions of that agency, but is someone denying that Primary Rate Interface is in everyday use? How odd. Jim.henderson (talk) 13:09, 1 July 2012 (UTC)
Yes, ISDN is available, but that's not what a narrowband modem works on - narrowband modems are designed to work on an analog PSTN, with the only gain from a digital network being lower line noise. Since 64Kbps is the fastest available on an analog telephone network, and 8Kbps are stolen to avoid an erroneous null signal, 56Kbps is the fastest data transmission over a single analog line of a PSTN. Sure, you could have multiple lines; sure, digital could have faster speeds; sure, the copper itself could carry faster data (DSL does). But a 56Kbps modem works on a single analog copper voice-only line, and hits the maximum speed on that line. (At the risk of furthering US bias in the article: Only 47 of 1000 Americans have installed ISDN lines at all, so no, ISDN isn't in everyday use in the US.)
Freeone3000 (talk) 21:40, 26 September 2012 (UTC)

Is error-correcting codes included or excluded in the bit rates for V.92 modems?[edit]

Does v.92 modems have Forward error correction (FEC), a.k.a. error-correction codes? Or does it only have error detection and automatic repeat request (ARQ)? Perhaps in slower modes than the 56kbps mode? And it that case, is the FEC code included or excluded in the bit rate indicated by the operational system?

I ask because I wrote the following in the bit rate article. Is it correct?

The connection speed of a network access technology or communication device is indicated by some operational systems. The connection speed of a technology that involves forward error correction typically refers to the physical layer net bit rate. ...
However, the connection speed of maximum 56000 bit/s in the downlink of a V.92 modem refers to the gross bit rate (the line rate), while the modem data transfer rate (the average useful bit rate) sometimes can be higher higher due to data compression, and sometimes lower due to bit-errors and retransmissions.

Mange01 (talk) 19:11, 26 August 2008 (UTC)

Yes, but we should remind readers that the 56K figure is an ideal or fantasy. Ignoring the overhead of v.42 error correction is only part of what makes it unreal. Jim.henderson (talk) 15:32, 1 September 2008 (UTC)

It's not unusual at all for people living a reasonable distance from the central office or a DLC to get 53K. Squidfryerchef (talk) 03:43, 2 September 2008 (UTC)
True enough, but the section in question suffers from more problems than that. The v.42 overhead paragraph presents an unsourced calculation based on a 56K line rate, which never actually happens. This paragraph should come after one that clearly explains the various line rates, the conditions that create them, and the likelihood of achieving them. Then there is a paragraph discussing the 115.2K port rate in a way that suggests far greater relevance than actually applies. It's a very poor section and likely to mislead many who don't already know about this stuff. Jim.henderson (talk) 01:45, 8 September 2008 (UTC)
Thnx for your response. Please check the rest of the discussion in the bit rate article about net bit rate and gross bit rate, and connectin speed, not only for phone modems. What speed does MS Windows show, i.e. what is the connection speed, when we are using 56K modems? Does it show 56K or 53K at most? Are there any forward-error correction codes in some modes of a v.42 modem? The error coding you talk about, is that error-detecion, used for ARQ? Mange01 (talk) 21:23, 24 September 2008 (UTC)
Ideal perhaps, but I wouldn't call it fantasy. I've seen a great many solid 56Kbps K56flex & v.90 connections 'in the wild', from my time working in the industry. You certainly can't expect it all the time, but it is certainly possible in the real world if conditions are good. There is an oft repeated myth that the FCC imposed a power cap which limits PCM to 53.3K. That's marketing crap some modem vendors started back in the day because *their* modems couldn't get past 53.3 without violating the power level limits, but it was not a limitation of the protocol, just their chipset. Others were hitting 54.6 & 56 with the same protocol - or higher on a clear DS0 link, though you'd be luck to catch that in the wild. MegaZone (talk) 07:10, 28 July 2009 (UTC)
I guess you didn't notice but our article in fact repeated this myth. I did some searching, and it appears you are correct so I've added a poor source (the original claim was unsourced as with most of the article). If you are aware of better sources, your welcome to improve the article Nil Einne (talk) 17:47, 19 September 2009 (UTC)

Ten bits per byte[edit]

Umm, this is sometimes true but not usually. It's part of a paragraph with precise calculations based on assumptions that are sometimes true but not usually. Seems to me, this article ought to discuss, not user throughput, but line rate, with links to articles about the various factors besides line rate that affect throughput. On the other hand, if someone has published a survey of actual results with 56K modeme, rather than calculations, then it would belong here. Jim.henderson (talk) 04:46, 16 December 2008 (UTC)

Well I haven't taken a survey, only made observations of my own experience. Back in the days of 2400 bit/s modems, they did use 10 bits per byte per the settings (8N1 aka 1 start bit, 1 stop bit, and 8 data bits) with a throughput that was only 240 bytes/second. ---- In today's world the maximum download speed I can get on a 53.3 kbit/s connection is only 5.1 or 5.2 kbytes/second. If it were 8 bits/byte then it should be getting 6.6 kbytes/second but that never happens, so I can make an engineering estimate that the modem is using the functional-equivalent of 10 bits per byte just like the old 80s-era modems used. ---- Theaveng (talk) 13:40, 21 April 2009 (UTC)
A possible but not likely explanation. Almost all dial-up Internet connections by "56K" modems are v.42 synchronouse ones, with an overhead of about 4%, not start-stop asynch connections with overhead of about 20%. The changeover to synch mode happened about the same time, and on the same modems, as the increase in line rate from 9.6K to 14.4K. Much more likely you, like everyone else with "56k" modems, are getting not merely a lower line rate than advertised by the name, but lower than the 53k theoretically possible. Namely, some 40-45b/s. Jim.henderson (talk) 03:50, 25 April 2009 (UTC)
Ah. A little research shows that it's worse than I thought. Yes, the calculation is incorrect, not arithmetically but in the assumption that 8N1 or similar start-stop protocol is in effect rather than Link Access Procedure for Modems. LAPM is almost always used in the past decade. Even were the calculation correct, a custom made engineering estimate would violate WP:SYNTH and have to be removed on that ground. So, no, all that calculation must be deleted as both erroneous and WP:OR. Jim.henderson (talk) 00:32, 28 April 2009 (UTC)
First off, when I connect it clearly states I have a 53.3 kbit/s line. So if we follow your reasoning, I should be seeing 53.3 kbit/s divided by 8 bit per byte == 6.6 kbytes/second download speed. In over ten years of using these modems, I have never seen that kind of speed. They always max-out at around 5.1 or 5.2 kbytes/second.
I have seen lower connections, like 44.0 kbit/s while traveling, and yes that translates to 4.1-4.2 kbyte/sec as you would expect with a 10 bit per byte protocol. Can you give us a GOOD explanation for that discrepancy between your assertion that there's 8 bits per byte, and the apparent 10 bits per byte seen in the real world? ---- Theaveng (talk) 17:17, 4 May 2009 (UTC)
Drat; been taking too many outdoor pictures in the nice weather and will be busy integrating them into other parts of Wikipedia and won't give this bit of electronic antiquity the time it needs. Yes, 8N1 remains commonplace for the DTE connection, that is computer to modem, since bandwidth is plentiful, thus overhead and error rate aren't important in that link. LAPM is used for the modem to modem connection through the phone line, where those are important considerations. The EC feature of v.42 is constantly at work, using in most connections a significant percentage of the channel capacity and I figure those are the missing bytes you are misattributing to the excess overhead of inefficient framing.
Anyway if you can find time to link reliable sources that say I'm right or wrong I'll be delighted to see them included in this article; otherwise the matter should be omitted. Eventually I'll photograph and upload everything of encyclopedic interest in my neighborhood and have time to devote to the joys of voiceband modemming. Meanwhile have fun and don't forget the perils of WP:OR. Jim.henderson (talk) 15:45, 7 May 2009 (UTC)

Comment originally in the article[edit]

In 8-N-1 connections (1 start bit, 8 data bits, No parity bit, 1 stop bit), which were typical before LAPM became widespread, the actual throughput is a maximum of 5.6 kilobytes per second, since ten bits are used for every byte, although effective throughput can be increased as high as 32 kbytes/s using internal compression, or 100 kbytes/s using ISP-side compression.

^ The above statement is incorrect, this applies only to the COM port to MODEM link, not the MODEM-to-MODEM or MODEM-to-DIALING RACK link, and of course from MODEM or DIALING RACK to COM port(s) again. The MODEM to MODEM link strips out the start bit, stop bit, and any 'COM port' link parity. This is almost purely a UART isolated form of serial communication.

The comment above was given by as of 06:51, 15 March 2010 (UTC) [1]

--Quest for Truth (talk) 13:16, 14 April 2010 (UTC)