Talk:RS-232/Archive 1

FALTY TEXT

Data signals for RS 232

Level Transmitter Receiver Logical 0 +5 V to +15 V +3 V to +25 V Logical 1 -5 V to -15 V -3 V to -25 V Undefine -3 V to +3 V

http://www.camiresearch.com/Data_Com_Basics/RS232_standard.html http://hw-server.com/rs232-overview-rs232-standard http://www.lammertbies.nl/comm/info/RS-232_specs.html —Preceding unsigned comment added by 212.181.101.10 (talk) 09:10, 15 February 2011 (UTC)

Analog vs Digital

Removed pending confirmation:

(Note that bits-per-second is not generally the same as baud; when techniques like Quadrature Amplitude Modulation are employed, more than one bit can be transmitted per baud. In most RS-232 applications, only very low bitrates have one bit per baud.)

Forgive me if I'm wrong, but isn't this person getting analogue modem signals confused with digital serial signals? -- Tim Starling 07:29, 25 Aug 2003 (UTC)

Sampling rates at the central office for voice lines prevent the baud rate from surpassing about 4 k baud; but the bit rate is higher because discrete signals represent multiple bits. But yes, this seems to go in another article.Waveguy 22:57, 25 Aug 2003 (UTC)

The quote above is correct, but probably doesn't belong here, because of what Tim said. It's really more relevant to the modem than to RS-232, although the two were almost synonymous in home computing (for the average user, at least), especially back when most modems were external. --radiojon 08:12, 2003 Sep 5 (UTC)

Serial Communications

The discussion of Serial communications would be better in that article (which also needs work). --Wtshymanski 21:34, 10 Apr 2005 (UTC)

Started tuning up the RS 232 article. Please leave the commented text till I have a chance to move it to serial communications.

You know, RS 232 doesn't really define a "bus" as I understand the term - it's more of a point-to-point connection, which is indeed one of the limiations I refer to in my contributions. --Wtshymanski 04:21, 11 Apr 2005 (UTC)

Name change from RS-232 to EIA-232

I wonder whether anybody would disagree if i change this article's name to EIA-232. Currently, a search for EIA-232 is re-directed to RS-232. IMHO, i think it should be the other way around. ie RS-232 should be re-directed to EIA-232. To support my opinion, below is a foot note from "Unix System Administration Handbook" 3rd edition page 94.

To be technically correct, RS-232 should be refered to as EIA-232. However, no one will have the slightest idea what you are talking about.

Please, say something as i plan to change the name sometime in future if this post doesn't generate objections.

OK, I object. The footnote you quote is true: Although EIA-232 is technically correct, nobody ever calls it that. Google gives 1,600,000 hits for RS-232, and only 57,100 for EIA-232; it's obvious the new name was never really accepted. And there are a lot of present (and probably future) articles that link to RS-232 that would need to be changed; not worth the effort. --Rick Sidwell 20:08, 3 May 2005 (UTC)

Object - RS-232 port is the more frequently used terms, though I've also seen it designated "EIA RS-232". Since this is what the standard was known as during it's heydey, I don't think a name change improves the utility of the article. A redirect from "EIA 232" to this article would work. --Wtshymanski 00:15, 4 May 2005 (UTC)

Thanks for feedback. I have to give in as i agree that changing the name will have too much ripple effect on other articles. gathima 00:33, 11 May 2005 (UTC)

• Why is EIA-232 correct? SchmuckyTheCat 02:04, 19 May 2005 (UTC)

It's not any more correct than RS 232. The current standard is TIA-232-F Interface Between Data Terminal Equipment and Data Circuit-Terminating Equipment Employing Serial Binary Data Interchange (ANSI/TIA-232-F-1997 (R2002) (see [1]- so changing the article title to an intermediate, also obsolete, description seems bootless. TOo bad they don't say "Recommended Standard" but when you think about it, what other kind could there be? I would expect that most of the serial ports out there are labelled "RS 232" since revision C was the mostly widely referenced, though often ignored, edition of the standard during the Great Computer Boom of the latter 20th century. --Wtshymanski 14:05, 19 May 2005 (UTC)

• That's what I thought. Thanks. SchmuckyTheCat 17:42, 19 May 2005 (UTC)

Personally I think the article should be called by the correct name (TIA-232) and redirect pages from RS-232 and EIA-232 put in place. The redirects will take care of the "no one will know to look for it" problem. An encyclopedia should not be so quick to bow to popular usage. However I'm not strongly motivated enough to do the change myself. Jeh 06:08, 16 June 2007 (UTC)

I still don't think it's a good idea - as I said two years ago. If the port has a label,it's usually labelled "RS 232", not "TIA-232". Let's call it what it's called. --Wtshymanski 03:26, 17 June 2007 (UTC)

I certainly agree with the people who say that RS-232 is the most recognized name. If you said to me "EIA-232", I would have to stop and think, but as long as I had the right context, I would figure it out pretty fast. If you said, "TIA-232", I would probably have no clue. But, that's not the point. The proper name of the standard today is apparently TIA-232, so that's what the article should be called. The other names should certainly exist as redirects, so normal people can find it when they do a search. -- RoySmith (talk) 22:05, 17 June 2007 (UTC)

What a thing is called is what people call it. Observe that a Google search on "RS232 -TIA 232" turns up literally millions of hits, while a search for "TIA 232 -RS-232" turns up "only" 50,000. What's written on the back of the box is usually RS 232 (if it's labelled at all), and the article mostly talks about the EIA days when it *was* RS 232. Please don't rename the article, make TIA 232 a redirect instead. Why give the article a name that is willfully obscure? --Wtshymanski 14:23, 18 June 2007 (UTC)

Encyclopedias should favor correctness over popularity. Calling it TIA-232 won't make it obscure; the RS-232 entries will still exist as redirects, so anybody who types that in will get to the right place. -- RoySmith (talk) 14:46, 18 June 2007 (UTC)

Agreed, encyclopedias should be correct. The correct name of a thing is what people call it. When you look up "cat" in the encyclopedia, is it under the correct species designation or under "cat" ? Correct is what people do; the TIA standards committee is a very very small subset of the people who refer to some kind of XXX-232 compatible port every day. And up to 1990 or whenever EIA renamed itself, RS-232 *was* the only name. Don't let this turn into another "kibibytes" debacle - Wikpedia is prone to historical revisionism! If the redirect is adequate, leave TIA 232 a redirect for the dozens who type that in, and leave RS 232 for the millions. --Wtshymanski 15:06, 18 June 2007 (UTC)

Somewhat off-topic, but why do you think kibibytes is a dabacle? -- RoySmith (talk) 16:07, 18 June 2007 (UTC)

Moved here from inappropriate section above:

I was redirected from my "EIA-232" search to the RS-232 page. This seems backwards. RS stands for "Recommended standard." EIA-232 is the standard that superceded RS-232. Shouldn't the page be renamed EIA-232 (or TIA-232), and searches for RS-232 be redirected to the EIA-232 page (or TIA-232)? I know RS-232 is what everyone knows it by, but it is no longer a "recommended standard." JackTinWNY (talk) 17:50, 31 March 2009 (UTC)JackTinWNY

Have a look at the debate above. Is there anything not covered there that you feel worth mentioning? CrispMuncher (talk) 18:07, 31 March 2009 (UTC).

Why?

Why is the {{ipstack}} template here? It doesn't clarify the article any and most uses of RS 232 have nothing to do with Internet protocols. Would anyone object if I made it go away? --Wtshymanski 03:51, 24 May 2005 (UTC)

I agree; most uses of RS-232 have nothing to do with IP. The template keeps getting added back, probably because RS-232 was included in it... --Rick Sidwell 19:45, 5 November 2005 (UTC)

i've seen a similar thing happening over at ppp (i personally think that one is on the oppoisite side of the line to this one but you see similar slow addition and removals there too). The problem is the case of protocols that can be used as part of a tcp/ip setup but can also be used outside of it. Plugwash 21:09, 5 November 2005 (UTC)

What type is an RS-232

Is an RS-232 a Duplex a Half-Duplex or a Simplex, and why. I'm at a loss here.

Since both the transmit signal and recieve signal can operate at the same time, you could say the interface as a whole is full-duplex; each circuit, though, only ever operates in one direction and so is simplex. --Wtshymanski 14:16, 31 October 2005 (UTC)

It is agnostic with regards to duplex-ness. Historically, it was used full duplex on low speed modems and half duplex on high speed ones. Individual pins on the interface are uni-directional. -- David Woolley (talk) 23:46, 30 December 2008 (UTC)

RS 232 is half duplex it can only manage communication at one direction at a time were as RS 485 manages communication in both direction simultaneously so it is full duplex. —Preceding unsigned comment added by 68.221.216.140 (talk) 21:28, 5 February 2009 (UTC)

Err... no. See the above for a slightly more complete answer but a complete RS-232 implementation is certainly full duplex, although of course all signals may not be present in a given implementation. The individual singals are simpelx rtaher than half duplex because they can't reverse themselves.

RS-485 OTOH is a real half duplex link. If you need full duplex communication on RS-485 then you need at least two links - full duplex is tricky to define on a multidrop bus.. CrispMuncher (talk) 22:11, 5 February 2009 (UTC)

To the IP editor - CrispMuncher is correct. RS232 absolutely supports both directions, literally at once. I've written enough drivers and applications for RS232 interfaces to know this first-hand. Now it is true that some early modems that had RS232 interfaces were half-duplex - I think one of them was the Bell 202? - and some of the signals you find in RS232 are there to manage that half-duplex channel. But that doesn't mean that RS232 itself was ever limited to just one direction at a time. As far as RS232 itself is concerned, the transmit data and receive data signals are completely independent of one another and can be used exactly simultaneously. Or to put it another way... this word, "simultaneously"... I do not think it means what you think it means. ;) Too bad I didn't save some of those serial line analyzer traces... Jeh (talk) 10:16, 6 February 2009 (UTC)

framing of bytes in the stream?

If RS-232 doesn't specify this does some other standard do it or are there just some common conventions? it sounds like it would be pretty important for interoperablity. Plugwash 12:34, 4 November 2005 (UTC)

There are several different line codes that frame bytes in the stream in a variety of different ways -- for example, HDLC. But they are in the tiny minority of RS-232 interfaces.

The vast majority of RS-232 interfaces use ASCII with "start" and "stop" bits around each character. The universal asynchronous receiver transmitter mentions this is "by convention", but doesn't give a name to that convention. The start-stop transmission article seems to imply it is part of the ASCII standard, although the ASCII article doesn't mention it. What is the name of this line code / convention? Who standardized it?

I don't know who standardized it, but I can tell you that it dates back to five-level teletype machines... at least! It is not part of the ASCII standard; ASCII has to do with what the eight data bits mean, not how the bits are framed. Jeh 14:50, 30 June 2006 (UTC)

Style

If something is interpreted "idiosyncratically", it's also "selectively" though not necessarily the reverse - I think the phrase is redundant. Similarly, "DTE Equipment" expanded as "Data Terminal Equipment Equipment" would be clearly redundant...so I think "DTE Unit|Device" is also redunant. --Wtshymanski 23:07, 22 January 2006 (UTC)

RS-232 connector wire colors

I had to find which pin on a DB-9 corresponded to which color wire insulation, and I was unable to find that info online. SO, with a pin tester, this is what I got. I'm not sure how standardized it is, but this is a radio shack DB-9 mate to female cable. So, here's the info, please fit it in somewhere appropriate if it's relevant:

pin1 - yellow
pin2 - orange
pin3 - red
pin4 - brown
pin5 - black
pin6 - teal
pin7 - blue
pin8 - purple
pin9 - gray

67.83.198.59 01:52, 22 March 2006 (UTC)

I strongly suspect its manufacturer specific and you'd be pretty mad to assume it was true without checking. Its not like USB where a special cable type is used (in a usb cable the data pair are twisted and the power pair are somewhat larger than the data pair). Plugwash 10:39, 22 March 2006 (UTC)

I can confirm that. It doesn't match on mine (generic DB-9 male to female cut in half). 81.107.46.167 07:58, 29 July 2006 (UTC)

EIA

Can someone check to see if EIA represents: Electronic Industries Alliance or Electronic Industries Association? They are similar, and easily confounded.

It's the Electronic Industries Alliance. Confirmed by their web site, including how it is designated in their bylaws. --Brouhaha 16:40, 31 May 2006 (UTC)

What are the Associated OSI Model Layer(s)?

--143.92.1.33 10:01, 1 August 2006 (UTC)

I've never found the OSI model to be particularly useful in understanding *any* communications system. Every single paper has that "apartment block" diagram and then that's the last you hear of the OSI model. I suppose RS 232 is physical layer - you can't get further down than defining what's a 1 and what's a 0, anyway. --Wtshymanski 17:30, 1 August 2006 (UTC)

Why is there any confusion? RS-232 is clearly layer 1. It specifies electrical things like signal levels, impedences, and slew rates. It specifies physical things like connectors and pinouts. While the article says, The standard does not define bit rates for transmission, it does define transmit and receive clock signals, so one device can tell the other what bit rate it is using. This all sounds like layer 1 to me. I'll agree that not everything always fits the 7-layer model, but I don't see any such problem here. -- RoySmith (talk) 02:42, 15 June 2007 (UTC)

focus

You know, there's a terrible tendancy especially in the electrotechnology articles on Wikipedia to put *everything* we know in one article. There's no need to cover everything pertaining to the topic in *every* article. That's what LINKS are for. Taken to the ultimate, we'd only need one article titled The sum total of human knowledge to date . --03:34, 13 August 2006 (UTC)

The (or anyway a) problem with LINKS is that we can't correct them (or wiki them). For example, I find the link for "RS232 Tutorial" at http://www.camiresearch.com/Data_Com_Basics/RS232_standard.html to be a terribly misleading disservice to readers, for trying to overrule standard definitions for "Received Data" and "Transmitted Data" vs. pins 2 and 3 with statements like, "Pin 2 on the DCE device is commonly labeled 'Received Data' ..."(!) NOT! [Pin 2 may be an "input" pin (socket) to DCE, and an output pin from DTE, but I've never seen it called "Received Data" except by the very confused seeking help.] Can we remove this link for that reason? PKlammer (talk) 21:30, 22 April 2009 (UTC)

if we actually knew where theese conventions came from we could possiblly move them there, but splitting them out into a general conventions used with RS-232 article when this article is not hugely long anyway seems kinda pointless. Plugwash 12:56, 13 August 2006 (UTC)

But we've got a perfectly good article on computer serial ports which is the place to talk about stuff like bit rates - which is NOT in the standard and so doesn't belong in this article. Baud rates are already shown there. Incidentally, do the rates 115k and higher actually work, or are they some PC tweaker stunt with no application in the real world? I've never seen a store-bought RS 232 device that recommended baud rates higher than 38.4 k (which isn't strictly allowed anyway). I haven't done the maths yet but would a compliant port even have TIME to slew up to 12 volts at 400,000+ bits per second? --Wtshymanski 15:26, 13 August 2006 (UTC)

Hmm that link in your first paragraph is red and the search doesn't seem to be finding anything. Its fine by me if this article becomes about the standard alone so long as there are clear pointers to another article which actually discusses real "RS-232" ports. Plugwash 16:57, 13 August 2006 (UTC)

The link in the article's first paragraph is correct, though, and I've fixed my red link above. Serial port is the place to talk about one set of baud rates. ==Wtshymanski 23:45, 13 August 2006 (UTC).

pins and signals confusion

I find it somewhat confusing that a DCE receives data on the TxD pin and transmits on the RxD pin. My edits that attempted to clarify this were reverted. The situation is made more confusing by the fact that some vendors mislabel their DCE ports to try to make them seem more sensible. Another source of confusion is that some signals/pins have no counterpart. TxD and RxD are roughly "opposites", as are DTR and DSR. DCD has no such counterpart, yet it is used in some null-modem cables! Can someone suggest a better way to clarify these issues? -- Austin Murphy 19:30, 16 November 2006 (UTC)

I think the sentence "The signals are labeled from the standpoint of the DTE device" makes it pretty clear. I personally don't find it confusing that some DCE signals don't have DTE counterparts. Think of the DCE as a modem; these "extra" signals are things you would expect, like "incoming call" (RI) and "connection with the remote modem established" (DCD). The latter is especially important; when it isn't asserted, the DTE thinks it is communicating only with the modem (e.g., for call setup), not the DTE at the far end. Not all software cares, but a general null modem needs to assume it does and somehow assert DTE when the remote device is connected (typically by connecting DCD on one side to DTR on the other).

Maybe a diagram showing both the local and remote DTE's that want to communicate with each other along with the two DCE's they are directly connected to using the RS-232 interface would add some helpful context. (It would be useful for the Null modem article as well.) After all, although the purpose of RS-232 is only to connect a DTE to a DCE, this is only part of the overall goal to allow the DTE to communicate with another DTE. Most often today, RS-232 is used to to directly connect two devices, so this subtlety is easily overlooked. --Rick Sidwell 01:24, 9 January 2007 (UTC)

RS232 as voltage source

How much current does the standard allow you to draw from a serial port? It's not unusual to "steal" voltage from it to supply small gadget and devices. TERdON 14:55, 6 January 2007 (UTC)

• The standard says (at least in Rev. C) that the terminator shall have an impedance between 3000 and 7000 ohms, and not more than 2500 pf of capacitance. But that's while maintaining all the other requirements for communication. The standard (up to Rev. C) doesn't say anything about powering devices from the interface. So, the answer is "Try it and find out" - from bitter experience not all serial ports have enough jam to drive all so-called "port powered" devices, which is why I don't buy them any more. --Wtshymanski 03:06, 14 January 2007 (UTC)
  • Afaict no power thief can ever be compliant, simply think of the case of connecting two power thieves to each other. A power thief always requires the port it is connected to to have more drive than it can provide to the signals back. Plugwash 15:34, 16 January 2007 (UTC)

Reference wanted

The standard (at least Rev. C) says the maximum slew rate is 30 volts per microsecond. If you are using +15 and -15 volt levels, sounds like 1,000,000 bits/second could be achievable...certainly more than 20,000 bits/second. Can anyone produce a citation saying that going faster than 20,000 bits per second exceeds the standard's limit on slew rate? I'm taking this out till I see a reference. --Wtshymanski 03:06, 14 January 2007 (UTC)

PCI-E RS232 card

[2] lol —The preceding unsigned comment was added by Towel401 (talk • contribs) 00:38, 3 February 2007 (UTC).

Whats so funny about it? people often want extra serial ports and some boards don't have very many PCI slots anymore. From the photo it looks like they've just combined a PCI serial controller with a PCI express to PCI bridge (there appear to be two high pin count chips on the board one in a QFP one in a BGA). Plugwash 00:46, 12 May 2007 (UTC)

Yea i know. I bought one. and they do use the pci to pci-e bridge. Towel401 21:42, 11 November 2007 (UTC)

Relationship to USB?

I think the statement, Today, RS-232 is gradually being superseded in personal computers by USB for local communications is grossly understating the case. I would say USB has completely taken over. Can you even find an RS-232 port on a typical PC that's sold today? -- RoySmith (talk) 02:15, 15 June 2007 (UTC)

lots of the big brands have removed them as part of a legacy free push but last i checked whiteboxes were still generally shipping with one. Finding a motherboard with more than one is pretty hard though (and ports on cards don't seem to behave quite the same as the ones on the motherboard in my experiance). Plugwash 11:35, 15 June 2007 (UTC)

Would it be fair to say, "RS-232 has mostly been superseded..." ? -- RoySmith (talk) 11:57, 15 June 2007 (UTC)

"has mostly been superseded" for consumer use, yes. But RS-232 is still in very widespread use in the technical computing, industrial control, etc., markets. In home theater and professional A/V installations RS-232 ports allow the computer to control processors, receivers, sources, etc... It's still a lot easier to put a serial interface on a device than a USB port, and it's FAR easier to write code (on both ends) to talk to a serial port than to a USB device. (Just for example, there is as much code in the USB keyboard and mouse driver stack in a modern PC's BIOS as there used to be in all of DOS!) Jeh 05:55, 16 June 2007 (UTC)

Always be very careful making these kind of assertions based solely on your own personal experiences. I know several places that still use RS232 and dumb terminals extensively, and it's actually quite common in EPOS applications for instance. Yes, my most recent PC still has serial ports and yes they still get used.

For one thing I have an A1 plotter with a serial interface that would be quite pricey to replace with a more recent model. My telescope also has an RS232 interface and that is still in production. I am not going to replace a telescope simply because someone thinks it would be a good idea to remove a useful port from a PC. 79.77.32.208 (talk) 12:05, 6 May 2008 (UTC)

I will have to agree, industrial settings seem to be almost exclusively using either serial or ethernet for communications. I have worked on an Allen Bradley controller for a 500 ton compaction press that had USB but it internally converted it to RS-232. Allso, a large portion of POS terminals and conveyor barcode applications use serial. Once again, some use USB but act as a serial port on the computer so that no special drivers are required, especially for situations such as a micro controller that can't use drivers. Unfortunately, I have no sources for this information, but it might be common knowledge if you work in those respective industries (bar code, automation etc)Bill --75.180.8.80 (talk) 08:17, 29 July 2009 (UTC)

DE-9 = self indulgent

You have got a technically correct name for the connector: TIA 574. Why persist with another name: DE-9, that is not historically important, nor (outside wikipedia) in common use? Conversely, how can you have a whole article without mentioning the historically relevant term which is still in common use? 150.101.166.15 08:26, 26 June 2007 (UTC)

I'm a bit confused here. Are you sure you mean DE-9 and not DB-9 (sometimes written DB9)?

RS-232 originally was on a 25-pin connector, commercially available as the DB-25. The blind alley of the 37-pin RS-449 connector was in the same family (DB-37) as well as the simplified 15-pin ISO 2110 connector. IBM introduced an even more simplified DB-9 for PC serial connectivity.

Rather painfully, I learned there was a subtle difference between the DB-25 and the ISO 2110. The male and female pins connected just fine, but the binding screws at the horizontal ends used an English screw thread on the commercial DB-25, but metric screw threads on ISO 2110. The threads were just close enough to cross-thread if you used a screwdriver forcefully. Drilling out the screws was not one of the happier experiences of my career. Hcberkowitz 19:51, 26 June 2007 (UTC)

> I'm a bit confused here. Are you sure you mean DE-9 and not DB-9 (sometimes written DB9)?

DE-9 is as correct as DB-25 is. The D refers to the "D" overall shape of the connector shell, the second letter is the physical size, and the number is of course the number of pins. A DB-9 would be the same size as a DB-25 but with only 9 pins, and would look very sparsely populated. I know that popular usage calls it "DB-9" but that is incorrect. Incidently the popular "VGA" connector would be a DE-15 by this scheme. Re. "TIA 574", that standard refers to the signal arrangements on the pins as well as to the physical connector; "DE-9" is just the connector. Anyway, I don't know that "DE-9" is "self-indulgent" but it is correct. Jeh 05:42, 27 June 2007 (UTC)

I found this page which verifies what Jeh is saying. D-subminiature seems to cover this well, but DB connector does a poor job of it. I suspect DB connector should just be deleted and redirected to D-subminiature. In any case, this is not really germaine to RS-232. -- RoySmith (talk) 14:49, 27 June 2007 (UTC)

Certainly. But nobody has addressed the the fact that DE-9 is (1) not historically important, and (2) not (outside Wikipedia) in common use. Wouldn't it be nice if the article was helpful and realistic instead of pretentious and geeky?150.101.166.15 06:43, 13 November 2007 (UTC)

But DE-9 is a standard and technically correct name that is in common use. If it isn't in common use amongst your associates that says more about your associates than it does about the connector. D-subminiature covers this very well and like it or not the details there are correct and in common use amongst the more technically focussed sections of the community (that doesn't necessarily include coffee table computer magazines). This is not a case of Wikipedia being correct and stuff everyone else, like the policy on 'kibibytes' for instance. 79.77.32.208 (talk) 11:58, 6 May 2008 (UTC)

Personally I like the IEC prefixes. But re DE-9 vs. DB-9, the connector makers and parts wholesalers do call the small 9-pin connector a "DE-9." We should too. Jeh (talk) 07:27, 16 June 2008 (UTC)

Wikipedia is winning. Last time I searched at Farnell, I got nothing for DE9, everything for DB9. (January 2007). This time it is split 2:1 in favour of DE9. You're still pretentious, but so is most of the rest of the world. Digikey is still overwhelmingly DB9, as are our other suppliers. references: http://search.digikey.com/, http://export.farnell.com/. Am I the only one not using my own imagination as the primary reference? Thanks for the comments about my associates.150.101.166.15 (talk) 06:21, 11 August 2008 (UTC)

Since when is being correct being "pretentious"? The name "DE-9" long predates Wikipedia and even the web and IBM PCs. I found them in an Allied Electronics catalog, back when all there was was a print edition. Today it appears in the part numbers from Cinch, Cannon, etc. See for example DE-9 at Allied Electronics, DC-37 at Allied Electronics, and ITT Cannon's D-sub catalog. In the latter please see the "shell size" designation on page 225 and the part number formats through the rest of the catalog. It was only ignorant PC folks who thought that every D-shaped connector was called DB-something. So it is not "Wikipedia is winning" but rather "some are realizing their mistake and going back to the traditional names". If Wikipedia played a small part in that, great! That is part of what an encyclopedia is for. Jeh (talk) 08:13, 11 August 2008 (UTC)

I've just checked the parts in our last order. None of them came from Cannon.

7 hits for DE-9, 194 hits for DB9, 08:29, 13 August 2008 (UTC) And yes, Cannon has there own part numbers. Spectrum http://www.specemc.com/docs/emi137-140.pdf makes a D-9 part with the part number 56-10x.xxx.x.x. Shall we all change to 56-10? 150.101.166.15 (talk) 08:39, 13 August 2008 (UTC)

It's not just the part number. Cannon invented the D-sub connector and the A, B, C, D, and E shell size designations. They were the sole supplier for some time. These are therefore the traditional names. That a distributor picks their own item number scheme that does not include the shell size letter is irrelevant. Jeh (talk) 16:48, 13 August 2008 (UTC)

Right, so traditional names are good now? Listen to yourself for a minute! The point is of course that DE-9 is neither the traditional name nor the name presently in common use. It is the Cannon name. 150.101.166.15 (talk) 03:29, 15 August 2008 (UTC)

Of course it is the traditional name: it is the name and notation chosen by the inventor (and, for some time, the sole supplier) of the connector. Whether or not it is in "common use" depends entirely on which catalogs you choose to look in. This is similar to the reason Wikipedia does not settle issues by "voting". Jeh (talk) 07:48, 15 August 2008 (UTC)

the whole rest of the world is marching out of step, and I am marching in step. 150.101.166.15 (talk) 08:39, 13 August 2008 (UTC)

A cute phrase but not a factual or logical argument. btw, please do not interject in-line comments as you did, without separate signatures, as it makes it appear that the other person is arguing with themselves. Make your own paragraphs for your own comments, please. Jeh (talk) 16:48, 13 August 2008 (UTC)

"It was only ignorant PC folks" Cute phrase but not a factual or logical argument. Btw, please do not edit the discussion to detach the clearly delimited rebuttals, shown in italics and with tildes, from the points to which they refer. 150.101.166.15 (talk) 03:29, 15 August 2008 (UTC)

It absolutely is a factual argument. I had occasion to specify and buy quite a few of those connectors from various suppliers before PCs came along and I never heard the term "DB-9" until after the PC-AT started using DE-9's. All the catalogs I dealt with at that time properly referred to them as "shell size E" even if the part numbers didn't use the DE- format. Now you're trying to rewrite history as if the PC folks were right all along.

Regarding your rebuttals, the problem with your format is that responses to the rebuttals, and replies to the responses, etc. will become impossibly cluttered and will soon obscure the original text. Please note that WP:TALK no longer supports the use of interspersed replies identified by small text, or by any other means. They do support interrupting long text (usually several paragraphs) with a particular notation, but that case does not apply here. Besides all that... I am simply asking you to please comply with my wishes as to how my paragraphs are to appear; I do not want others' comments interspersed in such a manner. If you like, quote my text or parts of it and then follow the quotations with your comments. Jeh (talk) 07:48, 15 August 2008 (UTC)

This discussion would be slightly more relevant at D-subminiature but it appears to have nothing to do with this article. --Wtshymanski (talk) 04:15, 15 August 2008 (UTC)

This article does mention the DE-9 connector. We probably could add "commonly called "DB-9"" to it. Jeh (talk) 07:48, 15 August 2008 (UTC)

RS = ?

The RS page says RS stands for "Radio Standard". Mebden 10:17, 11 August 2007 (UTC)

When I worked with EIA standards groups, I remember the cover pages defining "RS" as "Recommended Standard". Howard C. Berkowitz 11:28, 11 August 2007 (UTC)

Years later...In my 1985 "Data Communications Standards Library" by Telebyte Technology are reprints of various EIA standards current at the time. RS 232 C does not say what "RS" stands for on its title page. Most of the other reprints (334,383, 422A, 423A 449) don't explain either. The second title page for RS-485 April 1983 says "EIA Recommended Standard", so at least by 1983 "RS" stood for "Recommended Standard". The "Notice" at the front of 530 says something like "This EIA Recommended Standard..." but this particular standard published in March 1987 is actually numbered "EIA-530". So, insofar as "RS" needs to stand for anything in "RS-232", it would mean "Recommended Standard" . And in the era covered by these reprints, "EIA" always stood for "Electronic Industries Association" - singular "Electronic", plural "Industries". --Wtshymanski (talk) 05:09, 7 April 2012 (UTC)

3% baud rate tolerance?

I couldn't find anything relating to the 3% tolerance of the serial communication standard neither on the related article, nor under "asynchronous serial communication".
I was wondering what the standard actually says.
Years ago, the precision of serial communication speed was as good as the crystal used (roughly 20 to 50ppm) Recently I was surprised to find out that 2 PC I tested were off by 0.187% and 0.265%, which is 1870 ppm and 2650ppm respectively and way above normal crystal tolerances.
In the old days, things that had RS-232 ports used crystals which were explicitly chosen to run at a harmonic of 19.200 kHz, allowing a simple divider chain to generate all the common bit rates (110 bps was usually a bit off). Everything from the horizontal sync to the cursor blink rate probably came off the same chain. These days, it's easier and cheaper to do it in software, especially since clock rates on CPUs change so often. You wouldn't want to redesign the bit rate divider chain hardware just because you bumped the clock from 2 GHz to 2.16 GHz. -- RoySmith (talk) 14:38, 12 September 2007 (UTC)
I would be extremely surprised if a baud rate was generating by software. Bit banging isn't something computers do nowadays, and it would make communications extremely unreliable. A 0.2-0.3% is more consistent with the use of a ceramic resonator than of a crystal.
Regardless, this was anecdotal to show that baud-rates are less precise than you would expect. The main question is: what does the standard say about baud-rate tolerance, and if it doesn't say anything, does some other standard define it?
We can easily calculate that for 10 bit transmission (1start+8bit+1stop) the sender must be within roughly 6% of the transmitter in order to actually communicate all 10 bits. So if we relate that to a nominal baud-rate, each one can be no more than 3% off in either direction.
But If we add room for transmission rise and fall time over long capacitive lines, then what is the actual "standard" tolerance one must stay within?

Timing isn't part of the RS 232 standard. Looking at data sheets for various baud-rate generators used for serial ports, depending on the source crystal frequency and the available precision of the registers, you can't always hit every baud rate exactly. Also notice wacky rates like 115,000 bits per second are an artifact of the IBM PC serial port design - the exact value of the baud rates available in the PC ties into a large amount of late 20th century electrotechnology in un-obvious ways. But there was never a design objective saying "The IBM PC serial port must do 115,000 bits per second" - it only fell out as a consequence of the 14.818+ MHZ CPU clock crystal, and the peculiarities of the INS 8250 UART chip. And of course 14.818+ MHZ was yet another legacy of the NTSC color TV standard...--Wtshymanski 17:58, 12 September 2007 (UTC)

Timing accuracy isn't a precise part, but it comes up indirectly with the pulse rise and fall time specifications. Of course, that will vary with cable length and reactance, so the best one can approximate is holding the length to the nominal 50 feet, and still doesn't address reactance. Howard C. Berkowitz 18:37, 12 September 2007 (UTC)

Rise and fall time is not pulse width, nor pulse repetition rate. "50 feet" is not in the standard but 2500 pf is. Consider the rise and fall rate limits (and voltage swing), the maximum bit rate would be far above the 20,000 bits/second in Rev. C. Does anyone have Rev. F and if so, please tell us about the differences with previous editions? --Wtshymanski 18:55, 12 September 2007 (UTC)

I don't think the standard covers issues of framing and baud rates but from a practical point of view you don't want a design inaccruacy of more than about 3% from the specified baudrate. To avoid errors from sampling during the wrong bit period and considering that each RS-233 byte is 10 bits (including start and stop bits) the absoloute maxium difference tollerable is about 10%, cut that in half to allow for the condition where both ends are wrong in opposite directions and you get 5%. Take off a bit to allow for rise/fall times ringing etc and 3% is quite a reasonable figure to work to. I don't know what things are like in the PC world but in the microcontroller world the baud rate generator is generally a counter with the ability to program the rollover value and possiblly a couple of prescaler settings. You can choose a crystal to get exactly the specified baud rate but there is no real point and it may complicate other parts of the system to do so.

A practical example:

A pic18f452 is running at 40Mhz and we are given a specified speed of 9600 baud. The datasheet tells us that the formula we need is (Fosc/Baudrate/64)-1 (40,000,000 / 2,400 /64)-1 = 64.104166667 . But we can only feed the baud rate generator with an integer so we round that value off to 64. Working the above equation backwards This results in a baudrate of 40000000/64/(64+1)=9615.38 which is an error of 0.16%, well within the tollerable 3% and very similar to the errors the original poster of this thread gave. Plugwash 22:24, 12 September 2007 (UTC)

Ooops, cut my figures above in half, I forgot that you only have to be off by half a bit position to get an error. so that means 3% is not good enough if both ends are wrong in opposite directions. Plugwash 22:47, 12 September 2007 (UTC)