Notes on Serial Wiring and Cables ////////////////////////////////////////////////////////////////////////////// Notes on serial communication, adapted from "Managing UUCP and Usenet, 2nd ed." by Tim O'Reilly and Grace Todino. Sebastopol: O'Reilly and Associates, 1988. On the control lines of RS-232-C serial ports, a positive voltage is considered the "on" state, and a negative voltage is considered the "off" state. This is the direct opposite of the case for the data lines. A "normal" RS-232 connection, on DB-25 connectors, showing just the TxD and RxD and signal ground, (omitting any handshaking signals that may be required): +---------------+ +---------------+ | | | | | DTE 2|>>>----------------->>>|2 DCE | | | | | | (computer) | | (modem) | | 3|<<<-----------------<<<|3 | | | | | | 7|-----------------------|7 | | | | | +---------------+ +---------------+ A "crossover" connection, where the transmit and receive lines cross, as in a "null modem": +---------------+ +---------------+ | | | | | DTE 2|>>>------\ /-------<<<|2 DTE | | | \/ | | | (1st | /\ | (2nd | | computer) 3|<<<------/ \------->>>|3 computer) | | | | | | 7|-----------------------|7 | | | | | +---------------+ +---------------+ ** If you don't know whether a device is DTE or DCE, you can always tell ** by measuring the voltage on pins 2 and 3. The transmitter should ** always have a negative voltage, even when it is idle. If pin 2 is ** negative, the device is DTE. If pin 3 is negative, the device is DCE. ////////////////////////////////////////////////////////////////////////////// Newsgroups: comp.terminals References: <3D18E386.95FF66@yahoo.com> Message-ID: <3D269BCD.2000805@valid.net> Date: Sat, 06 Jul 2002 02:27:09 -0500 From: Andy Ball Subject: Re: Null modem cable wiring This is a wiring scheme Terminal Computer that has worked well for me in the past. DB25 DB9 1 FG (see notes) A lot depends on the equipment that you are 2 TXD -------\/------- TXD 3 connecting together, and 3 RXD <------/\------> RXD 2 there is really no substitute for a decent 4 RTS -------\/------- RTS 7 manual (sadly a rare thing 5 CTS <------/\------> CTS 8 these days). 20 DTR -------\/------- DTR 4 For instance, I read the 6 DSR <--+---/\---+--> DSR 6 other day that the Wyse 55 8 DCD <--' `--> DCD 1 terminals do not implement the RTS and CTS lines 7 SG ---------------- SG 5 commonly used for hardware flow control. Use screened cable and connect the screen to pin 1 at the terminal end only (this is the 'Frame Ground' pin). At either end you may not need to connect DCD (data carrier detect). It depends on whether the terminal or the software running on the host computer demand it. The gender of the connectors usually has no bearing on their wiring. I think at one time the convention was to use female connectors on equipment and male on cables (since equipment had power applied, female connectors were less easily shorted and are probably more robust than male ones). IBM chose to use male connectors on the PC's serial ports, perhaps to differentiate them from their DB25 female parallel port). I hope the above diagram is clear enough, and that you find it useful. - Andy. ////////////////////////////////////////////////////////////////////////////// Newsgroups: comp.terminals References: <3D18E386.95FF66@yahoo.com> Message-ID: Organization: Encompasserve Date: 28 Jun 2002 14:11:14 -0600 From: Simon Clubley Subject: Re: Null modem cable wiring In article <3D18E386.95FF66@yahoo.com>, Sark writes: [Since no-one else has had a go, here are some comments that may help you.] > > I currently have a DEC VT220 as the console on my Linux server. > > I am wanting to make my own cable, so I can get rid of this adapter nightmare. > > I went looking for schematics for a null modem on the internet, and was > quickly overwhelmed. What is the best/most reliable one? > > I am looking to create a cable that goes from a DB-25 male to a DB-9 male. > > Also, what are the electrical differences between terminals with male ports > and terminals with female ports? > > (I have heard of DTE and DCE, but I don't know what they really mean) > > On a related note, I have piles of these little DB-25<->RJ-45 adapters, > in both female and male, and most are pinned differently. Can I re-pin > these/move existing pins around to use standered Cat-5 ethernet cable to > connect terminals? Will Cat-5 even work? If it works with Ethernet it > should be enough for RS-232, but I don't know for sure. What sort of > cable should I be using? > [Some of your questions imply that you may not fully understand serial connections, so I've included some basic information below. Apologies if you already know it.] DCE = Data Communication Equipment. [For example, modems] DTE = Data Terminal Equipment. [For example, terminals and (usually) hosts] When wiring DCE to DTE, you use a straight through cable. When wiring DCE to DCE, or DTE to DTE, you use a cross-over [aka a null modem] cable. A straight cable is one in which each pin is wired to the pin carrying out the same function at the other end, so Tx is wired to Tx, Rx to Rx, and so on. A cross over cable is one in which each pin which is part of a pair is wired to the other cable in the pair, so Tx is wired to Rx, RTS to CTS, and so on. DB9 pinouts: ------------ 1=DCD 2=Rx 3=Tx 4=DTR 5=Signal Ground 6=DSR 7=RTS 8=CTS 9=Ring Indicator DB25 pinouts: ------------- 1=shield ground 2=Tx 3=Rx 4=RTS 5=CTS 6=DSR 7=Signal Ground 8=DCD 20=DTR 22=RI (Ring Indicator) It makes no difference if the terminal has a female or male connector; the concepts involved are still the same. Make sure that you correctly identify pin 1 (and hence the rest of the pins) however. There is no such thing as the "best" cable, IMHO; the type of cable that you need to wire depends on the devices that you are connecting together and your flow control requirements. There are two different types of flow control: software flow control and hardware flow control. Software flow control takes two control characters, Ctrl-S and Ctrl-Q, which makes these characters unavailable to any application [see (*) below] that you are using. Ctrl-S stops output and Ctrl-Q resumes it. They can be used by both the host and the terminal. If you are wiring up a DEC terminal to a system that uses software flow control and does not need additional pins to be connected, then you only need to wire up Tx, Rx and signal ground. Hardware flow control uses the wires in the cable to signal when to stop and start transmitting. RTS and CTS are usually used for this. Depending on the device, additional pins may need to be wired. I have never wired a terminal (only printers and modems) to a system using hardware flow control, so if you need to do this, ask someone else. :-) BTW, I am not sure if older DEC terminals can do hardware flow control only when talking to a host and then use hardware flow control to pause output at user request using Hold Screen. If you look, for example, in the VT420 manual, you are told that Hold Screen is disabled when you turn off X-On/X-Off and there's no mention about if modem control means just controlling the modem signals or using RTS/CTS to do hardware flow control in the absence of X-On/X-Off. [The VT510 has support for hardware flow control, but I've never tried it.] Anyone know the answer ? I have used both solid and stranded Cat-5 cable to connect terminals to terminal servers without apparent problems. As for the pinouts to RJ45 serial equipment, you need to read the manual for the equipment, as I understand that they are not consistent but as I only have one type of RJ45 terminal server on my sites, I can't verify this. (*) I am curious to know how people use Emacs over serial terminals when using the default Emacs key bindings, as Ctrl-S is incremental search. [I have tpu-edt on within Emacs and never use incremental search, so I've never had to solve this problem.] Hope these notes help, Simon. -- Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP Microsoft: The Lada of the computing world. ////////////////////////////////////////////////////////////////////////////// Newsgroups: comp.sys.sun.admin Message-ID: References: <3922f5b.0307290625.5a015d54@posting.google.com> <3F2697AC.C0FF1D1@ntlworld.com> Organization: I have a map of the United States that's actual size Date: Tue, 29 Jul 2003 17:03:07 +0000 (UTC) From: Greg Andrews Subject: Re: Connecting 2 e450s via serial port (fairly newbie-ish question) "Dr. David Kirkby" writes: | | I built such a cable from information at | http://www.obsolyte.com/sunFAQ/serial/ | The 25-25 pin cable they suggest does indeed work. Apparently, not all | null modem cables have all the connections, so that might be your | problem. | | Note, I'm sure there is an error in the pinouts at that site for the | 9-->25 pin cable. Pins 2 and 3 should be crossed. (I'm also sure I | pointed this out to them, but it does not seem to have changed.) | No, a null-modem cable between a 9-pin and a 25-pin connector does NOT cross over pins 2 and 3. That's because IBM reversed the signals on pins 2 and 3 when they invented the 9-pin RS232 port. On the 9-pin connector, pin 2 is RxD (Receive Data), and pin 3 is TxD (Transmit Data). -Greg ////////////////////////////////////////////////////////////////////////////// Newsgroups: comp.dcom.cabling Message-ID: References: <71f7ce7.0305112045.14993165@posting.google.com> <71f7ce7.0305121702.6285a120@posting.google.com> <3EC0520E.A88DD783@hotmail.com> <71f7ce7.0305130939.5f7d07c2@posting.google.com> Organization: Shaw Residential Internet Date: Wed, 14 May 2003 17:03:39 GMT From: Robert Side Subject: Re: Designing a Built-in Computer Desk, EMI shielding questions Al Dykes wrote: > In article <71f7ce7.0305130939.5f7d07c2@posting.google.com>, > Darren Best wrote: >>w_tom wrote in message news:<3EC0520E.A88DD783@hotmail.com>... >>> All this cabling advise is good in theory. Then when the >>> numbers are applied, it becomes mostly irrelevant. Only two >>> interfaces are susceptible to noise--RS-232 and Parallel port >>> printer cable. >> >>Good to know. I have two printers, both with parallel cables. My >>Palm base is RS-232. I'll try to keep these cables clear of the >>power. >> > Has _anyone_ here ever seen any interference between 120VAC power > cable and any computer wiring ? Just one example. Tell us. > I've never seen any. A professional cabling guy wired my wife's hotel. He ran a 120VAC in parallel with 4 UTP and 1 low-speed serial line (1200 bps). The UTP worked well, the serial line would get an induced return signal that seriously messed up the the hotel bar's POS system. Wrote this one up in this newsgroup over a year ago. See subject "RS232 communication problem" from March 7, 2002. The cabling guy wouldn't believe the 120VAC would mess up the slow serial line. He finally relented and moved the serial line away from the 120VAC. All things are well now. Rob -- Rob's paradox: If you have an equal choice between two options, you will choose the wrong option 90% of the time. ////////////////////////////////////////////////////////////////////////////// Newsgroups: comp.dcom.cabling Message-ID: <3EC28810.7519E47A@hotmail.com> References: <71f7ce7.0305112045.14993165@posting.google.com> <71f7ce7.0305121702.6285a120@posting.google.com> <3EC0520E.A88DD783@hotmail.com> <71f7ce7.0305130939.5f7d07c2@posting.google.com> Organization: Posted via Supernews, http://www.supernews.com Date: Wed, 14 May 2003 14:16:48 -0400 From: w_tom Subject: Re: Designing a Built-in Computer Desk, EMI shielding questions Robert Side wrote: >> Has _anyone_ here ever seen any interference between 120VAC power >> cable and any computer wiring ? Just one example. Tell us. >> I've never seen any. > > A professional cabling guy wired my wife's hotel.... It's not about slow signaling. Its about differential signaling (telephone, Cat5 Ethernet, USB) versus common mode signaling (Centronix parallel port, RS-232). For some reason, too many cabling 'experts' don't understand this so important concept. Even Cat3 does not solve common mode noise problems. But when the cable lengths are short, then electrical noise is irrelevant. For RS-232, noise is created by ground differences between two points AND by RFI/EMI because the RS-232 wire in an antenna. As noted previously: > All this cabling advise is good in theory. Then when the numbers > are applied, it becomes mostly irrelevant." But even differential cabling has common mode voltage limits. For example, RS-485 chips typically would withstand common mode voltage differences into 30 volt range. Additional comments--James Knot has accurately noted that safety ground for two circuits should not be shared. Sharing grounds on two separate 20 15 amp circuits was not advocated. In fact the post was specifically written not to advocate safety ground sharing. However two separate 20 amp circuits can be installed that share a common safety ground. They would be wired with a common neutral and must be installed in breaker box on separate phases. This should only be done by a professional who also verifies his work with the essential multimeter. But then 40 amps to an equipment room is just not reasonable. Why is a dedicated 20 amp circuit not acceptable? That should be more than enough power. I have seen RS-232 interfered by ground noise and AC power. Its called 300 feet of RS-232 run in shielded twisted pair. However it was only for text transfer AND was in complete violation of RS-232 spec But that is totally irrelevant to this installation--unless some wires travel to distant locations outside the room. More important to this installation is heat removal AND every incoming wire to the building must connect to central earth ground either via a direct hardwire or through a 'whole house' protector. Plug-in UPS provides no effective hardware protection. For AC electric, the 'whole house' protector must be at least 1000 joules. Some breaker box protectors are undersized. ////////////////////////////////////////////////////////////////////////////// Newsgroups: comp.dcom.cabling Message-ID: <3EC39D66.7BA4819@hotmail.com> References: <71f7ce7.0305112045.14993165@posting.google.com> <2207e6fa.0305120842.34043226@posting.google.com> <71f7ce7.0305121702.6285a120@posting.google.com> <3EC0520E.A88DD783@hotmail.com> <71f7ce7.0305130939.5f7d07c2@posting.google.com> <3EC19F0A.607D9ADA@hotmail.com> <71f7ce7.0305142104.706e21dc@posting.google.com> Organization: Posted via Supernews, http://www.supernews.com Date: Thu, 15 May 2003 10:00:06 -0400 From: w_tom Subject: Re: Designing a Built-in Computer Desk, EMI shielding questions Darren Best wrote: > > You went into a lot of detail here, and I appreciate it, but I'm still > a bit confused on the differences between safety ground and earth > ground, between noise and hardware protection. It's probably a long > and involved explanation, and also probably involves learning more > about the physics of electricy than I currently know. Is there a good > reference on the web that you can link me to, so that I can absorb at > my own pace? Thanks. > > Given that I will now probably be using only a single circuit for the > whole setup, it may not be necessary to address. Still, learning is a > good thing. Concepts of earth grounding were discussed in the newsgroup "misc.rural" Storm and Lightning damage in the country 28 Jul 2002 Lightning Nightmares!! 10 Aug 2002 However the solution is connection of every incoming utility to single point earth ground before that utility wire enters the building. Most important and missing protection is the 'whole house' protector on AC electric. Power strip surge protectors don't even claim to protect from surges that typically damage electronics--and cost 10 to 50 times more per protected appliance. Safety ground is for tripping the circuit breaker so that humans are not harmed. However it is also used to put all equipment at a common voltage to reduce noise. The term most often used to discuss this noise problem is 'ground loops'. I know of no simple and good explanation of ground loops and single point ground for noise reduction. In simple terms -- wire is not a conductor. Wire has a voltage difference (is an electronic component). If two peripherals are connected at different ground voltages, then noise exists. Therefore all components should be grounded (and we do this with the safety ground) to the same point so that they all connect to the same voltage point. Cited previously was the HP Laserjet II on a data cable switch box. Fundamental to understanding what happened: Laserjet and computer were not connected to same, single point, power connection. Therefore both had a zero volt reference that was different from the other. In other words, a voltage difference existed between both zero volt reference points, causing hardware damage when data cable was switched. Computers are so robust that this is only secondary consideration. The concepts of single point grounding only make sure and "again" that no strange problems or hums occur when you are done. Like all that worry about bundling power and data cables together--it's rarely is a problem. With a single 20 amp circuit, you should have single point grounding automatically. w_tom //////////////////////////////////////////////////////////////////////////////