With XLR connectors (whether three-pins, or more), it's standard that pin 1 is the ground. And it makes contact first, because the socket is designed that way, to reduce noise problems during plugging and unplugging. Usually pin 2 is the in-phase signal line, and pin 3 being the opposite phase signal. While there have been manufacturers that reversed that, they're quite uncommon, these days. And that doesn't matter when making up standard XLR to XLR cables, as you normally wire pin 1 to pin 1, pin 2 to pin 2, and pin 3 to pin 3. The wire doesn't care what signal passes through it. Where it does matter is when making adaptors between different types of connectors, or the internal wiring of equipment.
There's an old memory trick that I've always remembered from the first moment I was taught it: “X, L, R. 1, 2, 3. earth, line, return.” And while those letters don't actually mean that (they're a product code for the design of the connector), and “return” is a name you'd find in a current loop rather than a balanced line, they fit neatly into the mnemonic. If I were teaching someone that for the first time, I would have said “earth, line, reverse” (reverse meaning the reversed phase signal). There are other mnemonics for it, as well, but you only need to remember one, and I prefer one that directly uses words that relate to the signals in question. Though I can't think of an ordinary word for shield that starts with an X.
I've yet to find an XLR connector that doesn't label the pins, but sometimes you have to solder these up under conditions where it's hard to read them. You should be familiar with how pins one and two are next to each other along the flattened part of the insulator, in the middle of the plug, and pin three is by itself at the bottom of the curved part, but trying to work out which pin is pin one or two can be awkward when you can't read the labels. The locating notch in the shell is near the in-phase signal line. If you want a memory trick to remember that, you can call it the “signal indicator,” or remember it's “next to two” (playing on the words “to” and “two”).
As to what colour wires to use where, that's up to you, and doesn't really matter, so long as you wire the connectors properly. And most diligent people will test cables with a meter, rather than presume that coloured wiring has been used in the manner they expect it to, but it's common practice to do things in a consistent manner. If you consider the in-phase signal as positive, and the out-of-phase signal as negative (the inverse of the positive signal), then you'd pick wiring colours the same as if you were wiring for DC. For cables with red and black wiring, you'd use red for positive, black for negative. For red and white wiring, you'd use red for positive, white for negative. For white and black wiring, you'd use white for positive, black for negative. You should be able to see a pattern forming, above.
Some alternative names for the signal lines are:
Though “hot” and “cold” are commonly used, they could hardly be a more incorrect use of language. They have nothing to do with temperature, and there's nothing in balanced audio that says that one signal should be hotter (higher voltage) than the other. It's a misnomer that should be abandoned.
Positive and negative make sense, if you're talking about the phasing of that pin, or the polarity of the voltage to be found for an in-phase signal. But don't think of them as steady-state, or DC voltages. Balanced audio is alternating current, the signal cycles between positive and negative, positive or negative will only refer to the start of the cycle.
Normal phasing is thus: A positive pressure onto a microphone (vibrations pressing against it) will produce a positive going voltage, and a positive going voltage onto a speaker will produce a positive pressure (i.e. the speaker cone will move out towards you). The in-phase pin (2) carries this signal, the reverse-phase pin (3) will carry an inverted signal. Swapping the wires around will invert the response.
Phasing is important for combining multiple signals. For a single signal, it's probably going to be unimportant if the phase is correct, or reversed, as few things could tell the difference. But when combining signals (e.g. several microphones spread throughout a group of musicians), one signal reversed in comparison to another, causes signal subtraction, rather than addition. For some signals, that'll cause some annoying audio errors. Hence why it's important to wire leads up correctly (as detailed, above). If you need to deliberately reverse the phase of a signal, to compensate for a problem (such as microphone placement, or a signal fault somewhere else), that should be done with special leads, or using phase switches on your equipment. If you make a reversing lead, be sure to label it.
The age old question of whether to connect the connector shell to the shield wire can never be answered in a way that works for all occasions.
In the equipment, the shell should connect to the chassis. Which may also connect to signal ground, though perhaps through a resistor and capacitor. The shield wire should not, normally, be a part of the signal path, and whether the equipment connects the chassis to signal ground is another issue (equipment design, not cabling).
If you're connecting one patch lead between two pieces of equipment, then the shell doesn't need connecting to the shield, as the equipment should ground the shell with its chassis. This includes every microphone that I've ever seen, the connector shell electrically connects to the microphone body; and internally, the shield wire will go to the body of the microphone, ensuring the shell is grounded. But inside other equipment (than a microphone), connecting the shell to the signal ground may be subverting the grounding of that equipment, this can introduce noise. So I usually leave dedicated patch lead shells unwired.
If you're chaining leads together, then it's usually beneficial that shells in the middle of the cabling are connected to the shield. If they're not, then hum can get introduced into the wiring if they're close to an interference source, or if they touch something. On the other hand, if they're grounded, then hum usually doesn't get in, but you can get an earth loop if the connectors touch something else that's electrically conductive. The ideal would probably be an internally conductive connecctor, with the inside grounded, and the outside insulated. I've never seen a connector like that, but you could wrap tape around the outside. In practice, I find it's not much of a problem when connectors touch together, but I frequently find that hum gets into ungrounded connectors. So I ground the shell of female microphone leads, or any other long cables that are used for connecting to random equipment, and tape the cables to the floor near the join.
Since I usually find it beneficial to have the connectors grounded, I ground the female shells of most of my leads, unless I know that a particular lead is best not grounded (usually dedicated patch leads). And since I, now, have two outwardly identical cables that are wired differently, I label the connectors “isolated” or “grounded.”
There's only a need to ground one end, since joined connectors will, also, join the shells. So there's no need to do both connectors. I ground the female connectors, I have less problems that way.
One male end will connect to an input, and be grounded there. Connecting that chassis ground to the shield wire, as well, might mess up input stage grounding carefully designed to minimise noise, so I leave the male shells unwired. At the join between cables, the next male will be grounded by the female connector that it's plugged into.
One female end may be disconnected, laying on the floor next to a power transformer, metres away from the mixer, get picked up and handled, so it's good to have it grounded to minimise noise getting in. When another lead is connected to it, that other lead's shell will get grounded. When finally connected to a microphone, the shell and shield will be directly connected in the microphone, just a few centimetres away, so it's highly unlikely that you'll cause noise problems with the same wiring joined in the connector. Or, when connected to a powered instrument, like a keyboard, that instrument is probably ungrounded (most things are, these days), so it's often beneficial to connect the chassis to the shield grounding. But if connecting to grounded equipment (either grounded via its power cord, or because it's connected to other equipment, as well), you may have a noise problem if you connect shield to chassis, so having some short patch leads with isolated shells to connect, in between, may be necessary.
Since I usually want grounded connectors, and occasionally need to isolate a shell, I find it best to ground all my female connectors so that they “just work” when simply joined together. And I have a few “problem solving” patch leads to connect to the very end of cables, to deal with the unusual cases. I regard these problem solving leads as being similar to having phase reversal adaptors, and transformer boxes.
A fairly new trend is the use of plastic XLR connectors on equipment. The connector is electrically isolated from the chassis, and doesn't provide the usual grounding path. The chassis connector might not make any connection to the shell of the line connector plugged into it. So you're left with an isolated, ungrounded, metal shell around the wiring at the equipment. This may not be a problem, unless you're in an electrically hostile environment (but then you'll probably have noise problems, anyway). And you're probably not going to touch the connectors, except while repatching (where you might expect noises, anyway), unlike connectors in the middle of a cable, which may get dragged onto something, while they're being used. But it does the raise the issue that some people may want to ground the male end, in the cable.
All I could advise would be to test your equipment, to see what you might need to do. We have a Soundcraft mixer with plastic Neutrik connectors, and I'm yet to decide what to do about it. I dislike plastic XLRs, as they introduce a potential electrical noise issue I may have do deal with (the lack of shielding), and they're not very robust.
Sometimes you'll have to break away from using standard leads to resolve a problem such as earth loops, or to improve poor signal-to-noise issues when multiple shielded lines connect between multi-track equipment. You may need to lift the shield away from pin 1 on one side, or the other, on some of the cables. e.g. A recorder with two or more signal lines connecting to a mixing desk might have the first cable wired in the ordinary way, and all the other leads between them have the shield lifted off at one end. Which end, would be whatever produces the best results. Though you'd, normally, make that change at the same end (i.e. at all the female connectors, or all the male connectors). As always, special leads should be labelled.
The most common XLR adaptor lead you're likely to need, would be to go from an XLR to a ¼ inch phone plug. Normally, the tip will be the in-phase signal, and the ring will be the reverse-phase signal, with the sleeve being the shield. You won't get any option about the grounding of the shell, it's a mechanical extension of the sleeve.
This is really a question about audio signals, rather than XLR connectors, so see the balanced audio page for the whys and wherefores.
There are, at least, five common sources of unwanted electrical noise, and you may be subjected to more than one of them at a time, but a good balanced line should reject such noise sources quite effectively.
I've listed mains causes twice, based on different sounds and causes. For instance, you're very likely to hear a humming noise if you get a dynamic microphone near to a power transformer, or even when its cable passes close to mains power. And you're very likely to hear a hum if you get an earth loop. A buzz is the noise that you commonly hear getting into audio lines from lighting dimmers and fluorescent lighting. Radio interference is noise from radio frequencies (anything that's above ultrasonic) getting into the audio—it could be that you actually hear radio or television signals on the audio, or just annoying interference. And digital and motor noises can be buzzes, ticks, or whirrs generated by some of the equipment itself, such as computerised digital circuitry, or electrical hash from motors in the equipment.
For a number of those problems, simply repositioning equipment and/or cables, may be enough to reduce them, or even eliminate them. Using properly balanced signals and cabling, and taking care of how you power equipment (everything from the one source) helps minimise the effect of ground loops. And properly balanced signals and cabling, using best grounding and shielding practices, is your best defence against all sources of outside interference, and against internal equipment noises getting into the signals between equipment.
Using good quality cable, that's designed for balanced signals, is essential for dealing with hostile environments. Ideally the signal lines should be a pair of wires twisted around each other, or two pairs twisted around each other for star-quad cabling, with a shield that goes around the outside of them. Cable that has untwisted pairs will be poorer quality cable (e.g. as commonly used in round-shaped shielded headphone cable). And trying to use two individually shielded wires (e.g. figure-eight stereo cable) as a balanced pair, doesn't work well, either. While any two wires can connect the signals from one balanced signal source to a balanced input, it upsets the balance, and hampers the ability of a balanced signal system to reject noise. A balanced signal line is not merely two wires carrying a signal, it's two signal wires with the same electrical characteristics as each other, and with consistent characteristics between the two balanced lines and the shielding around them. The whole signal path needs to be balanced—the output stage, the cabling, and the input stage.