Some effects, such as dynamic effects (compressors, limiters, noise gates) are normally used in series with the signal (i.e., they interrupt the signal path). These are normally used on an insert in the individual channel (there are exceptions to this, such as buss compression). If you use these effects in parallel, they do not have the desired effect. Most of these effects were originally devised as what I think of as "housekeeping" effects; i.e., they were used to make the music conform to the limitations of the recording or broadcast mediums of the past. (compression was too keep the radio signals from jumping beyond their frequency spectrum, and to keep the needle from jumping the groove in record players, for instance.) Most of those limitations are gone (other than radio frequency limitations - which is why radio always sounds so compressed), but we still use the effects in the same way.
Other effects (mostly time domain effects, such as delay, reverb, chorus, harmonizers - yes, that is a time based effect) are mostly used in a parallel configuration, because we want to mix the original signal with the effected signal. (It sounds pretty weird if you can only hear the delay, and not the original guitar, obviously.) For these effects, the most common way to use them is through a Aux buss or aux send (same thing, different name). These can either be returned through an Aux return, or more commonly (if you have one to spare) through a channel strip (which gives you more control). These effects tend were from the beginning more ornamental in nature. With many effects - particularly digital multi-effects - it is possible to mix the dry and wet signals in the return signal, but it is always going to give you better sonic integrity to keep the original dry signal from the console, and set your effect to 100% effected (or "wet"). Again, with many of the original versions of these effects, this was the only option. The original delay unit was just a tape machine (it utilized the distance between the record head and the playback head to create the delay, and you adjusted the time by adjusting the speed of the tape machine). The only signal that comes out is what comes through the playback head. On a real live plate or spring reverb, there is a driver on one side of the unit, and a pickup on the other. (A real plate is just that - a large steel plate with speaker driver in the middle, and one or more piezo pickups placed around the plate to pickup up the new signal.) The output is has not dry component. Choruses and flangers are just variations on the tape delay concept (you push on the "flange" of the tape to speed and slow it down, thus changing the pitch of the delayed signal - the only difference between the two is the rate and depth of the variations, and the speed of the delays, but they are basically the same effect). With these older versions of the effect, you have no choice but to use the console to mix the dry with the wet. The old reason for using them like this are gone (well, reduced - it will still sound better to keep the dry signal in the console), but the habit of using them this way remain. Also, using an Aux Send means you can use the same effect box for multiple instruments. I, for instance, much prefer to use the same reverb for everything as it acts to unify the mix (to my ears).
A fun little application of that last: back in my days of working in a real studio, we had a really fantastic nice plate reverb. As in, a real live, honest to goodness, heavy as shit steel plate. It sounded fantastic. I used it more than anything else for reverb (it was a touch noisier than the Lexicon, but it sounded so much better). But I didn't want everything going through it to sit at the same depth of field - so I would set up a couple of delay lines (sometimes digital, sometimes tape machines) which would be sent back to the console, though not put into the two mix, and then would be sent out to the reverb through the AUX send. If you look at the more in depth settings for your digital reverb, there will be one called "pre-delay." This is exactly what pre-delay does - though obviously to everything coming in - and this is how it was done before digital effects. Even with modern digital effects, you can use the same idea. Say you only have one really great reverb plug-in, but you want to give each instrument it's own ambiance - give each instrument it's own delay line before going into the reverb. You can also EQ sends drastically differently from the dry signal. For instance, reverb tends to get bottom heavy if you use too much, which makes your mix sound muddy. So, put an EQ before the send, and roll off some (or, even, a LOT) of the low end, and you can add huge amounts of reverb, and you can get a really great shimmery ambiance that you could never do otherwise. (You may or may not like these sounds, but they are good things to experiment with).
Just because an effect is usually used in a particular way does not mean you need to, by the way. It is quite common to send a sub-mix of the drums through a compressor separate from the individual instruments, and use both in the mix. Using two different compressions on a vocal in parallel can be useful at times. Play with things. It's actually pretty hard to damage stuff, so why not try something weird?
Wow, I sure do like the look of my own typing, don't I?
Light
"Cowards can never be moral."
M.K. Gandhi
