So, with a little guidance from one of my luthiers books, and revising some of my first year university physics, I'm going to try to explain some stuff to the best of my understanding.
How does an acoustic guitar work? You pluck the strings, sound is transferred to the soundboard and you hear the note. Yeah okay fine, but we'll need a bit more detail if we are to work out what is happening at our saddles.
Okay, in this simplified picture you can see the string (under tension) goes across the saddle and down to the (or through the) bridge, with the bridge glued to the soundboard. For simplicity I'll say the soundboard has rigid/fixed nodes on either end, even though in real life it is shaped in a figure-8 and there are non-rigid dynamics occuring at the soundboard to side interface...okay, that's too much detail.
When you pluck a string, what happens? It vibrates, creating a standing wave between the point it touches the saddle, and the nut or at a fret if you have fretted a note. The fundamental frequency is of a wavelength the full distance, but there are also many harmonics added by many other things resulting in a complex timbre of the sound...but again too much detail.
Looking at a plucked string, you can see it vibrate, maybe up/down, maybe left/right, maybe in circles, if depends how you plucked it. There's the clue - the _motion_ of the vibration isn't what is directly causing the sound, it is *pressure* (tension) that is transferred to the soundboard via the saddle and bridge. Tension is of course directly related to the motion of the string, at the peak offset of the string vibration it is at maximum tension, passing through the lowest tension at zero offset, heading for the opposite peak. Think about it, if you pull a string away from it's resting position, you create more tension, it is pulling harder against the saddle (and the nut, but the nut is a lot more fixed and solid than the bridge).
Ah, so the tension goes past the saddle, and pulls the soundboard up and down then? No; no it doesn't. That would be like trying to pick youself up by your shoe laces.
The saddle is fixed in place hard up against the bridge, so all that can happen is the bridge/saddle _twists_ the soundboard. That's right, _twists_. The bridge/saddle ever-so-slightly pulls forwards with the tension of the string; and because it is connected to the soundboard it pulls that with it, creating a "bulge" behind the bridge, and a "dip" in front of the bridge.
The soundboard thus becomes an air pump - as the back bulges up it sucks some air in through the sound hole, and as it goes back to the resting position the air is expelled.
I hope I have explained that well enough that you understand, because now I can finally start talking about saddles.
The saddle sets the amount of torque that is transferred to the soundboard. If you have a really tall saddle, it is "easier" for the string tension to twist the soundboard, giving a bigger acoustic sound...if the saddle is lower then less torque is transferred, and you get less acoustic sound. (Let's totally ignore action, it isn't a function of creating sound in this context.) The soundboard can saturate if the saddle is too high; if you start putting a massive load on it then you won't get a linear tension response in the soundboard and the sound will start distorting. I'm back to too much detail.
As long as the saddle is firmly held in position, isn't made out of a material that absorbs vibration (for example a saddle made out of rubber) then the pressure is transferred reasonably efficiently and you have a faithful reproduction of the string's vibration to the soundboard. Different materials will add different nauces to the timbre of the sound, for instance a highly efficient material would be a steel saddle, but we aren't after an exceptionally "true" transfer, what sounds best is a compromise, a material that transfers torque but also absorbs a bit, giving a sound that we like. Again I'm heading out into details land, and my main point in continuously doing so and then stopping myself is that the creation of acoustic sound from a guitar is _incredibly_ complex, and all I really want to touch on is the absolute fundamental principle, firmly remembering there is more involved than just the basics.
Now, lets put an Under Saddle Transducer (UST) underneath the saddle. You're going to have to think about this too. The pressure being transmitted by the string goes over the saddle and then some heads down and back at an angle, so we get a ratio of the force is directed straight down through the saddle. Note that this is balanced by the force pushing up from where the string attaches to the bridge - can't pull yourself up by your shoelaces, remember? But there is still a squeeze going on between the saddle and the bridge.
Therein lies the beauty - even though the soundboard is twisting to get your acoustic sound and it has nothing to do with the saddle downwards pressure - there exists this place between the saddle and bridge where an exact "sample" of the string tension is available. By placing an electronic gizmo (UST) in there which turns pressure into a voltage, we have a pickup!
Now I said the join between saddle and the bridge has nothing to do with the acoustic sound - if the saddle was a bit loose, or the saddle slot a bit rough on the bottom it's going to make no difference to the acoustic sound (so long as the saddle isn't absorbing pressure, for example, minutely bouncing around). This however makes a lot of difference to trying to catch that sample pressure. When an acoustic guitar is built, unless it is going to have a UST installed, they probably aren't going to care about the saddle slot quite as much.
I think you are with me. Once I had fitted my pickup, I was getting poor response, and it came down to the saddle. Here are some symptoms, and some resolutions:
* The saddle was a bit loose in the slot, so once it was under tension, it lent forward a little bit. Doesn't affect acoustic sound, but suddenly all the downwards pressure was directed on the front edge of the saddle, and the UST wasn't getting the full pressure sample. This is fixed by buying a blank over-size width saddle, and very carefully on a piece of glass or polished marble/granite surface with sandpaper, "thinning" down the saddle to be the exact same width as the slot. Too loose and it will pull forward, too tight and pressure is lost into the walls of the slot as it is "jammed".
* Having a nice tight saddle will then next show you the saddle slot is not perfect. My slot in fact flares out a bit at the top - not much I can do about that other than having a new wider slot routed in. But all the laquer in the slot, from when the guitar was laquered, also caused it to be imperfect. Armed with a trimmer blade, I carefully scraped off laquer on the floor and the sides of the slot to ensure they were as flat as I could get them.
* To fight a non-flat slot from two fronts, I bevelled the bottom edge of the saddle, just a tiny amount. If the bottom of the slot still had some lacquer or generally non-90 degree bottom leading and trailing sides of the slot, that is, "radiused" sides, even only a tiny, tiny bit - and even if you don't see it - the saddle would be jamming in that last minute section, losing pressure. If there was going to be any non-flatness in the slot, that's where it would be, so bevel the bottom of your saddle.
* Even still, I went one further step and put a spacer in the bottom of the slot, underneath the UST. The spacer I made was two pieces of normal paper glued together cut to the same size as the bottom of the slot, acting as a kind of gasket, balancing out any imperfections. Although this gasket arrangement will absorb some of the pressure; the pressure gained by it being there is far more than that lost. Give this a try and see how you go - if the slot is good enough, you won't need to.
* Balancing. As in, how much pressure is exerted by each different string - if you play all of the strings with the same touch, are any louder than each other? Typically on a nylon string, the G string might be quite a lot louder than the rest...or maybe the D string. Combat that by taking some meat off the bottom of the saddle directly underneath the string - basically you are removing some of the ability for the pressure to be transferred from the string to the UST.
If the higher strings are too loud compared to the bass strings, then file your saddle to have a top profile where the bass strings are higher. This will result in a greater break angle from the string to the bridge tie off, which means more downwards pressure on the saddle.
Alternatively, if your saddle is already crazy high on the bass then add a sliver of paper on _half_ of the bottom of slot, which will take away a little bit of pressure from the trebles/direct it to the bass instead.
There is so much to talk about with saddles, bridges and the acoustic guitar's general ability to make a noise, and so much I don't know. I've gone on here for long enough, I'm sure to go on about it again in the future!
Oh, and as a final note - no UST is good enough by itself to get exceptional sound. You must always mix it with an AST/SBT or built in condensor mic - the UST is too focussed on the fundamental frequencies, an AST/SBT/mic picks up all that other good sounds the guitar is making :)
JAW