I was being stupid when I said it should glow. Something more like one of those simple rubberband force meters used in physics classes would work better:
http://www.practicalphysics.org/imageLibrary/jpeg200/317.jpg Except much smaller, and hooked up to some pivots so one end can point in a full hemisphere (you'd only need a full sphere if you plan to go above the sun, but at that point, the pointing device might not even point correctly anymore).
EDIT: Yeah, it would need two leveling tubes. I forgot that it could be tilted in the other direction. Only if you used tubes though. You could make a slightly curved "button" shape, with a circle or spot in the center. Then the bubble would be in the circle when level, and move any other time.
Okay, now let's see if I can explain the whole mechanism.
Assuming there's a horizontal magnetic field, you have a magnetic compass. That doesn't need explanation. It's needle is the M-needle. You have another needle that points to the tip of the Azure Sun. The C-needle. It's attracted to the crystal strongly enough that you could fashion a device to measure how hard it's attracted (force would weaken with distance). That would allow you to calculate the distance to the tip, which is your hypotenuse. The user wouldn't need to calculate it, only the crafter, who would add notches saying, "50 km", "60 km", etc. There's also a leveling device to make sure the angles are correct.
To determine your angle around the axis of the stalactite, you'd compare the angle of the M-needle and the C-needle. One end of the C-needle will be low (unless you're level with the crystal), so it would point to an angle. Various means could be used to make reading it easier, even having a second C-needle that doesn't point up at all, only towards the axis, or just having an adjustable dial that you'd point toward the crystal via sight, similar to the mechanism on our compasses.
To determine the angle to the crystal, you check the angle of the C-needle. Simple enough.
To determine your hypotenuse, you use the magnitude element of the C-needle, which would be pre-marked like a ruler or scale.
Then you get to do the math, which wouldn't be near as easy as a simple estimation based on eyeballing the sun, knowing the approximate elevation, and guessing your radial position based on walls, ceiling, or opposite side. For this device, you'd definitely need to be one of the educated. Though for just a coordinate system, if you based it off the tip of the crystal, you wouldn't need any calculations (other than subtraction between M-needle and C-needle) to get your actual coordinates, just to translate them into distances. But maybe if you grew up using this system, it wouldn't be so bad to estimate. We're used to cartesian coordinates because that's what we've always used. And basic conversion factors for general intervals might be commonly known (kinda like how some people can actually use English units. I just use feet, miles, and pounds. Anything else is metric). Especially within a certain region. So people on the first level might know approximately how far a degree is at the edge, and at every 1/10 of the way back to the wall or so. Then just estimate within them. Again, that wouldn't yield precise measurements or anything. Just trying to emphasize that it could be possible for the laykran to have a basic idea of distances expressed using a polar coordinate system, especially in his region.
But that's just a "simple" version of the compass, using minimal magic and gadgetry. If I sat down and thought about it, I could probably come up with a machine that would do most of the work automatically with minimal magic (mainly to counter friction and increase attractive forces). I feel safe in saying a needle can be fashioned that would point to the crystal, but otherwise, you could get away with "pointers" and various methods of alignment (it's a light source, so it wouldn't be too hard with the right equipment). In that case, distance would be the trickier part. Depending on the precision of your parts and alignment, maybe you could set up two separate "pointing" devices and triangulate it.
Also, I haven't worked out any of the actual numbers to see how much precision we'd actually need for these calculations. It may be that you'd need a huge compass to get a close enough reading. I'd have to dig up the dimensions of the cavern and crunch a couple numbers, but I'm too busy right now. The trickiest part, again, would be the distance mechanism. You'd need enough attraction to see a difference, but not so much that it get's yanked out of your hand, and also not so much that traveling a small distance uses up all the "travel" in it. Maybe it would be better to just add a separate magic-field-magnitude sensor that works via other means than attraction. I think that's why I originally said glowing, but that isn't precise enough.
Even if it turned out a pocket device wouldn't work, the more eccentric people might still have built gigantic devices in the name of science. Or even a Stonehenge type deal by the less modern types (everyone knows Stonehenge was originally built as a gigantic polar-coordinate uber-compass by people suffering from PlaneShift withdrawal during a server outage)
Maybe if I get bored on my coming 4-day weekend I'll do all that, and maybe even dig out Blender or a cad program to see if I could model the device. That could take a while though, but it would be pretty fun, and worthy of a student in my engineering heavy college. Actually, creating blueprints for an ubercompass that would only work in a video game world miles underground inside a giant stalactite probably wouldn't even be considered strange here. Maybe a little obsessive, and I guess the fact that the video game isn't Halo or WOW might be considered strange, but not by much.