I have been doing some work building a prototype digital underwater line compass for cave survey. A note is on:
http://www.cave-exploration.com/papers/Compass.pdf
The project still has some way to go, so the note describes what has been done so far and what I think needs to be done. The focus to date has been on ergonomics, because the number of azimuth "blunders" which appear in surveys suggests that this is the limiting factor at present. However the testing shows an accuracy of 1.4 degrees (RMS) or 1.25 degrees (absolute mean error), and furthermore the shape of the error curve and the high repeatability suggests that it would be little trouble to improve the accuracy by some fairly simple changes to the calibration routines.
I imagine that I am not the only person working on this sort of thing, and perhaps others will find the note useful. (I have also posted this on the Cave Diving Group BBS).
Regards,
Simon Richards
http://www.cave-exploration.com