The instructions seem pretty clear to me "as close as possible." Probably only Dynon can say how it works, but from aligning them in a couple of configurations I have no sense that you just need to get close and then it snaps in. I have tried section lines and compass roses and sightings with compasses. The more precise methods seemed to give the better results.
I have very good results by using the GPS method I mentioned. Previously, using less precise methods I didn't have as good results.
"As close as possible."
Let my try and give a general answer that applies (as far as I know) to about all magneto-restitive compass solutions.
There are two major kinds of calibrating a compass and these can be used combined as well (which is what we are doing).
a) Swing the compass through 360 degrees in a special measurement mode. This simple takes measurements of the X and Y axis sensors as it goes and takes note of the highest readings for each sign. Once done it uses this to "shift" the readings so they would be the same for both signes. You can picture the effect of the readings resulting in an oval or otherwise distorted circle (due to effects of both sensor inaccuracies and external effects). External magnetic "corruption" results in concentration and thus strengthening of the field the sensor sees.
The overall result of this calibration is to center the distorted circle which improves matters. However, while early on this is what we did (as this is what the manufacturers of the chips suggest (Honeywell)) we found that this was idealistic and not very useful in a real aircraft where you could have lots of different effects on the magnetic field all at the same time. You might find that you could fix an error of say 10 degrees to become 5 degrees, sometimes making other headings that where not so bad, worse at the same time.
b) The other method we then tried was to align the compass to a known magnetic heading and simply tell it what the heading is. Once the compass knows its error (for that heading) it can simply subtract the error. Do this for a number of known headings (N,E,S and W are quite sufficient) and then let the compass interpolate the four known heading errors over the full circle.
MUCH better.
So, to answer the question - the compass does not know where north (or any other heading) is so you must point it there as accurately as you can and then tell it. A cheap hiking compass (preferably with lubber line) is a good reference. GPS can be as well provided it can work out magnetic heading (for this it needs to know the exact magnetic variation at your location and this changes over time. Some GPS receivers (like ours) can do this by actually containing a model of the Earth magnetic field in detail and know it changes over time for each location).
One golden rule: Ensure that you have choosen the best possible location in your aircraft for the compass before you start calibrating. Try and find a location that shows only a small error (if you can get to less than 10 or even 5 degrees that would be great). If the compass is too close to a disturbing piece of metal (magnetised or not) it may concentrate the field too much leading to saturating the sensors at the affected headings (then no amount of calibration can do anything). The idea is to reduce the need for calibration to an absolute minimum. This automatically results in the best possible performance after calibration.
Lastly, just as a side note: If your aircraft is parked in the same location, in the same orientation all the time, some of its hard-iron metals will slowly be magnetised by the Earths magnetic field and this can result in your compass starting to show an error. You may need to recalibrate once in a while (or park your toy in different orientations after each flight if possible).
Hope this makes it a bit clearer.
Rainier
CEO MGL AVionics