Whoops - it goes to show I shouldn't work on flight planning problems and drink beer at the same time!
Sorry, I stuffed up in my method above by not being consistent with which way I applied the wind. On reflection, it should have gone like this:
1. Get boat's course and speed. From the centre grommet on the wind face, plot out in that direction the appropriate number of knots. (eg for boat going 270 at 20 kt, put 270 at the top and plot 20 kts straight up on the TAS scale. Make a mark.
2. Get the wind direction and speed. Make a mental line from the centre grommet in that direction (ie TOWARDS where the wind's coming from), and then parallel that from the first mark a distance equating to however many knots of wind there are. Make a second mark where you end up. (eg for the wind 180 at 30kt, go from the first mark 30 kts worth of distance (estimating from the TAS scale) in the same direction as 180 is from the centre point.)
3. Rotate the outer ring until the second mark is at the top. You can now read off the wind.
What I've done there is just add vectors, really. We're looking from the boat's frame of reference, and there's one relative wind velocity caused by the boat's speed, to which we must add the actual wind. If you 'step' into wind both times, you'll end up with the right answer. For the example I used above, it should have been about 220 at 37 kt.
Just working through another example - boat 330 at 25 kt, wind 210/15, I get the resultant 285/22.
Takes about 10 seconds.