Hi Alistair, the secondary must be correctly positioned under the focuser yes, but centering within the OTA is less important. As long as you still have a clear field of view across the entire chip.
For an astrograph you'd also preferably have a secondary large enough to 100% illuminate the entire FOV.
I like the star testing method as it is the only one to directly measure star shape/sharpness which is all we really care about. This also means I don't have any tools for collimation as one doesn't need any with this method.
Thierry Legault has an excellent tutorial here:
http://legault.perso.sfr.fr/collim.html
Another one is here:
http://popastro.com/documents/PA_jan...tartesting.pdf