When I first collimated the gs12 I found the collimation varied significantly depending on the angle of the scope-horizontal,45degs,vertical.
A quick inspection revealed the mirror floating around on the 3 springs supplied on the adjusting screws. I collimate at 45degs for convenience but moving the scope to vertical then horizontal for a final check. When positioned at vertical the springs supplied compress too easily under the full weight of the mirror upsetting the collimation .
An interim fix was to turn all screws in so the springs were 'nearly' fully compressed and collimating by turning out only the 2 screws required. This gave better support and reduced the variation in collimation at different angles.
A quick pm to another member confirmed the problem and agreed the springs are too soft.
A quick run around this morning and 3 new springs,slightly longer with a thicker diameter wire were found and fitted. I was only able to find coated steel springs,but will chase up some s/steel springs later.
The mirror now stays put and the collimation doesn't change at all.
I know collimation is hard enough to come to grips with at first, let alone having a mirror floating around.
Hope this post helps to remove some of the frustrations with collimation on these scopes.
Slice