Thanks for the encouragement and comments all. Last night was the first night that cloud did not get in the way.
I suspected I had slight flexure in both the optic's single mount attachment which stuffed up accurate guiding and the optical train due to gravity which distorted stars in the corners depending on the orientation of the mount. This is most probably due to the ten position filter wheel as it is a monster and asymmetric in weight distribution.
I tested this by pointing at Epsilon Hydrae which had the optic almost perfectly vertical and about to go through the meridian.
I took ten 2min green exposures. Here is the stacked result. 12MB
http://d1355990.i49.quadrahosting.co...5/flextest.jpg
Each individual exposure was as good as this stack apart from slight noise of course.
The solution will be to attach the PL16803 to two 10mm diameter rails on the dovetail with tight bushes mounted in a plate. This allows the Atlas focuser to still work but stop all flexure in the optical train.
Bert