Ok, Ray had mentioned at some point to have points much closer together than my previous model.
Made a model tonight concentrating either side of the Dec of the target with the points are very close together. 1 degree for Dec - 5 degrees in Ra.
The mode scatter graph also looks good.
I tried all combinations like model correction on - Dec arc and refraction corrections on and off. Also tried PEM on vs off.
Target still moves in 1 direction at a slower pace. about 45 degrees diagonal. Eccentricity is about 0.5 to 0.6 for 300s subs and 0.92" image scale.