PHD Guiding accuracy

[blinky]

I have been using PHD with EQMod for a while and am pretty sure the guiding is normally OK. Last night when imaging the comet though, after a 6 min exposure there was a little star trailing evident if I viewed the image at full scale. I enabled the graph output on PHD and I could see that it appeared to be overcompensating i.e. if it was a negative correction it would go below the centre line then the positive correction would go above the centre line. Anyway I changed the aggresivness and it seemed to settle right down, the meased value was about 0.4 but I still had the trailing.

Anybody any ideas on what else I could do? I will post an image this afternoon to show the effect but I am a bit stumped as to what to try next!