cronbachs alpha using repeated administrations on same subjects ?

This will be an easy question for someone but I just can't find the answer via google or my text books
I have a 30 item measure, an observation schedule, not standardised, and I would like to cite cronbachs alpha for it. I used it in an n=3 within subject design (psychology) where each participant was rated on the measure daily for 70 or so days (I wanted to do time series analysis but that's a different issue) My question is can I use the full 210 or so repetitions to calculate alpha, or just the first occasion for each participant ? I suspect that the former is problematic as the repeated measures will be correlated with each other, but with 210 repetitions you would expect intuitively that the result would be more robust.
thanks in advance