Well, the distance between the cells shouldn’t be the problem. Since you’re seeing preparation failure, it means the radio conditions (distance, interference, coverage, …) are not relevant. It purely a problem of congestion or HW issues.
If you look deeper in the pm viewer, you should find the details of the preparation failures (2 indicators : the radio problems and the BSS preparation problems, both for the “inc prep fail rate”)
You could also check out the outgoing preparation failure rate for the serving cell, in order to see how many ho requests are counted by both cells.
Also, you could check the “Adjacencies” indicators…
Well, all those are just checkings, I can’t give you an answer. The only time I faced this problem, we changed the whole BTS (it was a micro bts). Changing the SUM might help, who knows.