This depends on two factores mainly:

1) Do u see checkpoint not complete in alert files, that means checkpoint is happening very fast , increase the size of redo log files .
If you see in v$sysstat, that background checkpoints completed do not match with background checkpoints started, tune your DBWR.

2) How often you want checkpoint to occur, how large is your redo log file, how large is your db_block_buffers, depending on that you can tune your checkpoint.