What you should probably do here is delete your second and third loops. Then insert another loop surrounding your trials loop. This outer loop could be called, say, blocks. The blocks loop should be set to random, and linked to a very simple conditions file that just contains the names of the existing conditions files, e.g:
condition_file
conditionsB1.csv
conditionsB2.csv
conditionsB3.csv
Then in the inner trials loop, just put this variable name in the ‘Conditions’ field:
$condition_file
i.e. the outer blocks loop will select a new condition file randomly on each of its three iterations. The inner trials loop will then connect to a new conditions file on each of the three times it runs.
Each loop should have an nReps value of 1, which will yield three blocks of 48 trials.
NB this technique randomises the block order across subjects. A slightly different approach would be required if strict counterbalancing across subjects is needed.