Iron-rich meteorites are significantly underrepresented in collection statistics from Antarctica. This has led to a hypothesis that there is a sparse layer of iron-rich meteorites hidden below the surface of the ice, thereby explaining the apparent shortfall. As standard Antarctic meteorite collecting techniques rely upon a visual surface search approach, the need has thus arisen to develop a system that can detect iron objects under a few tens of centimetres of ice, where the expected number density is of the order one per square kilometre. To help answer this hypothesis, a large-scale pulse induction metal detector array has been constructed for deployment in Antarctica. The metal detector array is 6 m wide, able to travel at 15 km h-1 and can scan 1 km2 in ~11 hours. This paper details the construction of the metal detector system with respect to design criteria, notably the ruggedization of the system for Antarctic deployment. Some preliminary results from UK and Antarctic testing are presented. We show that the system performs as specified and should reach the pre-agreed target of the detection of a 100 g iron meteorite at 300 mm when deployed in Antarctica.