We study the problem of designing scheduling algorithms for multi-channel (e.g., OFDM-based) wireless downlink systems. While the traditional MaxWeight algorithm is throughput-optimal for this setting, it typically results in large per-user delays when the number of users and the available bandwidth is large. We present a class of algorithms called iLQF (iterated Longest Queues First) that solves this small-delay problem, in an appropriate large deviations setting. The results are validated through both analysis and simulations.