The transient electromagnetic (TEM) method is a widely used nonintrusive geophysical method for ground exploration. Due to the mutual inductance between the transmitter (TX) coil and the receiver (RX) coil, the primary field generated by the emission current reduces the accuracy of the near-surface detection. Because the feature signal carrying the target information is mainly distributed in the early secondary-field response, the expanded detection signal will reduce the near-surface investigation capability of the TEM device due to the following three aspects: the loss of the proportion of in the detection signal due to the excessively high primary-field response, the loss of due to the clipping loss, and the reduction of the noise margin in the case in which the detection signal is magnified. These problems are particularly significant in small-loop devices due to the tight coil distribution. The mutual inductance can be reduced by adjusting the relative positions of the TX and RX coils, a configuration called the weak-coupling coil design. We have analyzed the design principle of the weak-coupling coil design and developed a new design scheme — the crossing-loop design. Simulation results indicate that the crossing-loop design performs superiorly in terms of the detection sensitivity and the investigation depth, compared with the nonweak-coupling coil design and other weak-coupling coil designs such as the gradient coils, opposing coils, and the bucking coil design. The experimental results indicate that the crossing-loop design provides much better near-surface investigation capability than the central-loop device with the same TX coil, which is a typical nonweak-coupling coil design.