As　a　central　device　of　a　power　system,　the　safe　running　of　a　power　transformer　is　considered　to　be　significance.　Through　the　statistics　of　onsite　faults,　the　amount　of　the　transformer　winding　deformation　caused　by　short-circuit　faults　within　transformer　winding　is　maximum,　therefore　the　detecting　on　the　short　circuit　faults　is　important　to　improve　the　reliability　of　power　transformers　in　service.　In　this　paper,　a　data　acquisition　system　has　been　developed　based　on　a　signal　source　and　a　power　amplifier,　and　the　short　circuit　faults　are　simulated　by　using　a　three-phase　transformer　in　lab,　as　well　as　the　sweep　frequency　impedance　(SFI)　curves　are　extracted　and　analyzed　to　find　out　the　inner　relevancy.　The　result　indicates　that　the　resonance　frequencies　and　their　amplitudes　of　the　sweep　frequency　response　curves　vary　as　the　degree　and　type　of　short　circuit　faults　differ,　and　by　analyzing　the　relationship　between　each　curve,　some　preliminary　conclusions　can　be　drawn　here.　This　investigation　has　some　degree　of　guiding　significance　in　predicting　structural　faults　of　windings　in　the　transformers　and　is　the　basis　for　further　experimental　studies.