Sweep frequency response analysis is one of the few diagnostic tests that can detect mechanical damage to transformer windings without opening the tank. It does not measure insulation condition or turns ratio — it measures the physical geometry of the winding and core assembly by treating the transformer as a passive electrical network and mapping how that network responds across a wide frequency range. A winding that has moved, deformed, or developed a short gives a different frequency response signature than an intact winding, and SFRA finds that difference.
A transformer winding is a distributed network of inductances and capacitances. The inductance comes from the winding geometry and its relationship to the core. The capacitance comes from the turn-to-turn, disc-to-disc, and winding-to-ground distributed capacitances throughout the winding structure. Together they form a complex network with a characteristic frequency response — resonant peaks and nulls at specific frequencies that are determined by the physical dimensions and geometry of the winding.
SFRA injects a swept sinusoidal signal across a frequency range, typically from around 20 Hz to 2 MHz, into one terminal of the winding and measures the transfer function — the ratio of output to input — at another terminal. The result is a frequency response curve plotted against frequency on a logarithmic scale. This curve is the fingerprint of that specific winding in its current physical condition.
When something changes physically — a winding disc moves, turns collapse from a through-fault, core laminations shift from transport damage, or a loose connection develops inside the tank — the distributed inductance and capacitance values change. The frequency response curve shifts. Comparing a post-event curve to the baseline fingerprint reveals where the discrepancy is, and how severe it is, before the tank is opened.
Winding deformation from through-faults. When a transformer clears a close-in external fault, the electromagnetic forces acting on the winding conductors can deform the winding — radially outward on the outer winding, radially inward on the inner winding, or axially from unbalanced ampere-turn distribution. Deformation changes the winding geometry and produces a detectable shift in the mid-to-high frequency portion of the SFRA signature, typically above 10 kHz.
Shorted turns. An inter-turn short reduces the effective inductance of the winding section where it occurs. This appears as a change in the low-frequency inductance measurement and a shift in the resonant structure of the response curve. SFRA is not the primary test for shorted turns — excitation current testing is more sensitive to that specific fault — but SFRA can corroborate or flag a suspicion established by other tests.
Core movement. Core laminations that have shifted from a through-fault or transport shock change the magnetic coupling geometry between winding and core. Core-related changes appear primarily in the low-frequency end of the SFRA response, below 1 kHz, where the inductive impedance of the winding dominates. A shift in the low-frequency plateau of the response curve, without corresponding changes in the mid and high frequency regions, points to the core rather than the winding.
Loose or open connections. A loose bushing lead, a broken tap connection, or an open internal joint produces a sharp change in the frequency response — often a distinct shift in the high-frequency portion of the curve that is easy to distinguish from gradual deformation signatures. Open connections are generally easier to identify in SFRA data than deformation, because the change is discrete rather than distributed.
Transport damage on new or relocated transformers. SFRA is routinely performed as part of acceptance testing on large power transformers received after long-distance transport. A transformer that sustained shock or tilt during shipment may show internal damage that is invisible from the outside. Comparing the field SFRA measurement to a factory baseline measurement — which manufacturers increasingly provide — identifies any transport-induced changes before the transformer is energized.
SFRA results are only useful in comparison. A single curve with no reference means nothing — SFRA does not have pass/fail threshold values the way power factor or turns ratio testing does. The question is always: does this curve match the reference?
Time-based comparison compares the current measurement to a prior measurement on the same unit — ideally a factory fingerprint or a known-good baseline taken after acceptance. This is the most diagnostic comparison because it uses the same physical unit as its own reference, eliminating all manufacturing variation.
Phase-to-phase comparison compares the three phases of the same transformer to each other. On a three-phase unit with geometrically identical winding structures on each phase, the frequency response should be nearly identical across phases. A significant deviation on one phase relative to the other two points to a phase-specific problem. This comparison works even without a historical baseline.
Sister unit comparison compares the test unit to an identical unit from the same manufacturer, same design, and same rating class. This is the least reliable comparison method because manufacturing variation between individual units introduces differences in the curve that can be mistaken for damage. It is used when no baseline exists and no phase comparison is possible.
The most important trigger for SFRA is a through-fault event. Any time a transformer has cleared a close-in fault — whether the fault was on the secondary bus, a feeder cable, or a downstream transformer — the mechanical integrity of the winding should be verified before the transformer continues in service. The electromagnetic forces during a fault event can be substantial even when the transformer protection operated correctly and quickly.
SFRA is also standard practice on transformer acceptance testing, particularly for large units or units received after long-distance transport. Running a baseline measurement during acceptance gives you a reference fingerprint to work against for the entire service life of the unit.
Periodic SFRA as part of a comprehensive maintenance program — even without a triggering event — catches slow mechanical changes that develop over time from cumulative load cycling and thermal fatigue. On critical transformers where an unplanned outage has severe consequences, periodic SFRA is inexpensive insurance.
SFRA requires a baseline to be useful. Without a reference curve, the test generates data that cannot be interpreted with confidence. If your transformer fleet has no existing SFRA baselines, the first priority is to run and document baselines on critical units while they are known to be in good condition — even if no fault event has occurred. A baseline taken today is useful the next time something happens.
SFRA does not detect every fault type equally well. Early-stage winding deformation produces subtle curve shifts that require skilled interpretation. The test is sensitive to measurement setup — lead placement, connection method, and the presence of nearby metallic structures all affect the result. Results should be evaluated by a technician experienced with SFRA interpretation, not simply flagged by automated pass/fail software.
SFRA is one test in a diagnostic program, not a standalone assessment. A transformer with a changed SFRA signature should be evaluated alongside power factor test results, DGA, turns ratio, and winding resistance before a decision is made about continued service or repair. SFRA identifies that something has changed mechanically; the other tests characterize the insulation condition and electrical integrity of the same unit.
We perform sweep frequency response analysis on power transformers across Florida and the Southeast — post-fault assessment, acceptance testing, and periodic baseline programs. Send us your equipment details and we’ll respond within one business day.