The reason is that a transformer is not a passive object during testing. Applying direct current magnetises the core, applying a high voltage changes the charge distribution in the insulation, and running a tap changer changes the configuration. Each of those states affects the measurements that follow.

Why winding test order changes the readings

The measurements taken on a winding fall into groups by what they depend on. Ratio and resistance depend on the turns and the conducting path. Excitation current and the low frequency part of a frequency response sweep depend on the magnetic state of the core. Insulation resistance and dielectric loss depend on the insulation system and its moisture and charge state.

Where a measurement depends on a state that another test modifies, the order determines what is measured. The most significant interaction is through the core: direct current testing leaves residual flux, and residual flux changes everything that depends on the magnetic circuit.

The second interaction is through the insulation charge. A dielectric test leaves charge distributed in the insulation, and a subsequent measurement taken before the charge has dissipated can produce a value that reflects the charge rather than the insulation condition.

Tests that leave residual effects

The last row matters: the frequency response measurement is the one test that leaves no state behind, which is why it can be performed at the end of the sequence without affecting anything.

DC resistance before and after demagnetisation

Winding resistance is measured at every tap and every phase, and each measurement applies direct current and magnetises the core further. The measurements themselves are only marginally affected by the core state, because the resistance depends on the conducting path rather than on the magnetic circuit, but the accumulated magnetisation is what affects everything after them.

The practical arrangement is to group all the direct current work together, which means all the resistance measurements at all the required taps, and then demagnetise once. Spreading the resistance measurements through the outage means the core is magnetised repeatedly and the demagnetisation has to be repeated before each magnetic measurement.

The demagnetisation should be verified rather than assumed. A repeat excitation current measurement at low voltage, or a repeat low frequency sweep, confirms that the core has returned to the state the reference measurement was taken in.

Sequencing around ratio and excitation tests

Ratio testing is relatively insensitive to the core state, because it compares applied and induced voltages at a level where the winding relationship dominates. That gives it some freedom in the sequence, and it is usually grouped with the resistance measurements because both require the same access and the same tap positions.

Excitation current is at the other extreme. It depends directly on the magnetic circuit, so it has to be measured with the core in a defined state. That means after the direct current work and after demagnetisation, and with the same applied voltage as the reference measurement.

The pairing that produces the most useful evidence is excitation current measured immediately before the frequency response sweep, in the same core state. The two measurements then describe the same magnetic condition, and a deviation in one can be read against the other.

Where SFRA sits in the order

The frequency response sweep sits at the end of the magnetic sequence: after all direct current work, after demagnetisation, and after the excitation current measurement if one is being taken. Its low frequency region is dominated by the core, so a sweep taken on a magnetised core is not comparable with the reference.

Where both a pre-test and a post-test sweep are required, the pre-test sweep should be taken before any direct current is applied. That ordering gives the cleanest comparison, because the first sweep describes the transformer as it arrived.

The sweep also needs the tap position and the connection arrangement recorded. Those are not sequence questions but they have the same effect: a sweep taken at a different tap from the reference cannot be compared with it.

Recording sequence with the results

The test record should state the order in which the measurements were performed, the interval between the direct current work and the magnetic measurements, and whether demagnetisation was performed and verified.

Those fields are what allow a later reviewer to distinguish a condition from an artefact. A low frequency deviation in a sweep that was taken twenty minutes after demagnetisation with a verification measurement is a condition; the same deviation in a sweep taken immediately after a resistance test is more likely to be a residual effect.

Recording the sequence also makes it reproducible. The next outage repeats the order, which is what makes the comparison between outages valid rather than approximate.

Planning the sequence into the outage window

The sequence has a duration, and the duration has to fit the window. The direct current group takes time because each resistance measurement has to stabilise, and the demagnetisation and verification add to it.

The practical arrangement is to schedule the insulation measurements, which are largely independent of the magnetic sequence, alongside the magnetic group rather than inside it. Insulation resistance and dielectric loss can be taken while the core state is irrelevant, which removes them from the critical path.

Where the window is short, the temptation is to move the sweep earlier. That produces a result that cannot be compared with the next outage and costs more in the end than the additional time would have done.

What happens when the order is ignored

The visible consequence is a sweep that does not match its reference and an excitation current that reads high, both of which point at a winding problem. The investigation that follows usually establishes that the core was magnetised, and the outcome is a repeated test and a delayed return to service.

The less visible consequence is a reference that is worthless. If the first sweep taken on a transformer was measured on a magnetised core, every later comparison is against a baseline that does not describe the unit in a defined state.

The third consequence is a disagreement between tests that cannot be resolved. A ratio measurement and a frequency response sweep taken in incompatible states produce a combined evidence set that supports no conclusion, and the resolution is to repeat both. The transformer requirements that define what the measurements are checked against are set out in IEC 60076-1, the frequency response method is defined in IEC 60076-18, the dielectric test framework is published as IEC 60076-3, method documentation is published by instrument suppliers such as OMICRON, and applied interpretation is documented by Transformer Consultants with practice coordinated through CIGRE study committees. The winding test range is grouped under transformer winding test equipment within the wider power transformer testing range.