This paper describes the diagnostic measurements involved while performing impulse tests. For most applications during impulse tests, the impulse voltage and impulse current are measured. Because of the nature of impulse test it is very common that Electro Magnetic Interference signals (EMI) are generated. This EMI will influence the measurement results. Therefore these tests are done in a well-shielded high voltage laboratories. The relevant sensors e.g. voltage dividers, current shunts and other components used for the measurements, including traceability issues, will be discussed as part of the presentation.

This paper investigates the practice of correcting the results of no-load loss measurements of power transformers for the effect of distorted supply voltage waveform. A three-phase 700-MVA power transformer is tested twice: first with its low-voltage windings connected line-to-neutral, with nearly sinusoidal test voltage conditions, and then for the second test with its windings connected line-to-line with more distorted voltage waveforms. The test data obtained during the measurement was reproduced at the National Research Council (NRC) Laboratory to demonstrate the limits of the application of the loss correction formula currently used.

The paper reviews the practice of correcting the results of no-load loss measurements of large power transformers for the effect of distorted supply voltage waveforms. Data obtained during testing of large power transformers are reproduced and analyzed at the National Research Council laboratory to ensure that the proper corrected losses are determined. The instrumentation requirements for no-load loss measurements, particularly when the transformers under tests are energized line to line are discussed

The problem of correcting the results of no-load loss measurements of power transformers for the effect of distorted supply voltage waveform on the magnetic flux in the core and hence on both the hysteresis and eddy current loss components of core losses is presented. The instrumentation requirements for no-load loss measurements to ensure proper application of the correction formula are described.

The application of a computer-controlled current-comparator-based load loss standard to the in-situ calibration of two different types of high voltage three-phase transformer loss measuring systems at SMIT Transformers is described. The load loss standard simulates a high voltage inductive load at power frequencies which can provide load currents up to 2000 A at any power factor with a resolution of 1/spl times/10/sup -6/, and at test voltages up to 200 kV. It basically provides a known and stable power loss or reference power, determined by the test voltage and the power factor setting, to the transformer loss measuring system, such that its accuracy can be characterized over different voltage, current, and power factor ranges.

A new diagnostic tool nowadays available with digitally recorded waveforms by lightning impulse tests, is the so called transfer function. For the calculation of the transfer function, two waveforms are recorded. The first waveform is the impulse voltage and the second waveform is the impulse current. Both, waveforms recorded in the time-domain ve transformed to the frequency-domain and the transfer function is calculated. This paper proves that the transfer functions of a sequence of impulse tests with the same test circuit can be compared with each other. This issue has currently much attention all over the world. In our opinion, the differences found between transfer functions are mostly due to hard- and software which is not suitable for the determination of the transfer function. A statement about reliability is only possible if the used technology for determination of the transfer function is beyond any doubt. This paper will compare the transfer functions of the full impulse waveform and the chopped impulse waveform on the same terminal of the transformer and test circuit, which is the worst case condition for the comparison, to prove that comparison of transfer functions is possible.

A new Frequency Response Analysis (FRA) method is described which is being used on power transformers at the KEMA laboratories. This new frequency domain method is applied to digitally recorded time signals, is easy to use and provides high resolution frequency spectra for both magnitude and phase. This tailor-made FRA-method uses a low voltage special impulse source to avoid aliasing. Throughout the paper results from several measured transformers are presented.