Performing reliable and reproducible frequency

response measurements on power transformers

Table of Contents
Sweep Frequency Response Analysis (SFRA) has become a standard method to assess the mechanical and electrical
integrity of power transformers. It provides a high sensitivity to evaluate damages, but users often struggle to reach a
high reproducibility. This paper focuses on best practices to perform reproducible SFRA measurements.
Keywords
IEC 60076-18, power transformer, reproducibility, SFRA, Sweep Frequency Response Analysis

Abstract
The Sweep Frequency Response Analysis (SFRA) has become a standard method to assess the mechanical and electri-
cal integrity of the power transformer’s active part. It provides a very high sensitivity to evaluate possible damages after
transportation or for troubleshooting after a specific event such as a near failure with high short-circuit forces. However,
users often struggle to reach a high reproducibility which is essential for a reliable condition assessment. Deviations,
caused by reproducibility issues, can lead to a misinterpretation, unnecessary inspections or cost-intensive maintenance
activities. This paper focuses on best practices in order to perform highly repeatable and reproducible SFRA measure-
ments.

1 Introduction
The Sweep Frequency Response Analysis (SFRA) method was introduced to verify the integrity of the active part of a
power transformer. After manufacturing, power transformers are transported on-site, often over long distances using dif-
ferent types of transportation such as ship, train or truck. Both during transportation and loading from one vehicle to an-
other, the transformer might be exposed to mechanical shocks. Such shocks can also be caused by earthquakes or me-
chanical impacts due to short-circuit forces after a failure. All these impacts can lead to a deformation or partial move-
ment in the active part. Common diagnostic measurements such as transformer turns ratio including exciting currents,
short-circuit impedance at nominal frequency as well as frequency response of stray losses (FRSL) may have disad-
vantages regarding to their sensitivity to detect and prove mechanical deformations. For example, a buckling of a winding
does not typically influence ratio or insulation resistance measurements and is hard to detect in a change of capacitance.
Compared to them the SFRA is the most sensitive method for reliable core and winding assessment [1]. This paper
shows several best practices how to perform SFRA measurements in order to ensure highly repeatable and reproducible
test results.

2 Basics of the SFRA method

The SFRA method comprises a highly repeatable and reproducible frequency response measurement on a power trans-
former and the subsequent comparison with an existing fingerprint, also called a reference measurement [3], [4]. In prin-
ciple three methods are commonly used to assess the measured SFRA traces:
• Time-based (current SFRA results will be compared to previous results of the same unit)
• Construction-based (SFRA of one transformer will be compared to another of the same design)
• Phase-based (SFRA results of one phase will be compared to the other traces of the same unit)

The preferred method is the time-based comparison. However, the fingerprint or base-line measurement is in the major-
ity of the cases not available. Nevertheless, by a simple comparison of the SFRA plots of the phases or by a type-based
comparison, a successful assessment of the results can be achieved. Even if a fingerprint of the transformer is available,
the experience has demonstrated that the comparison has to be carried out carefully because in some cases the devia-
tions observed are not related to deformations, but to measurements under different conditions or due to measurement
mistakes [8]. For overcoming these misleading factors, the comprehensive time-based comparison concept is proposed
in this paper.
2.1 Frequency response measurement

The active part of a power transformer, consisting of the winding, core, insulation and connecting leads, forms a complex
electrical network as indicated in Figure 1. Such a network has unique characteristics which can be visualized by the fre-
quency response: A frequency-variable sinusoidal low-voltage signal of, for example 10 V, is applied to one terminal and
the response (U2) is measured at another terminal (Figure 2). In order to measure the amplitude, phase and frequency of
the injected signal, a reference measurement channel (U1) is connected to the same injection point as the source [2].
The frequency response consists of amplitude, ratio and phase difference between both terminals.

Figure 1: Simplified R-L-C equivalent circuit of a power transformer

The frequency response can be measured in different ways in order to gather more information for a sophisticated as-
sessment. The most common approach is the open-circuit measurement. Thereby, the frequency response is measured
between two terminals of the same voltage level, leaving all other terminals open. When shorting the terminals of the
other voltage level (for example, the low-voltage winding when measuring the high-voltage windings), a short-circuit
measurement is performed. A capacitive inter-winding measurement describes a test between two windings on the same
core limb (for example the high- and low-voltage winding), while all other terminals are open. An inductive inter-winding
measurement is also performed between two windings on the same core limb, whereas the measurement clamps are
mounted on each winding terminal and the other end of the winding is connected to ground.

Figure 2: Typical setup of a sweep frequency response analysis

2 | 12
2.2 Analysis methods for SFRA measurements

Depending on their main influences, different failure modes will be revealed stronger in different frequency ranges. As an
example, core phenomena will influence the low-frequency region whereas connection issues in the very high-frequency
range above 1MHz [1]. Experience shows that setup issues, such as not following the shortest braid concept can influ-
ence the frequency response even at 500kHz. However, it is difficult to provide a general table with shows the relation
between the frequency range and transformer characteristics as there are too many factors which are influencing the
frequency range (e.g. MVA rating, winding type, voltage level, etc.). Basic references can be found within the CIGRE
brochure. Different analysis tools can be used based on mathematical indices [5] or characteristic changes within the
measured curves [6].
For every analysis, a fingerprint or baseline measurement is necessary. If available, a comparison should always be
made to a previous measurement of the same transformer using the same configuration [7], a so called time-based com-
parison. Such a reference measurement can originate, for example, from commissioning tests or in-depth testing on site.
Alternatively, if no reference measurement from this transformer is available, the frequency response can be compared
with a sister asset. Sister assets typically have a very similar, but not identical, frequency response as shown in Figure 3.
Therefore, small deviations are acceptable and do not necessarily indicate a problem.

Figure 3: Frequency response measured in an open-circuit test on sister transformers (200 MVA, 230 kV interleaved disk winding)

In cases where even no trace of a sister asset is available, phase-to-phase comparisons might be applied. A good com-
parison is only possible for a symmetrical design, which is not exactly given for common designs. Even larger deviations
can be caused by constructive differences between phases. Phase-to-phase comparisons, therefore, require the greatest
amount of experience. Typically, the centre phase includes the most deviation, whereas the other two phases overlay
with reasonable similarity. The main deviations between the centre phase and outer phases are expected at lower fre-
quencies, which are mainly affected by the core due to the different flux paths.

3 | 12
 Figure 4: Comparison of the frequency response traces measured on three phases of the same transformer (200 MVA, 230 kV inter-
leaved disk winding)

Figure 5: Schematic of the recommended setup according to Figure 6: Example of shortest-braid connection concept -
IEC 60076-18 measurement cables (black) are connected to the bus-
hing terminal and the grounding braid to the terminal and
bushing flange

The IEC 60076-18 standard describes the recommended procedure for a proper and reproducible measurement setup in
detail (Figure 5). It is recommended to use double-shielded coaxial cables which are connected to the bushing terminal.
From here, a connection to the flange or tank should be installed on a low inductive ground, preferably using a flat, wide
aluminum braid instead of a simple wire. As explained in [10], braids have a large surface, a low inductance, and the
mesh reduces the considerable skin-effect above 80 kHz. As a consequence, the braid structure provides a better con-
ductivity for high frequencies, resulting in a more efficient noise suppression towards ground compared to the use of sim-
ple wires.
The length of the ground connection influences the frequency response. To achieve a high reproducibility, it is suggested
to use the shortest possible length by pulling the braid tightly along the body of the bushing as shown in Figure 6.
Besides the connection technique itself, it is important to establish a proper electrical contact between the terminal re-
spectively flange and the measurement clamp used. Cleaning the terminal and removing lacquer layers help to reduce

4 | 12
the contact resistance. Modern SFRA devices provide a ground-loop check to ensure proper connections with a low con-
tact resistance to ground.

4 Influencing factors of the frequency response

As discussed above, it is essential for a comparative method such as SFRA, to provide the highest confidence in exclu-
ding influencing factors related to the measurement setup or external factors which will be described within the following
chapter. For the sake of completeness, it is worth to mention that for the described use-cases the OMICRON FRANEO
800 SFRA test system was used and the end-to-end open circuit measurement was conducted on various assets.

4.1 Factors on the measurement setup

Shorting and grounding of tertiary windings and separate neutral terminals

The measurement type “open-circuit” or “short-circuit” defines whether the terminals of the opposite voltage level have to
be short-circuited or not. That means, when measuring the high voltage side, it defines whether to short circuit the low
voltage terminals or not.

Figure 7: Influence of grounded (blue) and ungrounded (red) tertiary on measured LV open-circuit curves

The measurement type does not provide information on how to handle separate neutral terminals or tertiary windings
which significantly influence the measured frequency response. This includes floating, closed or grounded tertiary win-
dings. Figure 7 shows the deviations between two open-circuit measurements made on the low-voltage windings with
grounded and ungrounded tertiary winding. Various deviations, especially within the area of mutual coupling (interaction
of the windings) on the measured frequency response can be observed. In general the frequency response for the mag-
netization inductance and parallel capacitance remain unaffected. Therefore, it is suggested to leave all other terminals
open and ungrounded as recommended in the IEEE Standard C57.149 [3].

Measurement direction
The measurement direction, meaning in the case of star-connected power transformers from phase to neutral or neutral
to phase, significantly influences the high-frequency behavior as shown, for example, in Figure 8. If not specified differ-
ently, it is suggested to have the source and reference lead connected to the phase terminal and the response lead to
the neutral terminal[3], [4].

5 | 12
 Figure 8: Influence measuring direction, source to phase (blue) vs. source to neutral (red)

Output voltage
In the low-frequency range, the frequency response is dominated by the core magnetization inductance and, therefore,
depends on the output voltage as shown in Figure 9. The residual curve is not affected by the output voltage as the
transformer windings can be considered as a linear system, which is, in principle, unaffected by the output voltage.

Figure 9: Effect of the selected output voltages; influence of different output voltages on magnetization inductance Lm

Grounding and measurement leads

Different procedures to connect the measurement and grounding leads are explained in IEC 60076-18 [4]. The most
common way is to use specially-designed clamps connecting the double-shielded BNC-cable to the phase terminal and
to apply a braid respectively wire for grounding the shield between the clamp and the bushing flange. As described in
section 3, using a braid instead of a wire will significantly reduce the noise influence, especially around mains frequency.
Furthermore, the length of the ground lead is essential for the low-frequency region as shown in Figure 10. When using

6 | 12
leads with a fixed length, the position of the lead influences the frequency response. Therefore, the concept of adapting
the ground lead in order to ensure the shortest path to ground, provides the highest degree of reproducibility.

Figure 10: Influence of different connection techniques, lowest noise influence and highest repeatability by using a ground braid in the
shortest path concept (blue), wire connection (green) and ground braid with a bigger loop (red)

Tap changer position and bushings

In accordance with the applicable standards, SFRA tests must be performed with the same transformer configuration,
including bushings or tap changer position. Sometimes test bushings are used during factory acceptance tests and the
final bushings are mounted on-site. When comparing the frequency response measured in the factory with the one
measured on-site, deviations will occur most probably in the high-frequency range. Changing the tap changer position
will cause continuous changes of the curve shape over a wide frequency range, as shown in Error! Reference source
not found.. It is suggested to perform measurements for each relevant phase on the lowest, highest and middle position
of the tap changer, while switching continuously from the higher to lower position Error! Reference source not found.,
Error! Reference source not found..

7 | 12
 Figure 11: Influence of tap changer position (tap 1: red, tap 2: green, tap 3: black) on the frequency response of a power transformer

Contact between bushing terminal and ground lead

One of the most typical connection errors is an unwanted contact of the ground lead with the bushing terminal. Such an
error influences the frequency response mainly on the higher frequencies as shown in Figure 12. It is recommended to
use an insulating cover for braids in order to avoid this short circuit.

Figure 12: Frequency response measured on HV side in an open-circuit test; proper connected (red) and short-circuit between termi-
nal and ground at the reference (blue)

4.2 Other influencing factors

Residual magnetization
Residual magnetization is a frequently occurring phenomenon caused by previous measurements, such as a DC winding
resistance measurement. This can be avoided by demagnetization before performing SFRA tests. The influence reveals,
in particular, in the very low-frequency range as shown in Figure 13 where the core resonance is shifted to the right. The
other parts of the SFRA trace are unaffected by this phenomenon. Therefore, residual magnetism can be simply identi-
fied and usually does not influence further analysis.

8 | 12
 Figure 13: Frequency response of a power transformer measured in an open-circuit test before (red) and after (black) demagnetiza-
tion

Insulating fluid

A power transformer should always be measured with the same configuration as it will be on site. This includes the insu-
lation fluid, as it significantly influences the frequency response. When comparing SFRA measurements of an unfilled
and oil-filled power transformer, as shown in Figure 14, a systematic shift of characteristic frequencies can be observed.
This is caused by the different dielectrics (air/gas with r,gas = 1 vs. oil with r,oil = 2,2) and roughly corresponds to the
theoretical value which can be calculated by the square root of the relative permittivity of mineral oil [11].

Figure 14: Measured HV traces with oil filled (green) and unfilled (blue) tank

Temperature
Ambient environmental conditions, such as temperature, can influence the measured frequency responses. However,
investigations have shown that the thermal coefficients for shifting the resonance points with temperature are very small
[12]. As a consequence, the shift can be neglected in the typical temperature range between 15°C (59°F) and 70°C
(185°F).

9 | 12
5 Conclusion
The importance of a suitable connection technique was pointed out. The advantages of the technique suggested by IEC
60076-18 in comparison with other techniques were discussed. Besides a high degree of reproducibility, the use of
ground braids instead of simple wires helps to avoid the influence of narrow-band noise around mains frequency and
increases the reproducibility, especially in the high-frequency range above 500 kHz.
Influencing factors of the frequency response were named, described and examples for the change in the curve shape
were given. This includes their influence on the measurement setup, such as shorting or grounding of tertiary windings,
measurement direction, output voltage, connection technique or tap changer position. Besides this, other effects were
discussed, such as residual magnetization, the influence of insulation fluid or changes due to temperature and humidity.

10 | 12
6 References
[1] Cigré WG A2/26, “Mechanical condition assessment of transformer windings using Frequency Response Analy-
sis (FRA)”, Brochure 342, Paris 2008
[2] T. Leibfried, K. Feser, Monitoring of Power Transformers using the Transfer Function Method, IEEE Transac-
tions on Power Delivery, Vol. 14, No. 4, October 1999, pp.1333-1341.
[3] IEEE Std C57.149, “IEEE Guide for the Application and Interpretation of Frequency Response Analysis for Oil-
Immersed Transformers”, 2013
[4] IEC 60076-18, “Power transformers – Part 18: Measurement of frequency response”, 2012
[5] NCEPRI, “Application Guideline for Transformer Winding Distortion Test Technology”, China 1999
[6] Cigré WG A2/26, “Mechanical condition assessment of transformer windings using Frequency Response Analy-
sis (FRA)”, Electra N°228, Paris 2006
[7] J. Christian, K. Feser, Procedures for Detecting Winding Displacements in Power Transformers by the Transfer
Function Method, IEEE Transactions on Power Delivery, Vol. 19, No. 1, January 2004, pp.214-220.
[8] Juan L. Velásquez, et al., “Noise in FRA Measurements: Sources, Effects and Suppression,” OMICRON DMPT
Workshop 2009.
[9] S. Tenbohlen, R. Wimmer, K. Feser, A. Kraetge, M. Krüger, J. Christian: The influence of grounding and con-
nection technique on the repeatability of FRA-results, Proceedings of the XVth International Symposium on High
Voltage Engineering, University of Ljubljana, Ljubljana, Slovenia, August 27-31, 2007.
[10] R. Wimmer, S. Tenbohlen, K. Feser, M. Michael Krüger: „Richtlinien für den Messaufbau für eine
hohe Reproduzierbarkeit der FRA-Messergebnisse“, ETG Fachtagung Diagnostik elektrotechnischer Be-
triebsmittel, Kassel, Germany 19.-20. September 2006
[11] J. Christian, R. Wimmer: “Comparability of transfer function results”, European Transactions on Electrical
Power 2006, issue 16, pages137-146
[12] R. Wimmer, K. Feser, J. Christian: “Reproducibility of Transfer Function Results”, XIIIth International Sympo-
sium on High Voltage Engineering, Delft, Netherlands, August 25-29, 2003

Leads
“SFRA has become a standard method to assess the integrity of the power transformers.”

“The connection technique is essential to achieve a high degree of reproducibility of SFRA measurements.”

“Adapting the ground lead to ensure the shortest path to ground provides the highest degree of reproducibility.”

“The use of ground braids instead of simple wires helps to increase the reproducibility.”

“Demagnetization before performing SFRA tests ensures reliable results.”As can be seen, modern SFRA test devices
and an appropriate software can provide reliable SFRA measurement even in harsh conditions and help to improve the
condition assessment of power transformer

11 | 12
OMICRON is an international company serving the electrical power industry with innovative
testing and diagnostic solutions. The application of OMICRON products allows users to as-
sess the condition of the primary and secondary equipment on their systems with complete
confidence. Services offered in the area of consulting, commissioning, testing, diagnosis and
training make the product range complete.
Customers in more than 150 countries rely on the company’s ability to supply leadingedge
technology of excellent quality. Service centers on all continents provide a broad base of
knowledge and extraordinary customer support. All of this together with our strong network of
sales partners is what has made our company a market leader in the electrical power industry.

For more information, additional literature,

and detailed contact information of our
worldwide offices please visit our website. 12 | 12

www.omicronenergy.com © OMICRON