The dielectric frequency response test (FDS or DFR) measures the capacitance and tan δ of paper-oil insulation in a sweep from kilohertz down to millihertz. The shape of the curve lets you estimate the moisture content of the cellulose and the oil conductivity without opening the transformer. It applies to power transformers, bushings and instrument transformers with oil-impregnated paper, and it is described in the IEEE C57.161 guide.
What the test is
Dielectric frequency response measures the capacitance and the tan δ of an insulation system at many frequencies, not only at 50 or 60 Hz. The sweep runs from frequencies in the kilohertz range down to very low frequencies, in the millihertz range. It is known as FDS (frequency domain spectroscopy) or DFR (dielectric frequency response); both names describe the same measurement. On the frequency response of a grid with high renewable generation, read frequency response with high renewable penetration.
The instrument applies a low AC voltage, far below service voltage, between two windings or between a winding and ground, and measures the current with a guard circuit. Because the voltage is low, the test does not stress the insulation. The result is a curve of tan δ against frequency.
What it measures and which defect it detects
In paper-oil insulation, the curve combines three phenomena: the conductivity of the oil, the moisture and aging of the cellulose, and the interfacial polarization between paper and oil, which depends on geometry. Each one dominates a different part of the sweep, and temperature shifts the whole curve along the frequency axis. The measured curve is compared with a model of the transformer insulation (spacers, barriers and oil ducts), and the fit returns two numbers:
- Moisture content of the cellulose, in percent by weight.
- Oil conductivity, linked to its contamination and aging.
That is the value of the test. Moisture in the paper accelerates its aging and lowers dielectric strength, and the classical estimate, based on oil moisture and equilibrium curves, is uncertain: water distributes itself between paper and oil depending on temperature and takes time to settle. CIGRE addressed this problem in its brochures 254 (2004) and 349 (2008).
The test also helps decide whether to dry out a unit, check afterward whether the drying worked, and explain a high tan δ at 60 Hz: whether it comes from wet paper, a conductive oil, or both.
It does not detect localized defects such as partial discharges, shorted turns or winding deformation.
Which equipment and voltage levels it applies to
| Level | Typical equipment | Factory | Commissioning | Maintenance |
|---|---|---|---|---|
| LV | Not applicable | — | — | — |
| MV | Large oil-filled power transformers | Occasional | As a reference | When moisture is suspected |
| HV | Transformers, autotransformers, reactors, OIP bushings | Occasional | Yes, as a fingerprint | Yes |
| EHV | Transformers and reactors, bushings, oil-paper CTs | Occasional | Yes, as a fingerprint | Yes |
Where it does not apply:
- Dry-type, resin-encapsulated transformers and insulation without impregnated cellulose: the moisture models are built for paper-oil systems.
- Low-voltage equipment and small distribution transformers, where the cost of the test is not justified compared with insulation resistance and oil analysis.
- Equipment with alternative liquids (esters): the measurement can be made, but the moisture estimate needs a model with the properties of that liquid; interpret it with caution.
On OIP high-voltage bushings, FDS helps assess the moisture of the condenser core, because low frequencies are more sensitive to it than tan δ at 60 Hz.
The standard behind it and how results are interpreted
The reference guide is IEEE C57.161-2018, Guide for Dielectric Frequency Response Test. It gives recommendations on instrumentation, procedure and analysis for liquid-filled transformers, in the factory and in the field, and it covers the factors that affect the moisture estimate: winding configuration, measurement setup, type of cellulose, low-molecular-weight acids and background noise. CIGRE developed it earlier in technical brochure 414 (2010), Dielectric response diagnoses for transformer windings, which estimates the errors due to pressboard type, oil conductivity and oil acidity.
There is no pass or fail acceptance criterion. The result is interpreted as follows:
- Estimated moisture of the cellulose, compared with the classification used by the software or the guide applied (dry, moderately wet, wet). The limits of these categories vary between sources; always state which one was used.
- Oil conductivity, consistent with the physicochemical analysis of the oil.
- Trend against the commissioning fingerprint or a previous test.
- Consistency with tan δ at 60 Hz, the oil analysis and the history of the unit.
What can distort the result
- Temperature. The curve shifts with temperature. The insulation temperature must be recorded accurately and entered in the model; an error of a few degrees changes the estimated moisture.
- Assumed geometry. The model needs the proportion of barriers, spacers and oil. A poorly chosen design value biases the result.
- Aging and acids. Aging products in the paper can look like moisture, as IEEE C57.161 and CIGRE 414 warn.
- Interference from energized lines, which matters more at low frequencies.
- External connections left in place, and wet or dirty bushing surfaces.
- Thermal equilibrium. Measuring right after de-energizing, while the transformer is still cooling, leaves the measurement under changing conditions.
What to do if the result is poor
- Check the temperature entered and the geometry model; repeat if there was noise.
- Compare with the oil moisture analysis and the power-frequency tan δ.
- If moisture is high, work out the source with the manufacturer (seals, breather, gaskets) and a drying treatment; measure again after the treatment.
- If the oil conductivity is high, review the dielectric oil analysis and consider reclaiming or replacing the oil.
- Check the conservator and silica gel breather, a frequent path for moisture ingress.
Instruments for this test
This test is performed with instruments from recognized manufacturers; these guides explain what each one measures. In the quotation we recommend the test and the instrument your case needs.
Dielectric response with Tiza Group
Our team performs insulation diagnostic tests on power transformers, reactors and bushings at high and extra-high voltage, in substations and industry, and interprets them together with the oil and the history of the unit. Field testing is done in Colombia, or elsewhere on request. See the scope under power transformer testing, read about the other tests in the primary equipment testing guide, or write to us.
References
- IEEE Std C57.161-2018, Guide for Dielectric Frequency Response Test.
- CIGRE Technical Brochure 254, Dielectric response methods for diagnostics of power transformers, 2004.
- CIGRE Technical Brochure 414, Dielectric response diagnoses for transformer windings, 2010.
- CIGRE Technical Brochure 349, Moisture equilibrium and moisture migration within transformer insulation systems, 2008.
- IEEE Std C57.152-2025, Guide for Diagnostic Field Testing of Liquid-Filled Power Transformers, Regulators, and Reactors.
Frequently asked questions
What is the difference between tan δ at 60 Hz and dielectric frequency response?
Tan δ at power frequency is a single point: it tells you the insulation has losses, but not why. The frequency sweep separates the effect of the oil, the cellulose and the geometry, and with a model it lets you estimate the moisture in the paper.
Does FDS replace oil moisture analysis?
No, it complements it. Moisture in the oil changes with temperature and takes time to reach equilibrium with the paper, so estimating cellulose moisture from an oil sample carries uncertainty. FDS estimates the moisture of the solid insulation from its own dielectric response, and the oil is still analyzed for its own reasons.
How long does a dielectric response test take?
It depends on the lowest sweep frequency and on the temperature: the lowest frequencies need very long cycles. Many instruments combine time-domain (PDC) and frequency-domain measurements to shorten the test.