Insulation resistance tests apply DC. Tan delta testing applies AC at or near power frequency, which is how the insulation is stressed in service, and measures how much the insulation departs from being a perfect capacitor. It is one of the standard diagnostic tests for transformers, bushings, instrument transformers, capacitors, arresters, cables and rotating machine windings.
What Is Being Measured
Insulation between two electrodes behaves mostly as a capacitor, so the test current leads the voltage by close to 90°. Losses in the insulation — conduction through moisture and contamination, dielectric polarisation losses and, at higher stress, partial discharge — add a small current in phase with the voltage. Tan δ is the ratio of the in-phase current to the capacitive current. The instrument also reports the capacitance itself.
- A rising tan delta points to moisture, contamination or ageing spread through the insulation.
- A change in capacitance points to a physical change. In a condenser bushing, for example, shorted grading layers increase capacitance.
- Tan delta rising with test voltage ("tip-up") can indicate ionisation in voids.
UST and GST Connections
A tan delta set has a high-voltage output, measuring inputs and a guard. How they are connected decides which part of the insulation is measured:
- UST (ungrounded specimen test): current is measured in a lead returned from an electrode that is not earthed, such as a bushing test tap. Currents to earth are excluded. This is used for bushing C1 insulation and transformer inter-winding insulation.
- GST (grounded specimen test): current is measured in the earth return, so all insulation from the energised electrode to earth is included. Guarding can exclude selected parts.
On site, transformer tanks are earthed, which is why reverse (GST) connections are normally used for transformer measurements. MEWOI's application note makes the same point: the forward connection is used only for fully insulated specimens.
Interference and Variable Frequency
In an energised substation, capacitive coupling from nearby live conductors injects power-frequency current into the test leads, and this can swamp the small loss current. Modern sets avoid it by testing at frequencies slightly away from mains, for example 45 Hz and 55 Hz, and combining the two results digitally. The interference at exactly 50 Hz is then rejected by filtering.
MEWOI's dielectric loss testers use this method. The supplier specifies an internal HV source up to 12 kV, capacitance accuracy of ±(1% of reading + 1 pF), tan δ accuracy of ±(1% of reading + 0.00040), automatic 45/55 Hz or 49/51 Hz dual-frequency operation, and a CVT self-excitation output for capacitor voltage transformers. One model is specified to keep that accuracy under interference of up to 200%.
Interpreting Results
Results depend on temperature, so record insulation temperature and use temperature correction where the manufacturer or standard provides it. Compare bushings against their nameplate C1 values, compare phases with each other, and track each item over time. A stable result in an acceptable range matters more than any single number.
The same principle is used to test insulating oil in a three-electrode cell — see Transformer Oil Testing — What Tan Delta, Interfacial Tension, Moisture and Acid Number Each Tell You.
Frequently Asked Questions
Is tan delta the same as power factor testing? They are closely related. Power factor is cos φ, the in-phase current divided by total current, while tan δ is in-phase current divided by capacitive current. For the low values found in good insulation the two numbers are almost identical.
Can it be used on cables? Yes. Long cables have high capacitance, so very low frequency (VLF) tan delta at 0.1 Hz is commonly used to keep the test supply a practical size.