VLF & Cable Testing

Tan Delta Testing: What the Number Means, and When It Misleads

VA-TEK Engineering VA-TEK Engineering 11 min read

Two assets return a tan delta of 0.42%. On a 20-year-old oil-impregnated cable that is a passing result and a reason to schedule the next test. On a brand-new resin-cast transformer bushing it is a reason to stop and open the unit. Same number, opposite conclusions — and that single fact explains why so much tan delta testing is misread.

Tan delta is one of the most useful measurements in electrical maintenance, and one of the easiest to misuse, because the number only means something when you know what it was measured on, at what frequency, and against what history. This guide separates those three things.

The short answer

  • What it measures: the ratio of resistive to capacitive current through insulation — how much energy the dielectric loses as heat. It is a bulk quantity, averaged over the whole insulation volume.
  • What it does not measure: insulation strength. A low tan delta does not prove a cable or a winding will hold voltage. It proves the material is not broadly lossy.
  • Why frequency matters: capacitive current scales with frequency, resistive current does not. Measuring at 0.1 Hz makes the resistive part a much larger share of the total, which is why low-frequency tan delta resolves loss that a 50 Hz reading buries.
  • What makes a value “acceptable”: insulation type, temperature, geometry, and the asset’s own history. There is no universal figure. Treat any single number quoted without those qualifiers as unreliable.

1. What tan delta actually measures, and what it does not

Every insulation behaves a little like a lossy capacitor. Apply an AC voltage and two currents flow: one charging the capacitance (90° out of phase with the voltage), one leaking through the dielectric as real loss (in phase with the voltage). The tangent of the angle between total current and the purely capacitive component is the dissipation factor — tan delta, or tan δ. It is dimensionless and usually quoted as a percentage or in units of 10⁻⁴.

Physically, that loss comes from dipole friction in the polymer, from ionic conduction in moisture and contamination, and from interfacial polarisation where materials meet. All three rise as insulation ages. Water trees in XLPE raise local conductivity; moisture in oil-paper raises ionic conduction; a contaminated resin surface adds a conductive path. Tan delta is sensitive to all of them, which is what makes it a good condition indicator.

What it is not is a strength test. It tells you nothing about the voltage at which the insulation would break down. A cable with a large population of small water trees can show a modest tan delta and still hold voltage; a cable with one badly made joint can show a perfectly normal bulk tan delta and fail next month. That second case is why tan delta is always paired with a partial discharge measurement rather than used alone — the two have opposite blind spots, and this comparison works through which test answers which question.

2. Why frequency changes the number completely

Capacitive current is proportional to frequency: Ic = 2πfCV. Resistive current is essentially frequency-independent at the frequencies in question. So at 50 Hz the capacitive current is enormous and the resistive leak is a tiny fraction of it; the loss angle is small and hard to resolve. Drop the frequency to 0.1 Hz and the capacitive current falls by roughly 500 times while the resistive current stays put — the loss angle grows by the same factor and the measurement becomes far more sensitive.

That is the whole reason tan delta and very low frequency testing are usually discussed together. The same 0.1 Hz source that stresses the cable also produces a much better-conditioned loss measurement. It is not a compromise; at low frequency the measurement is easier, not harder.

The practical consequence is that a tan delta figure is meaningless without its frequency. A value measured at 50 Hz and a value measured at 0.1 Hz on the same asset are different quantities. Instruments that offer several VLF frequencies (0.1, 0.05, 0.02 and 0.01 Hz are the common set) let you read tan delta at each one, and the change across frequencies is itself diagnostic — a strongly frequency-dependent loss points at specific ageing mechanisms that a single reading conceals.

3. Cable and transformer: two different tests sharing one name

This is where most confusion starts. “Tan delta testing” describes at least three distinct field procedures:

Application Typical energisation What the result tells you
Shielded power cable 0.1 Hz VLF, from a cable test set Bulk condition of the whole cable run; trend for replacement planning
Transformer windings and bushings Power frequency, from a dielectric loss / capacitance bridge Moisture and ageing in the insulation system; bushing condition
Rotating machine windings Power frequency, per phase or per winding Ground-wall insulation condition; contamination and moisture

They are not interchangeable. A cable set measures tan delta on a long distributed capacitance at 0.1 Hz; a dielectric loss bridge measures it on a lumped winding at 50/60 Hz with a guard circuit to separate the test object from stray capacitance. The instruments differ, the interpretation tables differ, and the historical data you are comparing against comes from a different tradition.

For cable work, the measurement usually ships inside the withstand source — the VLF-34TD and VLF-45TD each carry an integrated tan delta channel with 1.0 × 10⁻⁴ accuracy, so the withstand and the diagnostic reading come from one energisation. The VLF-30-90 family is the AC-withstand-only counterpart for commissioning work where a loss measurement is not required.

For transformers and bushings, it is a separate instrument. The TDC-3100 measures tan delta and capacitance at power frequency on apparatus rather than cable. If your programme covers both, that is two tools, not one — and a specification that asks for “tan delta testing” without saying which asset class it means will be quoted wrongly.

4. What is an acceptable tan delta value?

This is the single most common search around this subject, and the honest answer is that no universal number exists. Four things move the threshold:

  • Insulation type. Polyethylene, filled EPR and oil-impregnated paper do not share a loss line. A value that is normal for PILC would be alarming in XLPE. IEEE 400.2-2024 handles this by tabulating assessment values by insulation type rather than fixing one global limit — precisely because a universal figure does not exist.
  • Temperature. Loss rises with temperature, and it rises at different rates for different materials. A reading taken at 15 °C and one taken at 35 °C on the same cable are not comparable unless corrected.
  • Geometry and length. Tan delta is a bulk average over the whole insulation volume. A long run with many joints and a short factory length are not the same sample.
  • History. The most reliable threshold is the asset’s own previous readings. A cable that has sat at 0.3% for a decade is behaving; a cable that moved from 0.15% to 0.3% in two years is telling you something the absolute value hides.

So the defensible way to judge a reading is in this order: compare against the same asset’s trend, then against sibling assets of the same type and age on the same network, and only then against the standard’s table for that insulation type. A number quoted alone, without any of those three references, is not an assessment.

5. Tip-up and stability: the part the number hides

Because tan delta is a bulk average, the most informative thing about it is often how it changes. Two derived quantities carry most of the diagnostic weight:

Tip-up. Measure tan delta at two or more voltage levels and look at the increase. Loss that grows with voltage indicates discharges or non-linear conduction — a mechanism that is invisible in a single reading. A flat tip-up suggests the loss is linear and probably moisture- or contamination-driven rather than discharge-driven.

Stability over the hold. During a monitored withstand, loss should be roughly steady. A reading that drifts upward as the hold proceeds is a warning sign that the insulation is being degraded by the test itself, and it is one of the criteria a monitored withstand is designed to catch.

Both of these are structural arguments for measuring tan delta during a withstand rather than as a standalone spot reading: the voltage steps and the time axis are already there, and the derived quantities come free.

6. Running it in the field

Three practical points decide whether a field tan delta number is usable:

  • Disconnect everything else. Tan delta is sensitive to parallel paths. Surge arresters, voltage transformers, capacitors and terminations left connected will dominate a cable reading. Isolate and, where the instrument supports it, guard.
  • Record the conditions. Temperature, humidity, test frequency and voltage level belong in the record alongside the number. Without them the reading cannot be compared to anything later.
  • Test at more than one level. A single point gives you a number; two or three points give you a tip-up, and a series of readings gives you a trend. The incremental cost is minutes.

For cable specifically, the field procedure and the waveform constraint are covered in detail in tan delta testing at VLF — including why the waveform has to be sine, since a cosine-rectangular withstand source cannot support a loss measurement at all.

7. Where tan delta fits in a 2026 programme

Two documents changed the standing of this measurement recently. IEEE 400.2-2024 moved the criteria for a monitored withstand onto tan delta and partial discharge, and added assessment tables by insulation type. NFPA 70B-2023 made electrical maintenance an enforceable standard and requires the maintenance programme to select insulation-condition test methods from a list that names tan delta explicitly, and to document the reason for the choice. Neither document mandates a tan delta test on every asset. What changed is that choosing — or not choosing — it is now a decision you have to be able to defend.

For a commissioning job on new cable, tan delta adds little: the question is workmanship, and a withstand test answers it. For service-aged cable in a replacement programme, it is the measurement that ranks the fleet so the capital budget goes to the right circuit first. For transformers and bushings it belongs in the routine dielectric-loss schedule on its own instrument. Three different jobs, three different scopes, one name.

Frequently asked questions

What is a good tan delta value for MV cable? There is no single figure. It depends on insulation type, temperature and the cable’s own history. IEEE 400.2-2024 tabulates assessment values by insulation type for exactly this reason. Judge against the trend first, then against comparable assets, then against the table.

Can I measure tan delta with an ordinary VLF hipot? Not meaningfully. Tan delta requires measuring the phase relationship between voltage and current, which is a different measurement chain from a voltage source with a current display. Sets without the function can apply the stress; they cannot produce the number. That is why integrated tan delta models exist as separate products rather than as a firmware option.

What is tip-up? The increase in tan delta as test voltage rises. A flat tip-up indicates linear loss (typically moisture or contamination); a rising tip-up indicates discharge activity or non-linear conduction. It needs at least two voltage levels to compute, which is why multi-step testing is worth the extra few minutes.

Does tan delta replace a partial discharge test? No. They have opposite blind spots. Tan delta averages the whole insulation volume and is blind to a single localised defect; partial discharge finds localised defects but says little about distributed ageing. Partial discharge testing covers the other half.

Is a tan delta test destructive? The measurement itself is passive. The stress applied to obtain it may not be: on cable it is normally taken during a VLF withstand, so the risk profile is that of the withstand. On transformers and bushings the applied voltage is below operating stress and the test is effectively non-destructive.

VA-TEK builds VLF withstand sets and integrated VLF-plus-tan-delta systems for 5–46 kV shielded cable, and the TDC-3100 for apparatus dissipation factor. For a test-voltage recommendation against a specific cable schedule, request a quote or contact the engineering team. For the wider method, start with the VLF cable testing guide and the three-test comparison.

Sources: IEEE Std 400.2-2024; NFPA 70B-2023; IEC 60502-2:2014+AMD1:2024; CIGRE TB 728. Test voltages and assessment values must be confirmed against the current official standard before a live job. Tan delta thresholds vary by insulation type and are set by the tabulated values in the applicable standard plus the operator’s own historical data, not by a universal figure.

Stay Updated with VA-TEK Insights

Get technical updates on VLF testing, relay protection and transformer diagnostics — new products, application notes and test guides. No spam, unsubscribe anytime.

VA-TEK Technical Updates
💬