VLF Cable Testing: A Practical Guide to IEEE 400.2-2024 Field Testing

Very low frequency (VLF) testing is the default way to prove a shielded power cable system after installation or in service. It replaced DC hipot for extruded cable because DC leaves space charge in the insulation, and it replaced 50/60 Hz AC on site because a mains-frequency set large enough to charge kilometres of cable does not fit on a van. VLF sits in the gap: an alternating voltage, applied slowly enough that the current a portable set must deliver drops by two and a half orders of magnitude.

This guide collects what a test engineer actually needs to run a defensible VLF test to the current standard, IEEE 400.2-2024: the voltage to apply, how long to hold it, which waveform, when to add tanδ or partial discharge, and where the method is honestly weak. It is written for the engineer writing the procedure and the engineer reviewing the record — not for the marketing page.

1. Why 0.1 Hz works: the physics behind the method

A cable is a capacitor. The current needed to raise it to test voltage is set by the standard capacitive relation:

I = 2πf · C · V

where f is frequency, C is the cable capacitance and V is the test voltage. Because frequency sits directly in the numerator, dropping the test frequency from 50 Hz to 0.1 Hz cuts the current the set must supply by a factor of 500 (600 at 60 Hz) at the same voltage. A circuit that would draw tens of amperes at mains frequency draws milli-amperes at 0.1 Hz, and a source that would fill a shipping container becomes one that a two-person crew carries up a manhole.

The voltage is still alternating, so the insulation sees a reversing field rather than the unidirectional field of DC. That polarity reversal is what makes the test meaningful for extruded insulation: it keeps charge from accumulating the way it does under DC.

IEEE 400.2 defines VLF as any alternating test below 1 Hz. Commercially, 0.1 Hz is the standard choice; 0.05, 0.02 and 0.01 Hz are used to reach longer cables with the same set. The relationship above is also the reason for those lower settings: halving the frequency lets the set hold the same voltage on roughly twice the capacitance, which is twice the cable length.

2. The standards map: which document governs your test

Four documents cover most work, and the first question is always which one applies to the cable in front of you.

The 2024 revision of IEEE 400.2 is the one to work from now. It replaced the 2013 edition and made several changes that affect the record you keep: it unifies all test voltages on an rms basis (the older edition mixed peak and rms depending on waveform), it tightens the monitored-withstand criteria to tanδ and partial discharge, it raises the minimum HV withstand time from 30 to 60 minutes, and it adds an explicit requirement for a slow, controlled voltage rundown at the end of the test. If your procedure is a few years old, those four points are worth checking against the current text.

Where the cable sits at 30 kV or below and the spec is IEC, the after-laying test sequence walks through how the IEC options and the GB/T annex line up in practice.

3. Test voltage: reading IEEE 400.2-2024 Table 3

IEEE 400.2-2024 gives test voltage in a single table with three columns — installation, acceptance and maintenance — and the values are kV rms, phase-to-earth. Table 3 as published in the 2024 edition, for sine waveform:

Values are from IEEE Std 400.2-2024, Table 3 (sine waveform). Always confirm against the official standard before a job.

Three notes decide how the table is used:

• Voltage falls relative to rating as the class rises. The standard’s Note 2 records this deliberately, to offset the higher design stress in MV cables. A 15 kV cable is tested at a higher multiple of U₀ than a 66 kV cable.
• Maintenance is roughly 75% of acceptance. Note 3. If the value is not a whole number in the standard, take the table figure, not your own 75%.
• If the cable runs below its rated voltage, test at the actual operating class (Note 1). A 20 kV-rated cable energised at 11 kV is tested as an 11 kV cable.

The 66 kV trap

The number most often mis-read is 66 kV. Acceptance in Table 3 is 80 kV rms. Converted to peak (80 × √2) that is ≈ 113 kV peak; with the margin the standard’s own note implies for instrument output, the practical requirement is a set rated ≥ 115 kV peak, 120 kV class to be safe. A set that tops out at 90 kV peak — a very common ceiling in portable VLF equipment — can manage the maintenance level around 66 kV (60 kV rms) but cannot perform the acceptance test. Any quotation for “66 kV VLF acceptance” should be checked against the instrument’s peak rating before the cable is scheduled.

Peak or rms? Read the nameplate

Because Table 3 is now rms and many instruments are marked in peak, the two numbers in a bid are not always the same quantity. As a rule of thumb, a peak rating is the rms figure multiplied by about 1.41. VA-TEK’s VLF units state their output as peak; a comparison that places a peak-rated set next to an rms-rated set without converting will make the peak set look stronger than it is. Decide which basis the bid is on before comparing.

4. Hold time and frequency: 15, 30 or 60 minutes

IEEE 400.2-2024 sets the duration bands and they map to the purpose of the test:

The choice of frequency follows the cable length, not the cable class. With I = 2πfCV, a set has a fixed current ceiling; the way to hold voltage on a longer (higher-capacitance) cable is to lower the frequency. That is why most sets offer 0.1, 0.05, 0.02 and 0.01 Hz. Dropping from 0.1 Hz to 0.01 Hz multiplies the reachable capacitance by ten at the same voltage.

Two practical points. First, a longer test at a lower voltage is gentler than a shorter test at a higher one; the field data behind the standard favours it. Second, the hold time and the frequency are separate decisions — changing the frequency to reach a long cable does not license shortening the hold time. The duration-and-frequency article works through the trade-off with worked numbers.

5. Waveform: sine versus cosine-rectangular

Two waveforms are in field use:

• Sine VLF (true sine). The IEEE 400.2 default. The field distribution most closely resembles service, and it is the waveform required for tanδ and partial-discharge work.
• Cosine-rectangular (CR). A faster polarity reversal. It can drive more capacitive load for a given set and allows three-phase testing, but the standard gives its own voltage column: for CR, test voltage = sine value × √2 (Note 6). CR is a withstand waveform; it cannot support tanδ or PD diagnosis.

If the deliverable is a pass/fail record, either waveform is valid at its own table value. If the deliverable includes any diagnostic measurement, the waveform must be sine.

6. Adding diagnosis: tanδ and partial discharge

A withstand test answers one question — does the cable hold the voltage. It says nothing about a cable that passes while quietly ageing. The 2024 revision leans into this by formalising monitored withstand: run the hold while measuring tanδ and partial discharge, so one energisation yields both the proof and a condition record.

Tanδ (dissipation factor) is the workhorse for water-tree ageing. Water trees create conductive paths that raise loss, and at low frequency that loss is more visible, so 0.1 Hz tanδ responds to ageing that a mains-frequency measurement would smooth away. The reading is meaningful as a trend and against the standard’s stability criteria more than as a single absolute number. The tanδ article covers interpretation; a field-ready instrument such as the VLF-34TD or VLF-45TD combines the withstand source and the tanδ measurement in one unit for 5–15 kV and 5–25 kV cable respectively.

Partial discharge locates what tanδ cannot. A single void or a badly made joint is invisible to tanδ, which averages over the whole cable, but it is exactly what PD sees. CIGRE TB 728 recommends PD testing at ≥ 1.7 U₀ because most discharge inception voltages sit above 1.5 U₀ — test lower and you miss them. PD is added to a VLF set through external coupling (HFCT or acoustic), and the practical rule is to read the trend across a fleet rather than trust a single absolute pC figure.

7. VLF in context: why not DC, and where DAC fits

Why not DC. DC hipot was the standard field test for decades and is now avoided on extruded insulation above 5 kV in service. The mechanism is space charge: under DC the field distributes by conductivity, and injected charge is trapped in the polymer. On polarity reversal — at energisation, or in a wrong connection — that trapped charge adds to the applied stress and can approach twice the average field near the semicon. A seven-year EPRI study found maintenance-level DC on aged XLPE shortened remaining life, while factory-level DC on new cable did not. DC remains valid for laminated (paper/lead) insulation; the prohibition is specific to extruded insulation. The VLF versus DC article works through the comparison in detail.

Where DAC / oscillating wave fits. Damped AC (DAC, also called oscillating wave) charges the cable with DC and then discharges it through an inductor, producing a decaying oscillation typically between 20 and 500 Hz. Being closer to mains frequency, it gives a PD environment that some engineers prefer and it locates PD by reflection. Its trade-offs are honest: the frequency depends on cable capacitance and must be tuned by switching inductance, and a CIGRE field study found only weak agreement between the discharges seen under DAC and under sine VLF on the same cables. Both sine VLF and oscillating wave were brought into GB/T 3048.8-2025 as informative annexes, so neither is the “only” recognised method. The choice is a scope decision: VLF for a portable, standard-aligned withstand with tanδ available in the same box; DAC when PD characterisation closer to power frequency is the goal.

8. What the field evidence actually shows

VLF has a large field record, and it is worth knowing both what it supports and where it is thin.

• Proof tests do find real defects. A CIRED 2003 database of 17,435 VLF tests found hidden defects in about 12.5% of circuits, with roughly two-thirds of breakdowns appearing in the first 12 minutes and one-tenth only after 30 minutes — the practical case for the full hold time.
• VLF beats DC on outcomes. A DOE-funded NEETRAC programme measured on-test failure falling from 48% under DC to 8% under VLF, and later service faults falling by roughly 40% at 13 kV and 60% at 27 kV relative to DC-tested cable.
• Detected-defect rates are useful, not perfect. Utility programmes report roughly 15–29% of tested circuits flagged for attention. A flagship owner-side database (KEPCO, tens of thousands of systems) reports markedly lower fault rates on diagnosed cable than on untested cable.
• Longer at lower voltage wins. The same NEETRAC work found 30 min at a lower multiple of U₀ outperformed 15 min at a higher one.

Against that, the method has real limits that a credible procedure should state:

• A withstand test is a proof test. It trips on gross, concentrated defects. Distributed ageing — water trees, early electrical trees — will not necessarily break down, so a pass is not a clean bill of health.
• Tanδ is blind to isolated defects. It averages the whole length; the standard itself notes a very large population of water trees is needed before it responds.
• PD repeatability under VLF is modest. Inception readings vary with residual charge and test order; extinction values and trends are more stable than inception values.
• Claims that VLF causes cumulative damage are disputed. The most systematic evidence points to voltage magnitude rather than repetition as the driver of on-test failure, and the standard treats a correctly applied withstand as low-risk to sound insulation. Where a claim is associated with a competing test technology, read it with that in mind.

IEEE 400.2-2024’s move to monitored withstand is the standard’s own acknowledgement that the voltage alone does not tell the whole story.

9. Choosing a set by voltage class

Work the choice from the cable, in this order: voltage class → acceptance or maintenance → waveform → whether diagnosis is needed → then the current/load rating for the cable length. The table below maps acceptance voltage from Table 3 to the practical set rating.

The VLF hipot family covers 5 kV to 46 kV cable with one shared platform. The VLF-30-90 is the six-tap withstand unit spanning 30 to 90 kV peak for the 11–46 kV range; the VLF-34TD and VLF-45TD add tanδ, DC and sheath testing for diagnostic work on 5–15 kV and 5–25 kV systems. All are sine-wave, all read their rating as peak, and all four output frequencies from 0.1 to 0.01 Hz are standard. How to choose a VLF test set takes the same four questions further, including the load-rating reasoning that decides how long a cable one set can reach.

10. Frequently asked questions

What is the difference between 2U₀ and 3U₀ in VLF testing? These are shorthand for the multiple of the phase-to-earth voltage applied. Acceptance of new cable sits at the higher multiple, maintenance of aged cable at the lower. The controlling figures are the ones tabulated in IEEE 400.2-2024 Table 3, which express the same idea in kV rms and vary the multiple with the cable class — prefer the table to a remembered rule of thumb.

Can a 90 kV VLF set test a 66 kV cable? For maintenance at the reduced level, in many cases yes; for the acceptance test at 80 kV rms (≈ 113 kV peak), no. A 66 kV acceptance test needs a set rated around 115–120 kV peak. Confirm the instrument’s peak rating against the Table 3 figure rather than the cable’s nominal class.

How long should a VLF test run? 60 minutes at 0.1 Hz for new-cable acceptance and HV work; 30 minutes (15 minimum) for aged-cable maintenance. Test hold time is independent of the frequency used to reach a long cable.

Does VLF damage the cable? A correctly applied withstand at the standard’s voltage is regarded as low risk to sound insulation; the leading systematic evidence attributes on-test failures to voltage magnitude rather than to repeated applications. As with all high-voltage testing, follow the standard’s voltage, ramp and rundown requirements.

Can I do partial discharge with a CR waveform? No. Cosine-rectangular is a withstand waveform only. Tanδ and PD require sine VLF.

Putting it together

A defensible VLF programme is not a single setting. It is the right voltage from the class, the right hold time for the purpose, a sine waveform if anything is being measured, and — where ageing matters — tanδ or PD alongside the withstand so the record shows condition and not just pass/fail. The 2024 edition of IEEE 400.2 formalised the last point, and it is the direction the field is moving.

VA-TEK builds VLF withstand and diagnostic sets for 5–46 kV shielded cable. For a test-voltage recommendation on your cable, a load-rating check for a specific run, or the current datasheets, request a quote or contact the engineering team. Datasheets and application notes are collected in Downloads, and Technical Insights carries the deeper article series on VLF voltage, duration, tanδ and the DC comparison.

Standards cited: IEEE Std 400.2-2024; IEEE Std 400.4-2015; IEC 60502-2:2014; IEC 60229:2007; IEC 60060-3:2006; GB/T 3048.8-2025; CIGRE TB 728. Field data: CIRED 2003 (S.C. Moh); NEETRAC / CDFI programme funded by the U.S. Department of Energy. Test voltages are reproduced from IEEE 400.2-2024 Table 3 for reference; always work from the current official standard for a live job.