Instrument Transformer Testing: CTs, VTs and CVTs

Why a certified accuracy class is necessary and badly insufficient for protection.

Every revenue meter and every protective relay in a power system rests on one assumption that almost nobody questions: that the current and voltage arriving at the terminals are faithful scaled replicas of the primary quantities. Instrument transformers are the guardians of that assumption. A metering CT with 0.5% error on a 500 MW line represents 2.5 MW measured wrong in one direction or the other, every hour of every year, at a single metering point. A protection CT that saturates during an asymmetric fault hands the relay a distorted, understated current: the relay under-reaches, delays, or simply does not operate, and a fault that should have been cleared in 80 ms becomes the loss of a bus.

The central paradox of this equipment family is that metering and protection are opposite designs. A metering core is supposed to saturate early: above roughly 120% of rated current it should give up reproducing the primary and protect the instruments behind it from fault current. A protection core must not saturate at all inside its accuracy limit factor: at twenty times rated current it must still reproduce the waveform with a bounded composite error. The same unit frequently houses both cores around the same primary conductor, and the laboratory has to verify each against completely different requirements. This module covers what makes instrument transformers unique among everything else on the test floor — fine metrology expressed in tenths of a percent and minutes of arc, and transient behaviour that decides whether protection sees the fault in the first cycle.

What you will be able to do

• Place each part of the IEC 61869 series in its scope, including the 2023 structural change in which IEC 61869-1 Ed. 2.0 absorbed the general LPIT requirements previously held in IEC 61869-6:2016.

• Interpret metering accuracy classes (0.2, 0.2S, 0.5, 0.5S, 1, 3, 5) and protection classes (5P, 10P, PR, PX, TPX/TPY/TPZ) with the numerical ratio error, phase displacement and composite error limits at each test point.

• Explain why accuracy only exists relative to a stated secondary burden, and calculate the effect of burden on error.

• Run and interpret a CT excitation curve, locate the knee point by both the IEEE C57.13 45° tangent criterion and the IEC criterion (+10% V produces +50% I), and use the curve to screen for shorted turns.

• Size a protection CT for transient duty: DC offset, transient dimensioning factor Ktd, remanence, and the choice between closed cores and gapped cores.

• Describe the CVT circuit, the physical origin of ferroresonance and of slow transient response, and the tests that verify both.

• Build the full routine, type and special test plan for CTs, VTs and CVTs per IEC 61869-2/-3/-5, including partial discharge, withstand, temperature rise and short-time current limits.

• Recognise the low-power instrument transformer frontier — Rogowski coils, dividers and IEC 61850-9-2 sampled-value output — and what changes in the test bay.

Accuracy exists only relative to a burden

A real instrument transformer commits two errors at once. The ratio error ε is the percentage deviation in magnitude, ε = (Kn·Xs − Xp)/Xp × 100%, where Kn is the rated ratio. The phase displacement Δφ is the angular deviation between the reversed secondary phasor and the primary, expressed in minutes of arc or centiradians. For energy metering the angle matters as much as the magnitude: in a system running at 0.5 power factor, a 10-minute phase error in the CT introduces roughly 0.5% error in measured active power, which is the size of the class limit itself.

Burden is the load connected to the secondary, expressed in VA at rated current or voltage with a stated power factor. CT error grows with burden, because more load demands more induced internal voltage, more flux, and more excitation current. That is why the standard does not certify a single error value but an error inside a burden window — 25% to 100% of rated burden for metering classes, with specific rules for very low burdens where IEC 61869-2 uses a range from 1 VA up to the rated value. The practical consequence bites immediately: a class 0.2 CT with a 15 VA rated burden feeding a 0.5 VA electronic meter is outside its certified window. Its real error is unknown and possibly larger than its class. Burden over-specification, inherited from the era of electromechanical instruments, is now one of the most common causes of out-of-class revenue metering.

The excitation curve and two incompatible knee points

The excitation curve — RMS voltage applied to the secondary with the primary open, plotted against excitation current — is the magnetic fingerprint of a CT. It gives the knee point, the margin against saturation, and the health of the winding. It is also where two standards schools part company. IEC defines the knee as the point where a 10% increase in voltage produces a 50% increase in excitation current. IEEE C57.13 uses the 45° tangent on the log-log curve, or 30° for gapped cores. For the same core, the IEEE value typically comes out 20 to 25% lower than the IEC value. Confusing the two definitions when specifying or comparing CTs is a classic engineering error, and it cannot be fixed by applying a conversion factor: if the relay data sheet asks for an IEEE knee point, the curve has to be read by the IEEE criterion.

As a screening test the excitation curve is unbeatable for detecting shorted turns. A single shorted secondary turn collapses the curve — the knee point voltage typically drops to a fraction of its original value with dramatically increased excitation current. The measurement is always compared against the factory curve or against sister phases, and a divergence of more than a few percent in the knee region is an immediate alarm.

The transient: DC offset, Ktd and remanence

A real fault is rarely symmetrical. Depending on the point on wave at fault inception, the current carries a DC component that decays with the primary time constant Tp = Lp/Rp, typically 20 to 150 ms on transmission systems. The core integrates that DC component as a growing unidirectional flux, and the transient flux can reach 10 to 30 times the steady-state value. A CT sized only by its accuracy limit factor in symmetrical current saturates in the first cycles, exactly when the relay needs it most.

Transient sizing uses the transient dimensioning factor Ktd, which multiplies the required secondary voltage and is evaluated at the time to which fidelity is required — relay operating time, or the whole open-close-open reclosing sequence. For Tp = 100 ms and fidelity to 40 ms, Ktd exceeds 15, which is why transient cores are physically large. IEC 61869-2 answers the problem with three philosophies: TPX with a closed core and no remanence limit, TPY with small distributed gaps and Kr ≤ 10%, and TPZ with large gaps and a linearised core that reproduces only the AC component.

Remanence deserves its own paragraph because it is the invisible villain. After a fault with DC offset is interrupted, a closed core — TPX, or an ordinary 5P/10P — can retain 60 to 80% of saturation flux. Normal load flux swings around the retained level without removing it, so remanence does not decay in service. If a reclose onto a permanent fault happens seconds later, or a new fault happens weeks later, the core starts one magnetic storey up and saturates within a few milliseconds. Protection misoperation traceable to CT saturation and remanence appears consistently in utility misoperation statistics, and the engineering answer was written into the standard: PR and TPY classes with Kr ≤ 10% wherever automatic reclosing exists.

The CVT is a resonant circuit by construction

Above roughly 145 kV a purely inductive VT becomes expensive and bulky. The capacitor voltage transformer divides the system voltage across a capacitive stack (C1 in series, C2 at the tap, giving an intermediate voltage typically between 5 and 20 kV), compensates the divider impedance with a series tuning reactor that resonates with C1 + C2 at rated frequency, and delivers the result to a conventional inductive intermediate VT. The same capacitor stack doubles as a power-line-carrier coupling capacitor, which is the historical reason CVTs are everywhere on transmission lines.

An LC circuit feeding a saturable ferromagnetic core is the textbook recipe for ferroresonance. After a transient — energisation, load rejection, a removed secondary short — the nonlinear inductance of the intermediate VT can lock with the divider capacitance into a sustained subharmonic (typically one third of fundamental) or fundamental oscillation, with destructive overvoltage and overheating. This is not a manufacturing defect; it is an inherent property of the topology, which is why the standard requires a ferroresonance suppression device and a test that deliberately provokes the phenomenon. IEC 61869-5 applies and removes a secondary short circuit at unfavourable instants, at several voltages up to Fv·Un, and requires the oscillation to die away with the secondary voltage back within ±10% of steady state inside the specified suppression time, on the order of 500 ms, in every repetition.

Transient response is the other CVT-specific test. When a fault collapses the primary voltage to zero, energy stored in C1, C2 and the reactor keeps feeding the secondary for tens of milliseconds — the CVT lies to the distance relay at precisely the moment of the zone 1 decision. IEC 61869-5 disciplines this with transient response classes T1, T2 and T3, requiring the residual secondary voltage to fall below decreasing percentages at standardised instants from 10 ms to 90 ms after a solid primary short. The result depends heavily on the suppression device, and aggressive ferroresonance damping improves the transient at the cost of loading the divider and affecting accuracy. CVT design is that triangle of compromises.

The LPIT frontier: the iron problems disappear, new ones arrive

Low-power instrument transformers replace the iron core with low-power sensors: air-core Rogowski coils whose di/dt output is electronically integrated, split-core CTs with conditioning electronics, resistive and capacitive voltage dividers, and optical sensors. The output stops being 5 A or 115 V and becomes a low-voltage signal — 22.5 mV, 200 mV, 4 V — or a digital stream. IEC 61869-9 standardises the sampled-value interface over the IEC 61850-9-2 profile at 4 000 or 4 800 samples per second for protection and metering, 14.4 kHz for power quality, with PTP time synchronisation to IEEE 1588.

Syllabus

• Standards map: IEC 61869-1 (2023), -2, -3, -5, -9, and what the 2023 edition consolidated

• Ratio error and phase displacement: definitions, the excitation current as the physical cause

• Burden as a boundary condition: the 25–100% certified window and the electronic-meter trap

• Metering accuracy classes and the S classes down to 1% of rated current

• Inductive VT classes, the voltage factor Fv, and the 1.9 × Un for 8 h case

• Protection classes 5P/10P, accuracy limit factor, composite error, PR and PX

• The excitation curve: IEC and IEEE knee point definitions and why they are not interchangeable

• DC offset, transient dimensioning factor Ktd, remanence, and classes TPX/TPY/TPZ

• Routine tests: accuracy against a reference standard, applied voltage between sections, polarity, partial discharge, excitation curve

• Type tests: short-time thermal and dynamic current, temperature rise, lightning and switching impulse, RIV, extended accuracy

• The CVT: capacitive divider, tuning reactor, ferroresonance suppression, transient response classes T1/T2/T3

• Special and diagnostic tests: full saturation characteristic with remanence measurement, PRPD analysis, accelerated ageing

• Low-power instrument transformers: Rogowski coils, split cores, resistive and capacitive dividers, digital output

Laboratory work

Three bench exercises. First, excitation curve and knee point on a real CT: with a variable AC source and the primary open, sweep V against I from 10% up past the knee, locate Ek by both the IEC and IEEE criteria and compare the two numbers, then deliberately magnetise the core with a DC pulse, observe the shift, and demagnetise correctly with a decaying AC field. Second, ratio error against a reference: measure a class 0.5 CT against a 0.05 standard on a comparison bridge at 25% and 100% of rated burden, plot the points inside the class parallelogram, then swap the burden for a 0.5 VA electronic meter and discuss why the result leaves the certified window. Third, routine partial discharge on a cast-resin CT: calibrate the circuit with a 5 pC injection, run the IEC 61869-1 voltage profile, and compare a healthy unit (below 5 pC) against a unit with a known defect, reading the internal-cavity PRPD pattern. Directed study: for a 230 kV bay with 31.5 kA fault current, Tp = 80 ms and an open-close-open cycle in 300 ms, fully specify the protection CT — class, accuracy limit factor or transient class, and remanence limit — and justify every choice.

Key takeaway

An accuracy certificate describes symmetrical steady state on a virgin core. It says nothing about what a core with 70% remanence will do on the next asymmetric event. Specify the burden window you will actually operate in, read the knee point by the criterion your relay asks for, and treat remanence as a design parameter rather than a footnote.

Module 13 of the Atlas Energy Academy · Equipment Families · about 4 hours · Video series (8 × 10–15 min) plus bench practice · Prerequisites: Module 12 — Power Transformers