A protective relay never sees the power system directly. It sees only what its current transformer (CT) gives it. A CT looks like the simplest device in the substation, just a winding on a core, yet a saturated CT can make an overcurrent relay slow, a distance relay hesitate and a differential relay trip for a fault that is not in its zone. This guide explains how CTs behave under fault conditions, why they saturate, how to check a CT with one simple formula, and how to select and test CTs for protection.
1. How a CT works
A protection CT is usually a toroidal core with a secondary winding of N turns. The primary is the power conductor itself passing through the core (one turn). In an ideal CT the secondary current is simply:
A real CT needs some current to magnetise its own core. That exciting current is taken from the secondary current, so the relay gets slightly less than the ideal value. As long as the exciting current is tiny, the error is negligible. The trouble starts when it is not.
Fig. 1: CT equivalent circuit. The magnetising branch (Xm) steals current from the relay when the core is pushed too hard.
The key idea is that the CT must produce a voltage to push the secondary current through its own winding resistance, the leads and the relay:
That voltage is created by changing flux in the core. A core can only carry so much flux. Ask for more voltage than the core can produce, for long enough, and the core saturates.
2. The excitation curve and knee point
Every CT has an excitation curve: secondary voltage plotted against exciting current, on log-log axes. It is the CT's fingerprint and is measured in the excitation (knee point) test.
Fig. 2: Typical excitation curve. Below the knee, exciting current is small and the CT is accurate. Above it, a small rise in voltage needs a large rise in exciting current.
- Linear region: the CT reproduces the primary current faithfully.
- Knee point: IEC defines it as the voltage at which a 10% increase in voltage causes a 50% increase in exciting current. ANSI uses the point where the curve's tangent is at 45° (for non-gapped cores).
- Saturation region: the exciting current shoots up and the relay current collapses.
3. Two kinds of saturation
a) Symmetrical (AC) saturation
The fault current is simply too large for the CT ratio and burden, so even a pure sine wave needs more voltage than the knee. This is a design mistake and is usually avoidable.
b) Transient (DC offset) saturation
Most faults start with a DC offset: an exponentially decaying component added to the AC current, whose size depends on the instant the fault starts and whose decay depends on the circuit's X/R ratio. The DC component drives flux steadily in one direction, so the core needs far more flux than for the AC alone. In the worst case the flux needed is larger by a factor of about (1 + X/R).
Fig. 3: A CT saturating under a fully offset fault current (illustrative simulation). The output follows the primary at first, then collapses for part of every cycle until the offset decays.
Three things are worth noticing in Fig. 3:
- The CT is accurate for the first part of the fault. Saturation takes time to build up.
- Saturation affects mainly the half-cycles in the direction of the offset.
- The output recovers as the DC offset decays. The higher the X/R, the slower the recovery.
Remanence makes it worse. After a fault is interrupted, or after a DC winding-resistance test, flux can remain trapped in the core. A gapless core can hold a large fraction of its saturation flux this way. If the next fault pushes in the same direction, the CT saturates much sooner. Low-remanence cores (for example IEC class PR or TPY) are used where this matters.
4. Reading CT ratings
| Marking | Standard | What it means |
|---|---|---|
| C400 | IEEE C57.13 | The CT can deliver 20 × rated current (100 A for a 5 A CT) into the standard burden with no more than 10% ratio error. 400 V is the terminal voltage at that point. C100, C200, C400 and C800 correspond to standard burdens of 1, 2, 4 and 8 Ω. |
| 5P20, 15 VA | IEC 61869-2 | Composite error no more than 5% up to 20 × rated current (the accuracy limit factor, ALF), with rated burden 15 VA connected. 10P20 allows 10%. |
| Class PX | IEC 61869-2 | Defined by knee-point voltage Ek, exciting current at Ek and winding resistance RCT. Used for differential and high-impedance schemes where the designer needs the actual numbers. |
| TPX / TPY / TPZ | IEC 61869-2 | Transient classes, specified for performance with DC offset. TPY and TPZ have air-gapped cores with low remanence. |
An IEC 5P CT's voltage capability can be estimated as its secondary limiting emf:
5. The criterion to avoid saturation
For C-class CTs, the condition for avoiding saturation even with full DC offset (and no remanence) fits in one line. It appears in the IEEE CT application guide (C37.110) and in Stanley Zocholl's book Analyzing and Applying Current Transformers:
- If = maximum fault current in per unit of the CT rated current
- Zb = total secondary burden (CT winding + leads + relay) in per unit of the CT's standard burden
- X/R = X/R ratio of the primary fault circuit
The "20" comes from the rating itself: a C-class CT is rated for 20 times rated current into its standard burden. Without the (1 + X/R) term, the formula checks symmetrical saturation only; with it, it checks the worst-case DC offset.
For IEC CTs the same idea is written in volts:
With full offset: (1 + X/R) × If,sec × (RCT + Rb) ≤ Ek or E(al)
6. Worked examples
Example 1: a feeder where saturation can be avoided
11 kV feeder, maximum fault 8 kA, X/R = 5. CT 800/5 C400 (standard burden 4 Ω). Total burden 0.8 Ω (CT winding 0.4 Ω + leads 0.3 Ω + numerical relay 0.1 Ω).
(1 + 5) × 10 × 0.2 = 12 ≤ 20 ✓
No saturation, even for a fully offset fault.
Example 2: a power station switchboard where it cannot
11 kV switchboard close to generators, maximum fault 25 kA, X/R = 14. CT 1200/5 C400, total burden 1.5 Ω (CT winding 0.6 + leads 0.6 + relay 0.3).
Symmetrical: 20.8 × 0.375 = 7.8 ≤ 20 ✓
With offset: (1 + 14) × 7.8 = 117 ≫ 20 ✗
Maximum burden for no saturation = 20 / (15 × 20.8) × 4 Ω = 0.26 Ω
0.26 Ω is less than the CT's own winding resistance, so transient saturation cannot be avoided here by any practical choice of leads or relay. This is common near generators, where fault current and X/R are both high. The engineer then accepts that the CT will saturate during the offset and makes sure the relays are designed to cope: high-set elements that act in the first unsaturated part of the cycle, CT saturation detection in differential relays, and suitable slope settings.
Example 3: an IEC CT, 1 A versus 5 A
33 kV feeder, fault 12 kA, X/R = 10, CT ratio 400. Leads: 100 m each way of 2.5 mm² copper, so a 200 m loop of about 1.48 Ω.
| 400/1, 5P20, 10 VA, RCT 2 Ω | 400/5, 5P20, 15 VA, RCT 0.3 Ω | |
|---|---|---|
| CT capability E(al) | 20 × 1 × (2 + 10) = 240 V | 20 × 5 × (0.3 + 0.6) = 90 V |
| Secondary fault current | 30 A | 150 A |
| Total loop resistance | 2 + 1.48 + 0.1 = 3.58 Ω | 0.3 + 1.48 + 0.004 ≈ 1.78 Ω |
| Voltage needed, symmetrical | 30 × 3.58 = 107 V ✓ | 150 × 1.78 = 268 V ✗ |
| Voltage needed, full offset (× 11) | ≈ 1180 V (transient saturation) | ≈ 2950 V |
With the same leads, the 5 A CT saturates even on a pure sine wave, while the 1 A CT handles the symmetrical fault with margin. For a time-overcurrent relay that is fine: the offset decays within a few cycles, long before an inverse-time element trips. For a fast differential relay, the transient case still needs attention.
7. Calculating burden, and why 1 A wins
The total burden is the sum of three parts:
- CT winding resistance: grows with the number of turns, so higher ratios have higher RCT. Take it from the test report.
- Lead resistance: for phase-to-earth faults the current goes out on the phase wire and back on the neutral, so use the full loop (2 × one-way length). For balanced three-phase faults with star-connected CTs, the return currents cancel, so the one-way length is enough. Design for the worse case.
- Relay burden: a numerical relay takes a fraction of a VA. Old electromechanical relays could take several ohms.
Burden in VA rises with the square of the current. The same 1.48 Ω of leads costs 1.48 VA at 1 A but 37 VA at 5 A. That is why long lead runs in large substations and power plants use 1 A secondaries.
Practical rule: keep total lead resistance low (a common target is around 0.5 Ω or less for 5 A circuits) by choosing a larger cable size, and prefer 1 A CTs where leads are long.
8. What saturation does to relays
| Relay | Effect of CT saturation | Usual countermeasure |
|---|---|---|
| Time overcurrent (51) | Measured current is lower, so tripping is delayed; coordination margins shrink | Avoid symmetrical saturation; check grading at maximum fault level |
| Instantaneous overcurrent (50) | With a very low ratio CT on a high fault level, the relay may see too little fundamental current and fail to pick up | Do not use small ratio CTs on high fault levels; relays with peak or adaptive filtering |
| Differential (87T, 87G, 87B) | On an external fault, one CT saturates and the other does not, creating a false differential current and a possible maloperation | Percentage (dual-slope) restraint, CT saturation detection, matched class PX CTs, high-impedance bus schemes with a stabilising resistor |
| Distance (21) | Reduced current makes the apparent impedance larger: possible delay or underreach | Adequate CT rating; modern relays tolerate moderate saturation well |
Differential protection is the most sensitive case, because a relay that trips for an external fault takes out healthy equipment. This is why differential schemes are where CT selection deserves the most care.
9. CT selection checklist
10. Testing and safety
Commissioning tests
- Ratio and polarity: wrong polarity in a differential scheme guarantees a trip on load or on the first external fault.
- Excitation (knee point) test: confirms the curve matches the specification and reveals shorted turns.
- Winding resistance: needed for burden calculations.
- Insulation resistance and secondary loop burden measurement.
- Demagnetise after DC tests such as winding resistance, which can leave remanent flux in the core.
Never open-circuit a loaded CT secondary. With no burden to drive current through, all the primary ampere-turns magnetise the core and the secondary voltage can reach dangerous levels, enough to injure people and damage insulation. Always short the secondary (using test blocks or shorting links) before disconnecting a relay or meter.
11. Test yourself
1. What does C400 mean on a 5 A CT?
It can deliver 20 × 5 = 100 A into its 4 Ω standard burden with no more than 10% ratio error; the terminal voltage at that point is 400 V.
2. Why does DC offset cause saturation even when the AC current is within rating?
The decaying DC component drives the core flux in one direction. In the worst case the flux required is about (1 + X/R) times that for the AC current alone.
3. A CT circuit has 1.5 Ω of leads. What is the lead burden at 5 A and at 1 A?
5² × 1.5 = 37.5 VA at 5 A; 1² × 1.5 = 1.5 VA at 1 A.
4. For a phase-to-earth fault, should you use the one-way or loop lead length?
The loop length (twice the one-way length), because the current returns through the neutral wire.
5. Which relay is most at risk from CT saturation on an external fault?
A differential relay, because unequal saturation of its CTs creates a false differential current.
6. Why should a CT be demagnetised after a winding resistance test?
The DC test current can leave remanent flux in the core, which reduces the margin before saturation in the next fault.
Key takeaways
- The CT must generate a voltage of Isec × (RCT + leads + relay). Saturation happens when the core cannot supply it.
- DC offset raises the requirement by up to (1 + X/R); remanence reduces the margin further.
- Quick check for C-class CTs: (1 + X/R) × If × Zb ≤ 20.
- Always avoid symmetrical saturation. Transient saturation is sometimes unavoidable near generators, so the relays must be chosen and set to cope.
- Lead burden rises with current squared: use 1 A CTs for long runs.
- Never open-circuit a loaded CT secondary.
Have you seen a differential relay trip for an external fault because of CT saturation? Share what you found in the comments.
Further reading: Stanley E. Zocholl, Analyzing and Applying Current Transformers, Schweitzer Engineering Laboratories (2004), the source of the saturation criterion discussed here; IEEE C37.110, Guide for the Application of Current Transformers Used for Protective Relaying Purposes; IEEE C57.13; IEC 61869-2. The worked examples in this post use illustrative values; always use your own fault study, CT test reports and the relay manufacturer's CT requirements.




0 Comments