CT Burden Resistor Calculator

Size the burden resistor on a split-core or toroidal current transformer so that full-scale primary current just fills your ADC’s input swing — with the nearest standard value and what substituting it costs, the burden’s own dissipation, and a saturation check against the CT’s knee, because too large a burden makes the reading non-linear and low.

Last updated 24 Sep 2026

Burden resistor, standard value and saturation check

CT ratio + ADC swing → burden

How is the CT’s ratio given?

Full-scale primary current you want to measure (A rms)The largest current the meter has to read without clipping, not the average load. Size for the breaker, not for the kettle.

CT turns ratio, N (secondary turns per primary turn)2000 for a YHDC SCT-013-000 (100 A : 50 mA), 1000 for a Nuvotem Talema AC-1030. A CT rated as a current ratio is telling you its turns ratio: 100 A : 50 mA is 2000:1. Use the figures from your part’s datasheet; typical values vary widely between manufacturers.

CT rated primary current (A)Only used when the ratio is entered as a pair of currents.

CT rated secondary current (mA)The other half of that pair. 50 mA for a 100 A : 50 mA CT; 5 A or 1 A for a switchgear protection CT.

Is the signal biased to mid-rail?A CT’s output is alternating, so a single-supply ADC needs it lifted to the middle of its range. That leaves only half the reference for each half cycle.

ADC reference voltage (V)3.3 V for an ESP32 or a 3.3 V Arduino, 5 V for a classic Uno, 1.1 V if you have selected the internal reference. It is the reference that matters, not the supply.

Fraction of the available swing used at full scale (%)100% fills the ADC exactly at full-scale current and leaves nothing for surge or a mis-estimated maximum. 80–90% is a kinder design.

CT secondary winding resistance (Ω)48 Ω for a Nuvotem Talema AC-1030, 41.8 Ω for an AC-1020, 10 Ω for a Talema AS-103. It adds to the burden as far as the core is concerned, so it counts in the saturation check even though it is not part of your measurement. Use the figures from your part’s datasheet; typical values vary widely between manufacturers.

CT knee-point EMF, or its maximum rated secondary voltage (V rms)Above this the core saturates and the reading goes non-linear and LOW. Protection CTs publish a knee-point EMF; measurement CTs usually publish a rated burden instead, and 7.5 V here is the Talema AC-1030’s own figures multiplied out — 75 A maximum through 1000:1 into its recommended 100 Ω termination. Use the figures from your part’s datasheet; typical values vary widely between manufacturers.

CT secondary magnetising inductance (H) — 0 if you do not know itThis is what decides the ratio and phase error, and most energy-monitoring CTs do not publish it; a Talema AS-103 quotes 200 mH. Leave it at 0 and those two rows are simply left out. An LCR meter across the secondary with nothing through the core will tell you.

Supply frequency (Hz)50 or 60. Only used for the ratio and phase error, which get worse as the frequency falls.

ADC resolution (bits)10 for an Arduino Uno, 12 for an ESP32, 16 or 24 for a dedicated energy metering chip.

Standard value seriesA burden resistor should be 1% or better: its tolerance goes straight into the calibration.

The live conductor passes once through the core and is the whole of the primary; the secondary winding drives the burden resistor, and the voltage across it is what the ADC reads. The divider lifts one end of the burden to mid-rail so that an alternating signal fits inside a single-supply input, and the capacitor keeps that point stiff. The moving dots are the primary and secondary currents at full scale. The burden turns amber as its peak voltage approaches the available swing and red once it exceeds it; the transformer turns amber as the secondary EMF approaches the core's knee.

38.89ΩExample

a 1000:1 CT reading up to 30 A into a 3.3 V ADC biased to mid-rail, with 48 Ω of secondary winding resistance

One line, and the two things it assumes

I_sec = I_pri ÷ N
R_burden = V_{peak available} ÷ (√2 · I_pri,rms ÷ N)
= AREF · N ÷ (2√2 · I_max) for a mid-rail bias
E_sec = I_sec · (R_burden + R_winding) must stay below the knee
ratio error = 1 ÷ √(1 + ((R_b+R_s) ÷ ωL_m)²) − 1

N: the turns ratio. A CT sold as 100 A : 50 mA has 2000 turns on its secondary against the one turn of cable you clamp
√2: because the ADC sees peaks and the specification is in rms. Forgetting it is the classic factor-of-1.414 error
E sec: the EMF the core has to generate to push the secondary current through everything in the loop — the burden AND the winding’s own resistance
L m: the magnetising inductance. The current that goes into it is the current that does not reach your burden, which is the whole of the ratio and phase error

Worked example

a 1000:1 CT reading up to 30 A into a 3.3 V ADC biased to mid-rail, with 48 Ω of secondary winding resistance
At full scale the secondary carries 30 A ÷ 1000 = 30 mA rms, whose peak is 42.43 mA
A mid-rail bias leaves half the reference for each half cycle: 3.3 ÷ 2 = 1.65 V of peak swing
So the burden is 1.65 V ÷ 42.43 mA = 38.89 Ω, and the nearest E24 value is 39 Ω — 0.28% high, which brings full scale down to 29.9 A
That burden dissipates 35.1 mW and develops 1.17 V rms; with the winding's 48 Ω the core has to generate 2.61 V, which is 2.87× below the CT's 7.5 V limit
On a 12-bit ADC one step is 805.7 µV, so the meter resolves about 14.61 mA of primary current — before any noise, and before the fact that an rms calculation over a cycle averages many samples and does better than one step

Real current transformers and what they imply

A CT sold by a pair of currents is telling you its turns ratio: 100 A : 50 mA is 2000 turns. Where a secondary resistance is published it is worth having, because it counts against the core’s knee even though it contributes nothing to your measurement.
CT	Ratio	Secondary R	Notes
YHDC SCT-013-000	100 A : 50 mA (2000:1)	not published	split core; output clamped by an internal TVS, measured ratio 2025 on one sample; the emonTx uses 18–22 Ω and a high-sensitivity input uses 120 Ω
Nuvotem Talema AC-1030	1000:1	48 Ω at 20 °C	30 A nominal, 75 A maximum, 100 Ω recommended terminating resistor
Nuvotem Talema AC-1020	1000:1	41.8 Ω at 20 °C	20 A nominal, 60 A maximum, same 100 Ω termination
Talema AS-103	1:300	10 Ω maximum	300 Ω nominal termination, 200 mH minimum secondary inductance, 50 mA maximum secondary current
Switchgear CT to IEC 61869-2	e.g. 200 A : 5 A	varies	specified by a rated burden in VA and an accuracy class, not by a resistor

IEC 61869-2 accuracy classes for measuring current transformers

Accuracy classes are quoted AT a stated burden, in VA. A class 0.5 CT into twice its rated burden is not a class 0.5 CT any more. The hobby CTs above are typically ±1 to ±3% and are not classified at all — this table is here because the language turns up the moment you buy a CT for a distribution board.
Class	Ratio error at rated current	What it is for
0.1	± 0.1%	laboratory and reference measurement
0.2 / 0.2 S	± 0.2%	revenue metering; the S classes hold their accuracy down to 1% of rated current
0.5 / 0.5 S	± 0.5%	revenue and sub-metering
1	± 1.0%	industrial metering and instrumentation
3 / 5	± 3% / ± 5%	indication only

A current source with a compliance limit

A current transformer is not a voltage source and it is not a sensor with an output. It is a current source: the core forces the secondary ampere-turns to mirror the primary ampere-turns, and it will do whatever it takes to the secondary voltage to make that happen. The burden resistor is the only thing that decides what that voltage is. Put 39 Ω across it and 30 mA of secondary current makes 1.17 V; put 390 Ω across it and the same 30 mA tries to make 11.7 V, and somewhere on the way the core runs out of flux and stops cooperating.

Never open-circuit a live CT. This follows from the same fact and it is the one thing on this page that can hurt you. With no burden there is nothing to limit the secondary voltage, so the core drives itself hard into saturation and the secondary develops a high-voltage spike at every zero crossing — hundreds or thousands of volts on a switchgear CT, enough to break down the winding insulation and enough to kill. A CT secondary is short-circuited, not opened, when it is not in use; that is why metering CT terminal blocks have shorting links, and why a clamp meter’s jaws are opened, not its leads unplugged. Split-core hobby CTs of the SCT-013 kind usually contain a clamping device across the output for exactly this reason — the OpenEnergyMonitor teardown of the SCT-013-000 found a transient voltage suppressor, with earlier versions using a pair of 22 V zeners — but you should not rely on finding one.

Sizing. The chain is short. Full-scale primary current divided by the turns ratio gives the secondary rms current; multiply by √2 because the ADC sees peaks; divide into the peak voltage you have available, which for the usual single-supply circuit is half the reference because the signal has been lifted to mid-rail. That is the whole calculation, and the well-known compact form AREF · N ÷ (2√2 · I_max) is the same line rearranged. The only judgement in it is how much of the swing to use: filling the ADC exactly at your stated maximum leaves nothing for the moment the maximum turns out to have been optimistic.

Then check the core. The secondary current has to be pushed through the burden AND through the secondary winding’s own resistance, and the core has to generate the EMF for both. Compare that EMF with the CT’s knee point — the voltage at which a 10% increase in applied EMF needs a 50% increase in magnetising current, which is how IEC 61869-2 defines it. Past the knee the magnetising branch stops being a negligible shunt, the secondary current no longer tracks the primary, and the meter reads low and distorted. This is the failure that gets mistaken for a calibration error, and it appears first at high currents, which is precisely where an energy meter is least likely to be checked.

What this page leaves out: the CT’s own ratio and phase error are computed only if you know the magnetising inductance, and they are the reason a CT that measures current perfectly well can still get real power wrong — a 4° phase error at a power factor of 0.5 is a 12% error in watts. DC in the primary, which saturates a CT and is increasingly common downstream of half-wave rectified loads, is not modelled at all. For the rest of a meter see the energy meter pulse calculator, the ADC resolution calculator and the power factor calculator; for a shunt instead of a CT, the shunt resistor calculator.

Frequently asked questions

How do I calculate the burden resistor for an SCT-013-000 and an Arduino?

The SCT-013-000 is 100 A : 50 mA, which is 2000 turns. At 100 A the secondary carries 50 mA rms, or 70.7 mA peak. A 5 V Arduino biased to mid-rail gives you 2.5 V of peak swing, so the burden is 2.5 ÷ 0.0707 = 35.4 Ω — and the conventional choice is 33 Ω rather than 36 Ω, because a value above the ideal clips the top of the range. On a 3.3 V board the same CT wants about 23 Ω.

What happens if I disconnect the burden resistor from a live CT?

The secondary voltage rises until the core saturates, and a saturating core produces a very large voltage spike at each zero crossing. On an instrument CT that can be hundreds or thousands of volts: a shock hazard and enough to destroy the winding insulation. Short the secondary before disconnecting anything, and never rely on an internal clamp being fitted.

Why does my CT read low at high currents?

Almost always saturation. The core has to generate enough EMF to drive the secondary current through the burden and through its own winding resistance; when that EMF passes the knee, the magnetising current stops being negligible and the secondary current stops mirroring the primary. A smaller burden with an amplifier after it fixes it; a bigger burden makes it worse.

Should I pick the nearest standard value or the next one down?

Down, unless you have left headroom deliberately. A resistor larger than the ideal makes full-scale current clip against the reference; a smaller one just means the ADC is not quite filled and the last fraction of a bit is unused. Either way, put the value you actually fitted into your firmware’s calibration — the error is a pure scale factor and costs nothing once it is accounted for.

Does the burden resistor’s power rating matter?

Usually not — a hobby CT at full scale dissipates tens of milliwatts. It starts to matter on a 5 A switchgear CT, where 5 A through a 0.5 Ω burden is 12.5 W and the burden is a wirewound part, not a chip resistor. The page prints the figure so you can see which case you are in.

What is the difference between a metering CT and a protection CT?

A metering CT is designed to be accurate at and below rated current and to saturate quickly above it, so that a fault does not destroy the meter — that is what the instrument security factor FS is for. A protection CT is designed to stay linear well past rated current, which is why it is specified by a knee-point EMF. Using one for the other’s job gives you either an inaccurate meter or a relay that does not see the fault.

Related calculators

References

OpenEnergyMonitor, CT Sensors — Interfacing with an Arduino. The published worked example this page reproduces exactly: 100 A maximum, a 2000-turn SCT-013-000, 5 V AREF biased to mid-rail, giving 35.4 Ω and the advice to fit 33 Ω — “always choose the smaller value, or the maximum load current will create a voltage higher than AREF” — together with the compact form AREF × turns ÷ (2√2 × maximum primary current).
OpenEnergyMonitor, YHDC SCT-013-000 Current Transformer — a report on its properties. Measured ratio 2025 against a nominal 2000, phase error within about 4° with an 18 Ω burden and nearly 10° with 120 Ω, amplitude error under 1% to 650 Hz, and the internal clamp: two 22 V zeners on older units, output clipped at ±7.5 V on recent ones.
Nuvotem Talema, Current Sense Transformers & Inductors catalogue. The AC-1030’s 1000:1 ratio, 48 Ω secondary DC resistance at 20 °C, 30 A nominal and 75 A maximum primary current and 100 Ω recommended terminating resistor, and the AS series table used for the AS-103 figures quoted here.
IEC 61869-2:2012, Instrument transformers — Part 2: Additional requirements for current transformers. The source of the accuracy-class language used above, of the definition of rated burden in VA, and of the knee-point EMF definition — the EMF at which a 10% increase requires a 50% increase in magnetising current.
YHDC, SCT-013-000 current transformer data sheet (via Elecrow). 0–100 A input range, 100 A : 0.05 A turns ratio, ±3% non-linearity, grade B, 1000 V AC dielectric strength and an internal TVS on the output.