Hall Sensor vs CT vs Shunt Resistor
Hall sensors detect magnetic fields, current transformers transfer changing current, and shunts measure a resistive voltage drop. Hall and shunt systems can measure DC and AC; a conventional passive CT cannot measure steady DC.
Start with the waveform and isolation requirement. Then compare error at low current, heat, timing and the complete interface—not just the current rating on the first page.
Choose the architecture before the part number
Need steady DC? Start with Hall or shunt sensing. Need AC-only measurement? A suitable CT may fit well. Need isolation? Evaluate a rated Hall/CT assembly or a shunt with an added isolation barrier. None is universally the most accurate or lowest-cost option.
Scroll sideways to compare all three methods →
| Decision factor | Hall current sensor | Conventional passive CT | Shunt system |
|---|---|---|---|
| Steady DC | Yes, within range | No | Yes |
| AC and pulses | Within bandwidth and range | Within frequency, flux and reset limits | Within resistor and signal-chain limits |
| Isolation | Available in rated designs | Available through rated insulation | Requires an added barrier |
| Output principle | Field converted to a signal | Secondary current into a burden | Voltage across a known resistance |
| Auxiliary power | Usually needed | Not for the passive transformer | Usually needed for conditioning |
| Error concerns | Offset, drift, gain, external fields | Ratio, phase, burden, saturation | Tolerance, heating, offset, layout |
| Loss concerns | Conductor and electronics | Windings, core and burden | I²R plus electronics |
| Physical fit | Aperture, busbar or integrated conductor | Solid core, split core or PCB device | Series resistor and sense connections |
Compare the complete measurement channel: sensing element, conditioning, filter, power supplies, ADC and mechanical installation. A cheap resistor and a finished isolated transducer are not equivalent line items.
Two naming traps to avoid
A Hall position switch is not necessarily a proportional current sensor. And products called “DC current transformers” may use active compensation or fluxgate technology, unlike the passive CT discussed here. LEM distinguishes these magnetic sensing technologies.
Hall sensing: check the current near zero, not only at full scale
Current creates a magnetic field. A Hall element detects that field, including a static field, and electronics turn it into a useful output. That static-field response is why a suitable Hall current sensor can measure steady DC. TI explains the Hall-versus-transformer distinction.
Open loop and closed loop are different implementations
In a common core-based open-loop design, a core concentrates flux into the Hall element's gap. Coreless designs also exist. A closed-loop Hall transducer drives a compensation winding to oppose the primary field; the compensation current becomes the measurement.
Feedback can improve linearity, drift and dynamic performance, but adds circuitry and supply demand. Compare specific models rather than assuming every closed-loop device wins every specification. Source: LEM's open- and closed-loop Hall explanations.
A small offset can dominate standby current
Assume a 1.65 V zero-current reference and sensitivity of 40 mV/A:
VOUT = 1.65 V + (0.040 V/A × I)
Nominal outputs at −25 A and +25 A are 0.65 V and 2.65 V. Now add a residual 5 mV output offset. Its equivalent current error is 5 mV / 40 mV/A = 0.125 A. At a 0.5 A reading, that offset alone equals 25%.
Selection consequence: request absolute offset and drift, not only a percentage at a large current. Check output swing and ADC headroom as separate limits.
Zero-current calibration can remove an initial offset under known conditions. It does not automatically remove later temperature drift, noise or gain error. Nearby busbars, magnets and return-current paths also matter; qualify the installed geometry.
Manufacturer-documented example: TI's TMCS1100 lists a typical 1.8 mΩ input-conductor resistance. At 20 A, I²R would be about 0.72 W using that typical value, before temperature effects and electronics losses. This calculation is not a thermal approval. Check the device's safe operating area and PCB conditions. TMCS1100 Rev. C, electrical characteristics.
A CT needs the right waveform, burden and magnetic headroom
A conventional CT couples changing primary current into its secondary winding. Within its usable range, secondary current approximately follows the inverse turns ratio. A burden resistor converts that current into a voltage.
IS ≈ IP × NP / NS
VB ≈ IS × RB
IP and IS are current magnitudes, NP and NS are turn counts, and RB is burden resistance. Polarity markings determine direction. Real CTs add magnetizing-current, winding, ratio and phase errors.
DC changes are not DC measurement
A CT may produce a transient when DC changes, but it cannot sustain an output representing steady DC. DC bias can also reduce magnetic headroom and contribute to saturation.
Measuring only the AC ripple does not remove that concern. Check the permitted bias and waveform, or use an appropriate Hall or shunt channel.
More burden voltage is not a free improvement
Increasing burden resistance raises the output signal, but also the voltage demanded of the winding. Pulse duration, frequency, winding resistance and core properties then affect flux excursion. Pulse-sensing circuits need suitable reset between pulses; long pulses can droop or saturate.
A switching-converter CT is not automatically a 50/60 Hz metering CT. Coilcraft's current-sense transformer guide explains ratio, termination and reset constraints.
A 20 A RMS signal must still fit the ADC peaks
With one primary turn, 1000 secondary turns and 20 A RMS sinusoidal primary current, ideal secondary current is 20 mA RMS. A 50 Ω burden produces 1 V RMS and dissipates 0.020 W.
The voltage peak is about 1.414 V. Centering it at 1.65 V with a suitable interface gives a nominal 0.236–3.064 V span. That is not proof of compatibility with every 3.3 V ADC: allow for tolerances, required headroom, overload and transients.
Do not open an energized current-output CT secondary
A current-output CT can develop hazardous secondary voltage when its circuit is opened while primary current flows. Follow the manufacturer's burden, rated shorting and servicing arrangements. Hazardous-voltage work requires qualified personnel; this guide is not a live-work procedure.
Some voltage-output CT assemblies already contain a burden. Identify the exact output type and follow its instructions. Schneider Electric's open-secondary warning.
A smaller shunt reduces heat—and makes a smaller signal
A current-sense shunt sits in series with the measured path. Measure the voltage across its defined sense points and use Ohm's law. The resistor itself provides no galvanic isolation.
I = VSHUNT / RSHUNT
PSHUNT = IRMS² × RSHUNT
For steady DC, use the DC magnitude in the heating calculation. For pulses or AC, use RMS current for average resistive heating and check peak/pulse stress separately.
Scroll sideways to compare signal, heat and offset error →
| Shunt | Signal at 20 A | Heat at 20 A | Offset-equivalent error | Offset error at 0.1 A |
|---|---|---|---|---|
| 2 mΩ | 40 mV | 0.8 W | 5 mA | 5% |
| 0.5 mΩ | 10 mV | 0.2 W | 20 mA | 20% |
Reducing resistance by four reduces nominal signal and heat by four, but increases the current error from the same voltage offset by four. Choose resistance and amplifier performance together.
At 100 A, the 2 mΩ resistor would dissipate 20 W while that current flows. That is a calculated stress, not an approved overload. Continuous derating, actual cooling and pulse-load data determine suitability.
Explore the shunt signal, loss and offset trade-off
Change resistance while keeping current and offset fixed. The small-current field is a separate operating point for the percentage-error calculation. No component rating or safety approval is inferred.
Nominal shunt voltage40 mV
Nominal resistive loss0.8 W
Offset-equivalent current error5 mA
Offset error at the small current5%
Model: V = IR; P = I²R; current error = |input offset|/R. Resistance is fixed and nominal. Results exclude tolerance, TCR, self-heating feedback, gain error, noise, ADC error, thermal EMF, layout and dynamic behavior. Values are calculated locally and are not sent anywhere.
Kelvin connections and temperature can decide shunt accuracy
Kelvin sensing separates the small measurement current from the main load-current connections. This avoids including unwanted voltage drops from power traces, leads and contacts. Use the resistor's defined sense terminals or recommended pickup geometry.
Do not route load current through the sense traces. A precise resistor cannot correct a measurement taken from the wrong copper. Vishay's shunt guide covers Kelvin connections and power/pulse considerations.
Use element temperature, not ambient alone
The temperature coefficient of resistance (TCR), often stated in ppm/°C, describes resistance change with temperature. In a linear illustrative calculation, 50 ppm/°C over an 80°C element-temperature change contributes about 0.4% resistance change.
Self-heating can make the element hotter than ambient. Initial tolerance, terminal construction and amplifier drift are additional contributions. Vishay's TCR paper explains why copper terminals can matter at very low resistance values.
Accuracy, speed and isolation each need their own checks
Ask what the accuracy percentage refers to
Separate offset, gain error, nonlinearity, temperature drift and noise. Is a percentage relative to the reading, nominal current or full scale? The distinction can dominate low-current performance. More ADC bits cannot recover a signal already obscured by offset or noise.
Document which errors calibration removes and which remain. Check the smallest useful current as carefully as the largest; a range chosen only to survive startup can leave standby current poorly resolved.
Bandwidth is not protection response time
The path includes the sensor, amplifier, filtering, sampling and decision delay. Metering may need amplitude and phase accuracy; a protection function needs a complete response-time budget. A slow display and a fast control loop are different requirements.
Also separate RMS from peak current. An illustrative 10 A RMS waveform with crest factor 3 reaches 30 A peak. Assuming a sine wave would predict only about 14.1 A. Both the range and overload checks must use the actual waveform.
High common-mode tolerance does not mean isolation
TI's INA240 is a shunt amplifier with enhanced PWM rejection, a −4 V to 80 V input common-mode range, and a separate 2.7 V to 5.5 V supply range. The common-mode rating describes input voltage relative to its ground; it is not a galvanic isolation barrier. INA240 datasheet.
Isolation belongs to a rated barrier and its implementation
A suitable Hall sensor or CT can isolate the primary path from its output. A bare Hall element does not automatically provide a qualified barrier. Check working voltage, insulation category, creepage, clearance and installation conditions.
A shunt can instead feed an isolated amplifier. TI's AMC1300 transfers the measurement across an input/output barrier and uses supplies on both sides. The shunt remains in the power circuit.
A short-duration withstand test is not a continuous working-voltage rating. Component certification also does not certify the finished product. Hazardous-voltage designs need a qualified system-level review.
Which method fits your application?
Scroll sideways to review application decisions →
| Application | Candidates to evaluate | Question that decides the shortlist |
|---|---|---|
| Battery charge/discharge | Hall or shunt | Can it measure low DC current accurately in both directions? |
| AC feeder monitoring | Suitable CT; Hall where DC content matters | Do frequency, phase error, burden and installation fit? |
| Motor-drive feedback | Hall or shunt | Does it settle within the sampling window under PWM transients? |
| Converter pulse sensing | Pulse CT, shunt or Hall | Are peak current, pulse width, reset and delay covered? |
| Low-current, low-voltage board | Often a shunt channel | Is the useful signal above offset/noise without excessive voltage drop? |
A bidirectional battery monitor needs two operating points
The requirement includes steady charge and discharge current, so an ordinary passive CT is excluded. The engineering team then checks standby error and peak thermal stress separately. A shunt may satisfy accuracy but need thermal area or an isolated front end. A Hall solution may simplify the isolation path but still need low-current offset and placement checks.
What goes into the RFQ: the smallest current needing a useful reading, normal and peak current, overload duration, bidirectional output scale and the required barrier—not simply “20 A sensor.”
For retrofits, a split-core design may avoid disconnecting a conductor mechanically. That does not authorize energized installation. Follow the product and site's electrical safety procedures. A metering CT is also not automatically suitable for protection duties.
When the current reading looks wrong
Scroll sideways to see the review actions →
| Symptom | Possible cause | Review next |
|---|---|---|
| CT output disappears on steady current | DC outside the passive CT principle | Whether DC measurement is required |
| CT pulse droops or clips | Reset, saturation, burden or ADC limit | Pulse conditions and each stage's range |
| Hall reads current at zero | Offset, reference, temperature or external field | Known-zero condition and installed magnetic environment |
| Shunt reading shifts after warm-up | Self-heating, TCR or amplifier drift | Element and board temperature |
| Shunt result changes with connections | Unwanted contact or trace drop | Sense pickup points and load-current paths |
| Switching creates noise in the reading | Coupling, common-mode events or sample timing | Analog behavior versus conversion timing |
Use a reference instrument with suitable range, bandwidth and uncertainty. A stable RMS display does not validate fast peaks or protection timing. On hazardous circuits, an ordinary grounded oscilloscope connection can compromise isolation; measurement methods must be appropriately rated and used by qualified personnel.
Qualify the sensing channel before approving a substitution
- Define the waveform.DC, AC, pulses or mixed; direction, lowest frequency, RMS, peak, duty cycle and overload duration.
- Define useful accuracy.Smallest important current, error over temperature, percentage basis and planned calibration.
- Define the interface.Output scale, supplies, ADC range, filtering, bandwidth, phase and total response time.
- Define electrical and mechanical limits.Common-mode transients, isolation, aperture or conductor size, footprint, cooling and ambient conditions.
- Request architecture-specific evidence.Hall offset/drift; CT burden, frequency and saturation; shunt TCR, Kelvin layout and pulse derating.
Compare implementation cost, including conditioning, isolated power where needed, thermal area, calibration and assembly. Require engineering approval of alternatives; a matching current rating does not establish compatibility.
Send the measurement requirement with the part number
Include your approved manufacturer part number, quantity and delivery requirement, plus waveform, current range, accuracy, isolation and output interface. State whether alternatives are permitted and which specifications must not change.
YURUNOX is an electronic-component sourcing partner. Your engineering team retains responsibility for circuit selection, safety and substitute approval.
Hall sensor vs CT vs shunt resistor: FAQ
Can a current transformer measure DC current?
A conventional passive CT cannot measure steady DC. A changing DC current may create a transient, but that is not a continuous DC measurement. Some products called DC current transformers use active compensation or fluxgate technology. Check the actual architecture rather than relying on the product name.
Can a Hall current sensor measure both AC and DC?
A suitable Hall current sensor can measure both within its rated range and bandwidth. Check zero-current offset, sensitivity, temperature drift, peak range and output interface. A Hall position switch is not automatically a proportional current sensor.
Is a shunt resistor more accurate than a Hall sensor?
Not automatically. Shunt tolerance, temperature, connections, amplifier offset and ADC errors all contribute. Hall performance also varies by architecture and model. Compare the complete channel at the actual current and temperature, especially at the smallest current that matters.
Can a shunt-based measurement have galvanic isolation?
Yes, when the shunt voltage is transferred through a suitable isolated amplifier or converter. The resistor itself provides no isolation. The complete design must meet the required barrier, supply, spacing and insulation requirements; an isolation IC alone does not qualify the finished assembly.
Why does a CT need a burden resistor?
A current-output CT needs a suitable load. A resistive burden converts secondary current into a measurable voltage and affects signal amplitude, error and magnetic flux demand. Some voltage-output CT assemblies already contain a burden. Identify the output type before designing the interface.
What happens if a CT secondary is left open?
An energized current-output CT can develop hazardous secondary voltage. Follow the manufacturer's burden, shorting and maintenance instructions. Do not disconnect an energized secondary to test this behavior. Installation and servicing on hazardous electrical systems require qualified personnel.
Which current-sensing method has the lowest power loss?
It depends on the implementation. Shunts have resistive loss; Hall sensors and CTs can have conductor, winding, core, burden or electronics losses. An aperture Hall transducer is a different case from an integrated-conductor Hall IC. Compare complete-channel loss at the actual RMS current.
Can the same current rating make two sensing methods interchangeable?
No. Output scale, supply, isolation, bandwidth, burden, thermal behavior and mechanical connections can all differ. Define the waveform and full interface requirements, then qualify the proposed alternative. Matching amperes is not evidence of a drop-in replacement.
Technical references
Manufacturer documentation supports the principles and named examples. Use the current device revision and application conditions before release.
- LEM: Technologies and Innovations — Hall, fluxgate and hybrid distinctions.
- LEM: Hall Effect Current Sensors — open- and closed-loop principles.
- TI SSZT306: Hall Current Sensing in Power Supplies — AC/DC magnetic measurement.
- TI TMCS1100 datasheet, Rev. C — conductor resistance and thermal limits.
- Coilcraft: Current Sense Transformers for Switched-mode Power Supplies — ratio, burden and reset.
- Schneider Electric FA125574 — current-output CT open-secondary hazards.
- Vishay: Shunts and Current-Sensing Resistors — Kelvin sensing, power and pulse loading.
- Vishay 30405: Temperature Coefficient of Resistance — self-heating and construction effects.
- TI INA240 datasheet — PWM rejection and common-mode range.
- TI AMC1300 datasheet — isolated shunt-signal transfer.
Examples are educational calculations or identified manufacturer specifications, not YURUNOX test reports, customer outcomes or design approvals. This article does not provide instructions for live electrical work.