A thermistor is a resistor whose resistance changes significantly with temperature. It is used to measure temperature, compensate for thermal drift, limit startup current, and respond to overheating. Thermistors fall into two main categories: negative temperature coefficient (NTC) and positive temperature coefficient (PTC). Although both rely on temperature-sensitive resistance, they behave differently and serve different purposes. This guide explains their operating principles, specifications, differences from other temperature sensors, and practical applications.
Thermistor combines “thermal” and “resistor”. Unlike fixed resistors, thermistors deliberately exhibit a characterized resistance change with element temperature. Metal-oxide ceramics are common in NTC sensors; doped barium titanate in switching ceramic PTCs. Silicon and conductive polymers support other behaviors.
Packages include SMD chips, beads, leaded discs and protected probes. Thermal contact, airflow, insulation and mounting affect response time and the temperature actually sensed. Appearance does not identify NTC versus PTC.

Figure 1. Representative Thermistor Packages (Generic Illustrations)
NTC resistance decreases as temperature rises. Sensing NTCs use a specified R–T curve under low excitation. Inrush NTCs instead self-heat under load: high cold resistance limits startup current, then falls. Matching R25 does not make these parts interchangeable.
PTC resistance rises within its specified positive-coefficient region. Switching ceramic PTCs (often barium titanate) show a sharp rise near a transition and serve protection, threshold sensing and heating. Silicon PTCs, including TI TMP61, provide a more nearly linear sensing curve. Polymer PPTC resettable protectors, such as Littelfuse 1206L, heat under overcurrent and have distinct hold/trip characteristics. These three classes are not interchangeable.

Figure 2. NTC/PTC Electrical Symbols (−t°/ +t° Identify Coefficient Direction)
| Function / device | Operating behavior | Ratings and data to verify |
| NTC temperature sensor | Low-power sensing; R decreases with T | R25; R–T table; Beta interval/tolerance; response; temperature; insulation |
| NTC inrush limiter | Cold resistance limits startup surge; self-heating lowers R | Cold R; continuous current; pulse tests; derating; hot restart |
| Switching ceramic PTC protector | Resistance rises around switching temperature | Switching curve; voltage/current; trip; residual current; reset |
| Silicon sensing PTC | Nearly linear positive temperature coefficient | Reference R; TCR; tolerance; bias; temperature range |
| Polymer resettable PPTC | Fault current heats polymer; resistance rises | Ihold/Itrip; Vmax/Imax; trip time; derating; reset |
| PTC heater | Fixed-voltage heating decreases near transition | Voltage; thermal balance; power; mounting; insulation |
These categories describe temperature response, not complete application specifications. The manufacturer’s product family and datasheet determine what each device is designed to do.
Heating increases charge transport and lowers resistance in typical metal-oxide NTCs. Switching ceramic PTCs rise sharply near a transition; silicon sensing PTCs respond more nearly linearly. Figure 3 shows qualitative shapes, not measured device curves.

Figure 3. Qualitative NTC and Switching Ceramic PTC R–T Curves
A current-carrying thermistor dissipates power:
Excess heating biases a sensing device above its target temperature; inrush and protective devices exploit heating. The steady-state estimate requires a dissipation constant measured for the relevant mounting and environment.
Worked example: Figure 5 uses 3.3 V across a 10 kΩ fixed resistor and a hypothetical 10 kΩ NTC at 25°C. Both resistors carry I = 3.3 V / 20 kΩ = 0.165 mA. The NTC therefore dissipates (0.165 mA)² × 10 kΩ = 0.27225 mW. If, solely as an illustration, the installation-specific were 1.0 mW/K, the estimated steady-state rise would be 0.272°C; at = 5.0 mW/K it would be . These are assumed dissipation constants, not claims about the Vishay sensor.
Intermittent excitation can lower average heating if the sensing and sampling sequence allows stabilization. With an ADC, the source impedance of the divider and the ADC sample-and-hold capacitance also matter: acquisition time must allow the input to settle after channel switching or power gating. Where the ADC reference tracks the same stable divider supply, a ratiometric conversion (ADC code divided by full scale) suppresses much of the supply-amplitude dependence. This does not remove resistor, ADC nonlinearity, or thermal errors.
R25 is an NTC’s specified resistance at , not a constant over temperature. Its tolerance applies near the reference point. Two 10 kΩ parts can differ substantially elsewhere on their R–T curves.
Beta (K) approximates an NTC’s R–T slope over a stated interval, such as B25/50 or B25/85. Under the constant-Beta model, relative sensitivity is:
At (298.15 K), B = 3950 K yields , or −4.44. This local slope does not apply uniformly across temperature. Higher Beta raises relative sensitivity at a fixed temperature, not necessarily measurement accuracy.
For an NTC, the Beta equation uses absolute temperatures T and T₀ (K):
A hypothetical sensor with R25 = 10 kΩ and B = 3950 K gives:
| Temperature | Calculated R (hypothetical model) |
| −10°C | 58.25 kΩ |
| 0°C | 33.62 kΩ |
| 25°C | 10.00 kΩ |
| 50°C | 3.588 kΩ |
| 85°C | 1.087 kΩ |

Figure 4. Resistance–Temperature Curve of a 10 kΩ NTC Thermistor
Use the manufacturer’s full R–T table or validated Steinhart–Hart coefficients for greater accuracy:
A, BSH and CSH depend on the product and resistance units; BSH is not the Beta value. Follow the manufacturer’s fitting range and equation convention.
The Vishay NTCLE100E3103JB0 is an actual radial-leaded NTC sensor, distinct from the hypothetical 3950 K example. Its ordering code identifies a 10 kΩ R25 with ±5% resistance tolerance; the published series table lists B25/85 = 3977 K with ±0.75% B tolerance. The manufacturer’s nominal R–T table is therefore the correct source for its readings—not Figure 4.
| Characteristic | Vishay NTCLE100E3103JB0 |
| R25 | 10 kΩ nominal |
| R25 tolerance | ±5% (“J” ordering code) |
| B25/85 | 3977 K |
| B25/85 tolerance | ±0.75% |
| Nominal R at 0 / 25 / 50 / 85°C | 32,554 / 10,000 / 3,605 / 1,070 Ω |
| Operating range at zero power | −40 to +125°C continuously (series datasheet) |
| Mechanical form | Radial-leaded, solder-mounted; do not infer insulation from coating |
To read Vishay’s table, locate the column headed NTCLE100E3103 (the 10 kΩ series entry), then select a temperature row. At it reports 3,605 Ω nominal; at it reports 10,000 Ω. The adjacent TCR and B-tolerance-deviation columns explain temperature-dependent behavior and uncertainty. A listed nominal R does not include the full ±5% part tolerance or assembly error; see Vishay’s tolerance-combination guidance.
Mounting restrictions matter. The coating has no specified insulation properties, and the family datasheet says the parts are not intended for potted applications. Vishay’s dedicated handling sheet limits soldering heat/time and minimum distance from the body according to lead material, advises against pin-in-paste reflow, and prohibits bending the leads closer than 4 mm to the coated body. Do not substitute these generic material rules for a verified drawing when configuring the production footprint.
| Specification | Why it Matters |
| Thermal time constant | 63.2% response to a temperature step under test conditions |
| Dissipation constant | Steady power for a 1°C rise; depends on installation |
| Operating temperature limits | Rated temperature limits and applicable derating |
| Insulation and isolation | Explicitly rated coating, housing and voltage isolation |
| Surge/current ratings | Product-specific continuous current and pulse ratings |
| PTC switching/reset | PTC trip and reset behavior under stated conditions |
For temperature measurement, NTC sensors compete with RTDs, thermocouples and IC sensors. These differ in sensed quantity and interface requirements:
| Feature | NTC Thermistor | RTD | Thermocouple | IC sensor |
| Sensed quantity | Resistance | Metal resistance | Small thermal voltage | Analog or digital signal |
| Curve | Nonlinear | Relatively linear | Nonlinear | Often calibrated internally |
| Typical strength | High local sensitivity | Repeatability, stability | High-temperature capability | Integrated signal conditioning with analog or digital output |
| Key design issue | Bias self-heating; R–T data | Lead/excitation errors | Cold-junction compensation | Supply/interface and location |
| Typical use | Battery / PCB monitoring | Industrial process | Furnace / high-temperature | Board / system monitoring |
Accuracy, response and cost depend on the specific part and installation. This comparison excludes inrush NTCs, protection PPTCs and PTC heaters.
Figure 5 uses a hypothetical 10 kΩ, B = 3950 K NTC from Vout to ground and a 10 kΩ resistor from Vout to 3.3 V. Assume a high-impedance, settled ADC input. This is not the Vishay datasheet example.

Figure 5. Low-side 3.3 V NTC Divider, Vout Falls as Temperature Rises
Step 1: At , the Beta model gives
Step 2:
Step 3: Recover resistance from the measured voltage:
Using the unrounded voltage gives 3,588.18 Ω, or approximately 3.59 kΩ. If the ADC reference equals VCC, the same expression can be applied using the measured ADC fraction instead of absolute voltage.
Step 4: Apply the inverse Beta equation:
With T₀ = 298.15 K, R₀ = 10,000 Ω and B = 3950 K, this gives T = 323.15 K = . This ideal calculation is not a system-accuracy claim.
A practical accuracy claim requires a component-level and system-level error budget. Estimate each source in its own units and operating conditions before combining independent errors. In particular, ±5% R25 tolerance, fixed-resistor error, ADC gain/offset/INL and reference mismatch, model-fitting error, self-heating, and thermal contact are not the same quantity.
| Error source | What to check / effect |
| Sensor R25 and B tolerances | R25/B tolerances change resistance and curve shape |
| Fixed resistor | Tolerance in divider resistor skews resistance inversion |
| ADC / reference | Gain, offset, INL, reference mismatch and settling |
| Curve conversion | Beta fit or lookup-table residual error |
| Self-heating | Dissipation raises thermistor above target temperature |
| Thermal contact / placement | Sensor placement and imperfect thermal coupling |
For the hypothetical 3950 K sensor at , . A small 1% relative resistance error corresponds to , or about locally. This is only one sensitivity conversion—not a total system accuracy specification. The sign depends on whether measured resistance is high or low. For reliable reporting, test the assembled circuit over temperature and document any calibration.
Electronic components change behavior with temperature. A thermistor placed in an appropriate resistor network can counter part of that variation. For example, a temperature-dependent bias or reference circuit can use the thermistor’s changing resistance to offset an unwanted drift elsewhere in the circuit.
Successful compensation depends on matching the required correction curve rather than merely matching a resistance at room temperature. If the thermistor and the affected component change at different rates, the correction will be accurate only over a limited span. Circuit designers therefore evaluate the full operating-temperature range and the tolerances of all relevant components.
A cold NTC in series with an input supply limits the capacitor-charging current at startup. As load current heats it, resistance decreases and the normal-operation voltage drop becomes lower than its cold-start value. Figure 6 depicts an AC-input arrangement with the NTC ahead of a bridge rectifier and the polarized bulk capacitor on the rectified DC bus. The arrangement is illustrative; actual mains input design requires appropriate fusing, isolation, grounding, and approvals.

Figure 6. Series NTC Before AC Bridge Rectifier
A separate illustrative 24 V DC source charging a discharged 4,700 capacitor stores at full voltage. This is capacitor stored energy, NOT proof that a specific NTC will dissipate exactly 1.35 J. Energy is shared among source resistance, wiring, charging path, the NTC and other components, and depends on waveform and switching conditions.
Suitability requires cold resistance and peak-current evaluation, manufacturer pulse-test waveform and repeated-pulse limits, continuous current and ambient-temperature derating, steady-state temperature, and the shortest expected off/on restart interval. A still-hot NTC may not limit a second surge adequately. Never select a current limiter solely because its nominal energy rating exceeds 1.35 J.
A protection PTC changes to a high-resistance state when self-heating under a fault exceeds its thermal balance. The datasheet must be read as a protection characteristic: Ihold is the maximum current that does not trip under a specified ambient condition, while Itrip is the minimum current that trips under the specified test condition. Between these limits, the response depends on temperature and time; trip-time curves, maximum voltage/fault current, ambient derating, and residual current after tripping all matter.
For a documented polymer example, Littelfuse’s 1206L020/30 PPTC lists Ihold = 0.20 A, Itrip = 0.42 A at still air, Vmax = 30 V DC, and a maximum time to trip of 0.10 s at a specified 8 A test current. The manufacturer’s temperature-rerating table lowers the hold current to approximately 0.14 A at . These figures belong only to that particular PPTC product and its stated conditions, not to silicon sensing PTCs, ceramic PTC sensors, or heater elements.
Residual current flows while a tripped PPTC remains powered. Remove the fault and let it cool to reset. It is not automatically a substitute for a safety fuse; sensing ceramic PTCs instead signal a controller.
A ceramic PTC heater self-limits power at near-constant voltage as resistance rises . Mounting and airflow determine its thermal balance; use heater-specific insulation and voltage ratings.
Troubleshooting should separate an actual sensor fault from the connector, wiring, ADC, or conversion setting. Work on a de-energized isolated circuit. Disconnect power, discharge all relevant capacitors safely, and verify the voltage is zero before resistance or continuity testing. For any mains-connected appliance, use an appropriate service procedure and qualified personnel. Do not use the ohmmeter on a powered circuit.
First identify the exact sensor and datasheet. Disconnect at least one lead or isolate the sensor from parallel circuitry, allow the element to stabilize at a known temperature, and measure resistance with a suitably low test current. Repeat at a second stabilized temperature using a calibrated temperature reference and the correct R–T curve. Do not judge a sensor from a single room-temperature reading alone; compare expected values with tolerance, and do not heat a live assembly without adequate safeguards.
After the sensor checks, reconnect and inspect connector contact resistance, lead continuity, solder joints, cable pinout, the fixed resistor, ADC reference, code scaling, and the selected curve or coefficients. Recheck the input voltage under normal operation using proper electrical precautions. In Figure 5 only, a low-side NTC that opens pulls Vout toward 3.3 V through Rfixed; a shorted NTC pulls Vout toward 0 V. Other divider topologies reverse or change these symptoms.
| Observation (Figure 5 circuit) | Likely cause and discriminating test |
| Vout near 3.3 V | Open lower NTC path: test sensor, connector and return wire |
| Vout near 0 V | Short NTC/lower wire, or missing upper supply; test both |
| R correct off board; wrong Vout | Check fixed resistor, wiring, reference and ADC settling |
| R correct at two temperatures; wrong displayed °C | Check Beta/R–T data, orientation, software and units |
| Intermittent reading | Check contacts, moisture, leads and solder joints |
| Persistent offset or sluggish response | Check self-heating, thermal contact and placement |
Confirm open/short symptoms with safe low-voltage measurements or a simulated resistor; do not swap parts on an energized circuit. Passing two-point resistance checks does not exclude ADC or firmware errors.
A thermistor replacement is not determined by package shape or a shared “10 kΩ” marking. Compare R25, resistance tolerance, the complete R–T table, Beta value AND its reference interval, conversion model, operating temperature, thermal time constant, dissipation constant, lead arrangement, package size, chemical exposure, electrical insulation/isolation rating, and attachment method. For inrush and protection products, current, voltage, surge, and reset ratings are additional hard requirements.
Mismatch example: assume a genuine hypothetical NTC has R25 = 10 kΩ and B = 3950 K. At an actual its resistance is about 3.588 kΩ. If unchanged software incorrectly decodes that resistance with B = 3435 K, the inverse Beta equation reports approximately —about error. The two devices may have the same nominal 10 kΩ value at , yet are not equivalent throughout the range. This calculated example is separate from Vishay’s NTCLE100E3103JB0, which has B25/85 = 3977 K and its own datasheet curve.
Confirm any factory calibration, lookup table and firmware thresholds before substitution. A matching package does not guarantee electrical compatibility; an unrated coating is not verified insulation.
Thermistors turn temperature changes into useful resistance changes. NTC devices are widely used for sensing, compensation, and startup-current limiting, while suitable PTC devices support temperature detection, circuit protection, and controlled heating. Their behavior is defined by material properties, R–T characteristics, and thermal conditions. Understanding these factors—and checking the relevant manufacturer specifications—makes it easier to apply thermistors correctly in electronic equipment.