Choose your task
An NTC thermistor has a negative temperature coefficient: its resistance falls as temperature rises. This calculator models a voltage divider made from the thermistor and a fixed resistor, then checks how an analog-to-digital converter (ADC) can measure its output.
- Find a Resistor: compare standard resistor values and choose a fixed resistor for your thermistor voltage divider.
- Check a Resistor: enter the fixed resistor you already use to check ADC range, temperature resolution, current and power. The calculator uses your value until you enter another value or select an alternative.
Quick start
- Enter your thermistor data. Choose Datasheet R25 + Beta, manufacturer Steinhart–Hart coefficients, or Import resistance–temperature data.
- Set the operating temperature range. Enter the coldest and hottest temperatures you need to measure. Check the sensor’s rated range in its datasheet.
- Configure the supply and ADC. Enter the divider supply voltage, ADC bit depth, reference and thermistor connection. A ratiometric reference must physically share the divider supply.
- Choose or check the fixed resistor. Set a Selection priority for Find a Resistor, or enter Your fixed resistor value for Check a Resistor. Enter circuit limits from your design requirements and component datasheets.
- Review and export. Use Circuit & Results to inspect the circuit and curves, Detailed Results to inspect the numerical checks, and C Code & Export to generate the ADC-to-temperature conversion.
Choose thermistor data
- Datasheet R25 + Beta: enter resistance at 25°C (R25), the Beta coefficient in kelvin, and their tolerances. A single Beta value is an approximation over the selected temperature range.
- Steinhart–Hart coefficients: enter the manufacturer’s A, B and C coefficients for resistance in ohms. This option accepts coefficients; it does not calculate or fit them from measurements.
- Import resistance–temperature data: use a manufacturer R–T table, with optional minimum and maximum resistance columns for manufacturing tolerance.
Nominal means the central values you entered, before component tolerances are applied. The nominal model is used for the main voltage curve and generated temperature-conversion code.
The resistance fields accept formats such as 10K, 10,000 or 10 kΩ; a number without a unit means ohms. Fields format the value when you leave them or press Enter.
A datasheet label such as B25/85 or β25/85 means Beta was specified between 25°C and 85°C. Enter the Beta coefficient from that datasheet row. The Datasheet Beta temperature interval records this reference information only; it does not change calculations or set the operating range.
R25 and Beta tolerances
These are component manufacturing limits, not temperature tolerances. An R25 of 10 kΩ with ±1% tolerance may be 9.90–10.10 kΩ at 25°C. A Beta coefficient of 3950 K with ±1% tolerance may be 3910.5–3989.5 K.
R25 tolerance shifts the resistance scale. Beta tolerance changes the curve’s shape; in the Beta equation anchored at 25°C, it has no effect at exactly 25°C. The calculator combines these limits independently with fixed-resistor and applicable supply/reference tolerances.
Tolerances can change the recommended fixed resistor. For an unchanged resistor value, the nominal curve and generated C code still use nominal R25 and Beta. Entering zero tolerance assumes an exact value; it does not establish that a real part has no uncertainty.
Use the tolerance for the exact orderable part. For example, the Vishay NTCSI0603E3 datasheet lists a 10 kΩ / B25/85 = 3950 K option with R25 tolerance ±1% and Beta tolerance ±3%. Do not substitute a family-wide range or assume the calculator’s defaults match your part.
Import a resistance–temperature table
- Paste the table or upload a CSV, TSV, TXT or XLSX file. For a workbook, choose the worksheet containing the data.
- Map the columns by their position: the first column is 1. Choose the temperature and resistance units used in the file.
- Map the nominal resistance column. If the file includes tolerance bounds, map both the minimum and maximum resistance columns; otherwise leave both blank.
- Inspect the preview, then choose Apply Imported Data. Temperatures must be unique, and resistance must decrease as temperature rises.
The calculator interpolates within the imported R–T table and does not extend it beyond its endpoints. Missing minimum/maximum columns mean thermistor manufacturing tolerance is not included for that table.
Leave room for tolerance analysis. Component variation can make a reading near an operating endpoint correspond to a temperature outside the table. To measure through 85°C, for example, you may need manufacturer data extending above 85°C. The same issue can occur at the cold endpoint.
If More data needed appears, review the stated table range and calculated conversion coverage. Use Adjust Temperature Range to narrow the operating range, or Edit Imported Data to supply a wider table. The required margin depends on the data and tolerances. ADC/reference problems have separate guidance; a wider table alone will not fix them.
Understand the results and circuit limits
- Temperature resolution is the temperature change represented by one ADC count, in °C/count. A smaller value means a finer temperature step; it is not measurement accuracy.
- Maximum tolerance error is the largest magnitude of the calculated component-tolerance error. The negative and positive errors need not be equal.
- Estimated temperature error combines sampled component-tolerance effects, a local half-count ADC contribution and nominal self-heating when a dissipation constant is supplied. It is not a guaranteed total accuracy specification.
- Firmware conversion error is the difference between the generated conversion and the nominal thermistor model. It is separate from the sensor’s physical accuracy.
For Check a Resistor, the limit report identifies each failed check and links to its input. Review the calculated value and relevant datasheet limit before changing the circuit or requirement. Resistor series and search resistance bounds only control the search for alternatives; they do not disqualify your entered resistor.
Electrical limits can pass while the temperature error estimate is incomplete, for example when tolerance readings leave an imported table’s coverage. Circuit results remain visible where possible. Firmware export requires the configured circuit limits to pass; generated code converts the nominal model.
If no resistor qualifies, first identify whether the issue is an electrical limit, missing table coverage or an ADC/reference conversion problem. Use Check a Resistor to inspect a particular value. Do not raise limits simply to remove a warning.
For self-heating, use the datasheet dissipation constant for the actual mounting and airflow. Values in mW/K and mW/°C are equivalent for a temperature rise. For example, 0.644 mW ÷ 2 mW/°C gives 0.322°C rise; at 6 mW/°C it gives about 0.107°C. A component power rating is a separate limit, not a permitted temperature error.
Worked example: Vishay NTCLP100E3103H with a 12-bit ADC
This example checks a 10 kΩ fixed resistor using the manufacturer’s resistance–temperature table for the Vishay NTCLP100E3103H. The sensor data come from pages 1 and 3 of the Vishay NTCLP100 datasheet, revision 22 January 2026.
The datasheet specifies R25 = 10,000 Ω ±3%, B25/85 = 3977 K ±0.75%, a zero-dissipation operating range of −40°C to 85°C, and a dissipation factor of 6 mW/K. The main example uses the R–T table rather than a single-Beta approximation. R25 and Beta are not entered again when importing this table; the imported resistance values and bounds define the model.
1. Import the manufacturer’s R–T table
- Download the NTCLP100E3103H example R–T CSV. It contains all 26 datasheet rows from −40°C to 85°C.
- In Thermistor model, choose Import resistance–temperature data. Upload the downloaded CSV, or paste its contents.
- Set Temperature column to 1, Nominal resistance column to 2, Minimum resistance column to 3, and Maximum resistance column to 4.
- Choose temperature units °C and resistance units Ω. Leave First row contains column headings selected.
- Check the preview, then select Apply Imported Data.
The nominal resistances come directly from the datasheet. The minimum and maximum resistance columns are derived bounds: each nominal value is multiplied by (1 − R-TOL/100) and (1 + R-TOL/100), using the R-TOL percentage listed for that row in the datasheet. The CSV has four columns and does not include a separate tolerance-percentage column. These bounds are not separate measured curves.
2. Enter the circuit settings
The following 5 V supply and ADC configuration are example circuit choices, not sensor datasheet specifications. Use them only with an ADC whose reference and input ranges permit this 0–5 V measurement. The ratiometric reference must physically share the divider’s 5 V supply.
The selected operating range is −20°C to 80°C. Importing the full −40°C to 85°C table leaves data beyond both operating endpoints for the tolerance calculations.
3. Enter the resistor and circuit limits
Choose Check a Resistor and enter the fixed 10 kΩ resistor. Under Resistor selection → Resistor search range & circuit limits, enter the limits and dissipation constant below.
Dissipation constant (D) = 6 mW/°C comes from the datasheet’s 6 mW/K dissipation factor; the units are equivalent for a temperature rise. Use the appropriate value for the actual mounting and airflow.
The 1% resistor and supply tolerances, 1 mA current limit, 1 mW power limit, 0.25°C allowed rise, and 10 kΩ ADC source impedance limit are illustrative design assumptions. Replace them with your component, measurement and ADC requirements. The 1 mW measurement-power limit is not the thermistor’s maximum datasheet power rating.
The datasheet lists 250 mW at 55°C and provides a power-derating curve. That rating changes with ambient temperature. Use the lower measurement-power limit to control self-heating, and check the derated component rating separately.
All E-series and Selection priority control the alternative-resistor comparison only. They do not change or disqualify the 10 kΩ value entered in Check a Resistor. Find a Resistor can recommend a different value for the same circuit requirements.
4. Check the nominal results
These are nominal values before component tolerances. At 25°C, divider current is 250 µA, thermistor power is 0.625 mW, and ADC source impedance is 5 kΩ. With D = 6 mW/°C, the nominal self-heating estimate is 0.625 / 6 = about 0.104°C. The circuit-limit report includes the entered tolerances, so its maximum values can be higher.
At the calculator’s 1°C temperature-check step, all configured circuit limits pass for this example. This does not establish guaranteed temperature accuracy. The ADC codes use the calculator’s ideal convention, round(4095 × Vout / 5 V); the rounded midpoint code alone does not represent exactly 25°C.
Optional alternative: enter the datasheet R25 and Beta values
To compare the single-Beta approximation with the imported table, keep the same circuit settings and choose Datasheet R25 + Beta. Enter:
B25/85 describes the interval used to specify the Beta coefficient. It does not set the operating temperature range: keep Minimum temperature = −20°C and Maximum temperature = 80°C for this comparison.
A single 3977 K Beta coefficient is an approximation. Its curve differs from the manufacturer’s table, particularly below 25°C. The checkpoints above use the imported R–T data, so the Beta-only option will not reproduce every checkpoint.
Practical NTC interface: protection and filtering
These circuits add input protection and filtering to the thermistor voltage divider. DTVS diverts input transients, RS limits current into the clamps, DH and DL clamp the ADC signal, and CF filters noise and supplies brief ADC sampling current.
The calculator chooses or checks Rf. The added TVS, series resistor, clamp diodes and capacitor require separate selection for your ADC and expected faults. RS ≈ 2 kΩ and CF = 100 nF (0.1 µF) are example starting values to evaluate for a slow thermistor interface, not universal recommendations. Verify them using the checks below.
On a narrow screen, scroll each diagram horizontally to read the labels, or open its full-size SVG.
DTVS: start with the normal signal voltage
Choose a low-leakage TVS with a reverse working or standoff voltage, VRWM, above the highest normal VDIV at its connection, including supply tolerance and operating conditions. VBR is breakdown at a specified test current; putting VBR above the signal voltage alone does not establish low leakage.
VC is the remaining voltage during a specified transient. Coordinate it with RS, the clamp diodes and component pulse ratings. Check leakage at the actual signal voltage and highest temperature, and choose ESD or surge ratings for the expected disturbance. Place DTVS near the connector with a short transient-current return.
RS: balance fault current, leakage error and settling
Too little series resistance can allow excessive fault current into DH, DL or the ADC. More resistance reduces that current, but slows capacitor recharge and increases voltage error from leakage. With no input current, RS does not change the ideal, fully settled DC divider voltage.
Values above 10 kΩ can affect acquisition and leakage error, but there is no universal cutoff. Check the ADC’s source-resistance limit, acquisition time, sample rate and external-capacitor guidance using RS + (Rf in parallel with RNTC), not RS alone. Verify resistor pulse-voltage, energy and continuous-power ratings.
Once DH is conducting, VADC is approximately VCLAMP + VF. For a first positive-pulse estimate, use I ≈ (VTVS,clamp − VCLAMP − VF) / RS when the numerator is positive. VTVS,clamp is the remaining voltage at the TVS and VF is DH’s forward voltage. Check the resulting diode, rail and ADC injection currents and the resistor’s pulse ratings. Fast-transient behavior still needs validation.
DH and DL: check forward voltage and leakage
Start with low-leakage Schottky diodes. DH’s anode connects to VADC and cathode to VCLAMP; DL’s anode connects to ADC ground and cathode to VADC. Check forward voltage at the fault current and temperature against the ADC’s pin-voltage and injection-current limits.
A Schottky label does not guarantee conduction before the ADC’s internal clamp. Temperature-dependent leakage can shift the reading. Additional resistance between the clamp node and ADC pin may be needed to limit remaining injection current; include it in the source and settling checks.
VCLAMP: use a rail that can safely accept current
Use an ADC-compatible rail or clamp circuit that can absorb injected current without an unsafe voltage rise. The ADC’s 5 V analog supply is a possible choice here only if the ADC permits that arrangement and the supply handles the fault current.
VREF is not automatically suitable: many references cannot absorb current. Check powered-off conditions: injection can back-power a rail, and a powered clamp rail may not protect an unpowered ADC. Follow the ADC manufacturer’s protection circuit.
CF: filter noise and supply sampling charge
Place CF close to the ADC input with a short ADC-ground connection. It filters noise and supplies charging current to the ADC’s internal sampling capacitor. Too little capacitance can allow sampling disturbance or incomplete settling; the reading may be high or low, depending on the previous sample and ADC architecture.
Too much capacitance slows startup and signal changes. Allow recharge between samples and check acquisition timing and channel switching. Select effective capacitance, tolerance, leakage and voltage rating for the actual circuit.
Check the total source resistance and filter response
For a first-order estimate with nominal components and no ADC loading or protection leakage:
Rsource = RS + (Rf × RNTC) / (Rf + RNTC)
τ = Rsource × CF
fc = 1 / (2π × τ)
For the worked example’s 10 kΩ Rf and −20°C to 80°C R–T data, adding RS = 2 kΩ and CF = 100 nF gives τ ≈ 0.312–1.106 ms and fc ≈ 144–511 Hz. These filter estimates do not replace an ADC acquisition and settling check.
The earlier 10 kΩ source-impedance limit checks only the bare divider. Adding RS raises the cold-end nominal total to about 11.06 kΩ. A local capacitor can help sampling, but the chosen ADC must support the complete network. The calculator’s divider Pass result does not validate the added protection and filtering components.
Further reading: TI: How to Select a Surge Diode, Analog Devices: Protecting ADC Inputs, TI: ADC Sampling and Charge-Reservoir Filtering, and Analog Devices: Using Analog Temperature Sensors with ADCs.
Optional buffer and signal conditioning
A unity-gain buffer, also called a voltage follower, can drive another circuit while drawing very little current from the thermistor divider. An ideal follower reproduces the divider voltage, helping prevent an added load from changing the temperature reading. Connect the protected divider signal to the noninverting (+) input, feed the output back to the inverting (−) input, and drive the additional load from the output.
Choose an amplifier that is stable at unity gain. Check its input voltage range, linear output range and load-current capability at the actual supply voltage, temperature and load; a rail-to-rail label does not guarantee accurate operation exactly at either supply rail. Follow the manufacturer’s supply-bypass recommendations.
- Allow for amplifier error. Input bias current, leakage, input offset, drift and noise can affect the reading. Use worst-case specifications over the operating temperature range.
- Check loading and settling. ADC capacitors, cables and long traces can make a buffer ring or oscillate. Follow the amplifier manufacturer’s isolation or compensation guidance and verify settling before sampling.
A gain-and-offset stage changes the signal according to Vout = A × Vdivider + B, where A is gain and B is an offset voltage. It can use more of an ADC’s range, but it also changes the ADC-code-to-temperature relationship. Check that the scaled output stays within the amplifier and ADC ranges across the full temperature and tolerance range. Rebuild the conversion for the complete circuit, or correctly undo the gain and offset before converting. Voltage scaling does not make the thermistor response linear.
This calculator models the direct resistor divider. It does not model buffer loading, amplifier errors, active gain or offset, or amplifier stability. A real unity-gain buffer needs an error allowance; adding gain or offset requires an updated conversion rather than using the direct-divider firmware unchanged.
See Texas Instruments: Buffer (Follower) Circuit and Analog Devices: Avoiding Instability Due to Capacitive Loading for design guidance.
Generate and use thermistor C code
Open C Code & Export after reviewing the circuit. Choose an integer lookup table, floating-point lookup table or floating-point equation. Imported R–T data supports lookup tables; equation export uses the Beta or Steinhart–Hart model.
Choose how to build a thermistor lookup table:
- Automatic: meet an error target adds entries to reduce conversion error. Review the achieved error; the requested target may be too small for the selected output scale.
- Fixed temperature spacing places entries at the requested temperature steps, subject to distinct usable ADC codes.
- Fixed number of entries uses the Lookup table entries setting. That setting does not control the other table-building methods.
Leave export temperatures blank to use the operating range, or enter a smaller interval inside it. Integer output uses degrees C multiplied by 10, 100 or 1000. At scale 100, 2500 means 25.00°C. The optional Fahrenheit function uses the same scale. Extra stored digits do not increase hardware accuracy.
Download C Code saves the conversion functions and a commented integration example. C99 means the 1999 C language standard. The example main() uses a sample ADC code; replace it with a reading from your hardware and enable the example only when needed.
Configure the ADC bit depth, divider supply and reference to match the exported design. Allow the ADC acquisition capacitor to settle, call the conversion routine with a raw ADC code, and check its returned status before using the temperature. Invalid or out-of-range codes alone do not establish an open sensor, short circuit or other wiring fault. Validate the complete assembly at known temperatures.
For lookup-table exports, Lookup Table CSV contains the points used by the C conversion. For equation exports, Model Checkpoints CSV contains nominal reference points for comparison; the equation code does not use them. 1°C Reference CSV provides a separate resistance, voltage and ADC check table at 1°C steps, plus the export endpoints.