Frequently Asked Questions

How NTC thermistors work, how they are tested and how to choose between them and other sensors — plus the commercial questions buyers ask us most. Answers are written by the engineers who make the parts.

Technical

How NTC thermistors work — and how to use them

The five technical questions we are asked most, answered in full detail.

NTC stands for negative temperature coefficient: as the part gets warmer, its resistance falls. The relationship is not a straight line but an exponential curve — and that steepness is exactly why a thermistor resolves a tenth of a degree with an ordinary 10-bit ADC.

Why the resistance falls. The sensing element is a sintered ceramic, a spinel-structure oxide of manganese, nickel, cobalt and copper. Current flows by electrons hopping between Mn³⁺ and Mn⁴⁺ sites in the lattice. Warming the ceramic does two things at once: it frees more charge carriers and makes each hop more likely. The result is resistance that drops close to exponentially with temperature — a sensitivity of about −4 %/°C at 25 °C, where a platinum RTD manages roughly +0.39 %/°C.

The curve in one formula. R(T) = R25 · exp[ B · (1/T − 1/298.15) ], with T in kelvin and B the material constant taken from two calibration points: B = ln(R1/R2) / (1/T1 − 1/T2). A typical chip is coded 103F3950 — 10 kΩ at 25 °C, ±1 % tolerance, B = 3950 K. At room temperature that part loses about 440 Ω for every degree it warms.

  • R25 and tolerance: ±1 % (F), ±2 % (G), ±3 % (H) or ±5 % (J). One code letter, two very different prices.
  • B value: 3435 K and 3950 K are the volume values; a wide B tolerance shows up as spread once the part leaves 25 °C.
  • Dissipation factor δ (mW/K): how much power it takes to warm the body by one degree — a few mW/K for a small bead.
  • Thermal time constant τ: seconds in still air, far less in a stirred liquid or clamped to metal.
  • Operating range: −40 °C to 105–125 °C on the standard enamel-wire build (MF52B), and up to 200–250 °C on the MF52C high-temperature version.

Measure it cold, not hot. Never measure resistance with a current high enough to heat the part. Self-heating biases the reading downward, so catalogue values are defined under zero-power conditions: keep dissipation under roughly 0.1 mW, which for a 10 kΩ part means a test current in the tens of microamps.

NTC and PTC are not the same device. A PTC (positive temperature coefficient) part — barium-titanate ceramic, or the polymer PPTC sold as a resettable fuse — holds a low resistance until a switching temperature, then jumps by orders of magnitude. That jump makes it a protector, not a thermometer. The two are mixed up constantly in supplier literature: the resettable fuse protecting a lithium battery is a PPTC, not an NTC.

Where they are used: battery pack and BMS temperature sensing, charger and appliance overheat protection, automotive harness and connector temperature sampling, cold-junction compensation for thermocouples, and self-heated liquid-level or airflow sensing.

Two questions sit behind this one: how the parts are tested in our factory, and how you verify a delivery on your own bench.

In-house. Every assembly gets a 100 % continuity and R25 check against the ordered tolerance band, plus a visual inspection of the bead, the lead exit and the tinned length. B value, dielectric strength, pull force and thermal response follow a sampling plan per lot, and the lot code ties the test report back to the production record.

On receipt. Four checks catch nearly everything:

  • Room-temperature point. With the part at 25 °C ±1 °C, measure across the two leads on a resistance range. A 10 kΩ ±1 % part must read 9.90–10.10 kΩ; a ±3 % part, 9.70–10.30 kΩ.
  • Second and third point. Ice-water (0 °C) and boiling water give additional points — solve for B and compare with the datasheet. Remember that the boiling point falls with altitude.
  • Wait for the reading to settle. A small bead still needs seconds in still air, and holding the leads between your fingers shifts the value upward.
  • Null out the test leads on low-resistance parts, or use a four-wire method. A few ohms of lead is irrelevant to 10 kΩ but fatal to a 100 Ω part.

Judge by percentage, not by a fixed window. “Within ±2 Ω” only makes sense for one nominal value: ±2 Ω is 0.02 % of 10 kΩ but 2 % of a 100 Ω part. Work from the tolerance code instead.

Incoming inspection. For larger orders we can supply the R25 distribution — mean, minimum and maximum per lot — so your AQL sample is based on real numbers rather than a guess.

All three measure temperature. They differ in physics, in where they are accurate, and in what they cost.

An NTC is a ceramic whose resistance falls exponentially. A platinum RTD (Pt100, Pt1000) is a metal whose resistance rises almost linearly. A thermocouple generates a small voltage at the junction of two dissimilar metals — the Seebeck effect — and needs cold-junction compensation before the reading means anything.

Sensor comparison at a glance
SensorSensitivityTypical rangeAccuracy you can expectRelative cost
NTC thermistor≈ −4 %/°C at 25 °C−40 to 150 °C±0.1–0.5 °C after calibrationLowest
Platinum RTD (Pt100 / Pt1000)≈ +0.39 %/°C, near-linear−200 to 600 °C±0.1 °C or betterHighest
Thermocouple (K, J, T)≈ 41 µV/°C (type K)−200 to 1300 °C±1–2 °C before cold-junction compensationModerate

Linearising an NTC. The exponential curve can be handled three ways: the B-value equation for a quick estimate, the Steinhart–Hart polynomial 1/T = A + B·ln R + C·(ln R)³ for a proper fit across the range, or a lookup table with piecewise interpolation in firmware. If you need an analogue output, a series and parallel resistor across the NTC flattens the curve over a window such as 25–50 °C, at the cost of sensitivity outside it.

Lead length decides more than you think. In a two-wire harness, a Pt100 lets the copper lead resistance add to the measurement — which is why three- and four-wire RTDs exist. A 10 kΩ NTC with 120 mm of enamel wire does not care: 0.5 Ω of lead is 0.005 % of the reading. For long harness runs and battery packs that difference is decisive, and it is why NTC leads in those applications.

Rule of thumb. Narrow range, small size, fast response and low cost → NTC. Absolute accuracy, laboratory or process duty, long-term stability → platinum RTD. Above 300 °C, on exhausts or in a kiln → thermocouple.

The negative temperature coefficient of semiconducting materials was noticed in the nineteenth century — Faraday recorded it in silver sulphide in 1833 — but for a hundred years the effect stayed a laboratory curiosity because the materials were not stable enough to use.

The device became an industrial component in the 1950s, when the manganese–nickel–cobalt spinel ceramic system was worked out and could be sintered reproducibly. Since then the story has been one of materials and process control rather than new physics.

  • Materials. Mn₃O₄–NiO–Co₂O₃ spinel bodies set the resistance level and the B value; copper and iron dopants shift the curve; glass and epoxy encapsulation extended humidity and temperature life.
  • Form factor. Pressed discs and rods gave way to beads on leads, then to SMD chips and thin films — smaller sensing elements with shorter thermal time constants.
  • Manufacturing. Hand-soldered leads were replaced by tape-cast wafers, automatic dicing, machine screening, auto-tinning and enamel coating. That is the process we still run, in 13 steps.
  • Today. Battery management and automotive programmes pushed the requirements to lot-to-lot consistency, AEC-Q200-style qualification and full traceability, under an IATF 16949 quality system.

What we build sits in the mainstream of that history: MF52B enamel-wire and MF52C high-temperature bead thermistors, 10K / 50K / 100K Ω, B value 3435 / 3950, in lead lengths from 30 mm to 120 mm — and the wire-harness assemblies built around them.

Every MF52 part on our line follows the same route. Each step is a place where tolerance is either won or lost.

  • 1. Raw material. Oxide powders are weighed and blended to the target resistance and B value.
  • 2. Casting and wafer dicing. The blended ceramic is cast, sintered and sliced into wafers.
  • 3. Chip forming. Wafers become individual chips with electrodes.
  • 4. Lead cutting. Lead wire is cut to the drawing length.
  • 5. Automated cutting. Batch cutting with dimensional control.
  • 6. Manual screening. Chips are sorted by measured value — this is where ±1 % grades come from.
  • 7. Fixture forming. Leads are formed in a jig so the bead position repeats part to part.
  • 8. Auto-tinning. The tinned length, typically 2–5 mm, is set at this step.
  • 9. Enamel coating. The lead is insulated for board and harness routing.
  • 10. Machine per wire. One machine places and bonds leads per wire, keeping bead geometry consistent.
  • 11. Inspection. 100 % R25 measurement and visual check, with B value and pull force on sampling.
  • 12. Crimping. Terminals are crimped to the harness end; crimp height and pull force are the controls.
  • 13. Custom build. Connectors, housings, sleeving and packaging to your drawing, then final QA and packing.

Sintering sets the B value, screening sets the tolerance grade, and crimping sets the mechanical reliability of the assembly. Those three steps are why two suppliers can quote the same 10K B3950 part and behave very differently five years later in the field.

Ordering

Ordering, samples and delivery

The commercial questions, answered the way we actually work.

Samples are typically ready in 3–5 working days after the drawing and materials are confirmed. If a custom mould is required, add 7–10 days for tooling.

Yes. We build from your drawing, BOM, 3D model or a physical sample. For cross-referenced parts we confirm the connector and terminal specification before quoting so there is no substitution risk.

Our factory operates to ISO 9001 with IATF 16949 capability for automotive programmes. Materials are UL, CE, FCC and RoHS compliant, and test reports are supplied with the shipment.

Every assembly receives 100% continuity and resistance testing plus a visual inspection. Pull-force and temperature response testing are applied where the specification requires it, with records retained against the production lot.

For new customers we typically work on 30% deposit with the balance against the bill of lading. Repeat customers can move to open account terms after a review period.

We ship by express, air or sea depending on volume, and supply commercial invoice, packing list and certificate of origin. Freight forwarder details can be provided for your own booking.

Yes. Inner bag labels, carton marks, barcodes and custom packaging are all supported. Send your label template with the order and we will match it.

Contact your account engineer immediately. We will confirm the defect with your documentation, and either replace the affected units or issue credit, depending on the finding from the root-cause analysis.

Still have a question?

Our engineering team replies to technical questions within one working day.

WhatsAppOpen chat