If you are wiring a 3D printer hotend, a DIY reflow oven, or an ESP32-based environmental monitor, you are likely using a 10K NTC thermistor with a Beta (B) value of 3950. At exactly 25°C (77°F), this component yields a baseline resistance of 10,000 Ω. However, because NTC (Negative Temperature Coefficient) thermistors respond non-linearly to heat, guessing intermediate values will ruin your PID tuning or trigger false thermal runaway faults. Below is the definitive reference data for the 10K 3950 profile.
How to Read This NTC Thermistor Resistance Chart
Before scrolling to the numbers, understand what the columns actually mean for your microcontroller setup. The Nominal Resistance column represents the mathematical ideal based on the Steinhart-Hart equation and the manufacturer's stated B25/85 Beta parameter. This is the column you use to generate lookup tables for firmware like Marlin or Klipper. The Min/Max Tolerance columns represent the physical manufacturing bounds (typically ±1% at 25°C, widening to ±3% or more at temperature extremes). These bounds define the hardware fault-detection thresholds in your code.
| Temp (°C) | Temp (°F) | Nominal Resistance (Ω) | Min Tolerance Bound (Ω) | Max Tolerance Bound (Ω) |
|---|---|---|---|---|
| -20 | -4 | 70,620 | 67,089 | 74,151 |
| -10 | 14 | 42,510 | 40,384 | 44,636 |
| 0 | 32 | 27,219 | 25,858 | 28,580 |
| 10 | 50 | 18,085 | 17,180 | 18,990 |
| 20 | 68 | 12,491 | 11,866 | 13,116 |
| 25 | 77 | 10,000 | 9,900 | 10,100 |
| 30 | 86 | 8,057 | 7,815 | 8,299 |
| 37 | 98.6 | 6,015 | 5,774 | 6,256 |
| 40 | 104 | 5,326 | 5,113 | 5,539 |
| 50 | 122 | 3,602 | 3,422 | 3,782 |
| 60 | 140 | 2,471 | 2,322 | 2,620 |
| 80 | 176 | 1,235 | 1,148 | 1,322 |
| 100 | 212 | 678 | 623 | 733 |
| 120 | 248 | 402 | 365 | 439 |
| 150 | 302 | 198 | 178 | 218 |
Applying Derating and Tolerance to Your Installation
A common mistake on the bench is assuming the nominal resistance is the only number that matters. Which column applies to your installation depends entirely on whether you are writing control logic or safety logic.
For control logic (like mapping an ADC voltage to a temperature readout on an LCD), use the Nominal column. Microcontrollers use this ideal curve to calculate the Steinhart-Hart coefficients. However, for safety logic (like a MINTEMP or MAXTEMP hardware interrupt), you must use the Min/Max bounds. If your firmware expects exactly 678 Ω at 100°C, but your specific thermistor reads 720 Ω due to standard manufacturing tolerance, a rigid safety threshold will trigger a false thermal runaway shutdown. Always program your fault thresholds using the Min/Max columns, widened by an additional 2% to account for ADC reference voltage drift.
How Power Derating Modifies the Base Value
Thermistors are not passive; they require current to be read, which generates internal heat. Manufacturers specify a maximum power dissipation (often 50mW at 25°C). As ambient temperature rises, the thermistor's ability to shed heat drops. According to standard Vishay NTCLE100 datasheets, power derating begins linearly above 25°C, reaching 0% allowable dissipation at the component's maximum rated temperature (usually 125°C or 150°C for glass-encapsulated beads).
If you are reading a thermistor at 100°C, you cannot push the same current through it as you do at room temperature. If your microcontroller's internal pull-up resistor is too low (e.g., 1K instead of 4.7K or 10K), the increased current at high temperatures will cause self-heating. The thermistor will read a lower resistance than the chart dictates, making the system think it is hotter than it actually is. Always use a voltage divider with a high-value series resistor (4.7K to 10K) to keep measurement current below 100µA.
What the Chart Cannot Tell You
A resistance chart is a static electrical model. It completely ignores the physical reality of thermal transfer. When debugging a sluggish or erratic temperature reading, the chart will not help you with these three physical limitations:
- Thermal Time Constant ($\tau$): The chart tells you what the resistance should be at 100°C, but not how long it takes to get there. A bare 10K 3950 bead might have a time constant of 1 second in still air, while the same die potted in a brass 3D printer heater block with thermal paste might take 15 seconds to stabilize. If your PID loop derivative (D) term is too aggressive, this physical lag will cause massive temperature oscillations.
- Thermal Coupling and Gradients: If you measure the resistance of a thermistor embedded in a PCB trace, the chart assumes the thermistor die is at the exact same temperature as the copper. In reality, the epoxy body and the lead wires act as heat sinks. The die might be 2°C cooler than the ambient air you are trying to measure.
- ADC Resolution Limits: At 150°C, the resistance drops to 198 Ω. On a standard 10-bit ADC with a 5V reference and a 10K pull-up, the voltage change between 140°C and 150°C is only a few millivolts—often less than the noise floor of the ADC. The chart implies infinite precision, but your hardware limits effective resolution at temperature extremes.
Thermistor Resistance Chart FAQ
How do I calculate thermistor resistance for temperatures not on the chart?
You must use the Steinhart-Hart equation or the Beta parameter equation. For a standard 10K 3950 NTC, the Beta equation is sufficiently accurate between 0°C and 100°C. The formula is $R = R_0 \cdot e^{B \cdot (1/T - 1/T_0)}$, where $T$ and $T_0$ are in Kelvin. For high-precision applications spanning wider ranges (like cryogenic or high-heat reflow profiling), use the full three-coefficient Steinhart-Hart equation. You can extract the exact A, B, and C coefficients for your specific batch using a Steinhart-Hart calculator and three measured data points.
Why does my multimeter read a different resistance than the 25°C chart value?
If your multimeter reads 10.4K Ω when your room thermometer reads 25°C, you are likely experiencing one of three issues. First, standard 10K 3950 thermistors have a ±1% to ±2% tolerance at 25°C, meaning 10.2K is perfectly normal. Second, your multimeter's test leads have their own resistance (often 0.2 to 0.5 Ω), though this is negligible for a 10K sensor. Third, and most commonly, the thermistor is suffering from self-heating or thermal gradient errors. If you are holding the thermistor in your fingers, or if it is sitting near a warm power supply, the die temperature will not match the ambient room temperature. Suspend the thermistor in free air away from drafts and heat sources for 5 minutes before taking a baseline reading.
Can I use a 10K 3950 thermistor chart for a 10K 3435 thermistor?
No. While both thermistors read exactly 10,000 Ω at 25°C, the Beta (B) value dictates the slope of the resistance curve. A 3950 Beta value means the resistance drops much faster as temperature rises compared to a 3435 Beta value. At 100°C, a 10K 3950 reads roughly 678 Ω, while a 10K 3435 reads closer to 900 Ω. If you use a 3950 lookup table in your Marlin firmware for a 3435 physical sensor, your hotend will severely overheat because the firmware will misinterpret the higher resistance as a lower temperature. Always verify the B-value printed on the component packaging or datasheet before flashing your configuration.
Does wire length affect the resistance chart accuracy?
For a 10K NTC thermistor, wire length is rarely an issue because the sensor's base resistance is so high. Adding 2 meters of 22 AWG copper wire introduces roughly 0.1 Ω of series resistance, which is entirely invisible against a 10,000 Ω baseline. However, if you are using a low-resistance thermistor (like a 100 Ω PT100 RTD, which is often confused with NTC thermistors by beginners), wire resistance will completely invalidate the chart. For standard 10K NTCs, focus on shielding the wires from EMI rather than worrying about voltage drop.
