Conversion methods¶

How the driver turns a voltage into a temperature, which equation each method uses and how accurate each is for the NTCG163JF103FT1S.

The pipeline¶

flowchart LR
    A["ADC: V_NODE"] --> B["Divider: R = R_SERIES · V / (V_REF − V)"]
    B --> C["Beta, or the datasheet table (LookupTable)"]
    C --> D["+ calibration_offset"]
    D --> E["EMA filter (optional)"]
    E --> F["min/max check"]

Every reading follows these steps in ReadTemperatureCelsius(): average sample_count voltages, solve the divider, convert, add the calibration offset, filter, and check the result against min_temperature and max_temperature.

Beta equation (what the driver uses)¶

1/T = 1/T25 + ln(R / R25) / B          T and T25 in kelvin, T25 = 298.15 K

NTC::ConvertResistanceToTemperatureBeta() implements it, and the driver calls it for NtcConversionMethod::Mathematical, for Auto (the default) and whenever the lookup table does not cover a reading. B is a two-point fit: the datasheet's B25/85 makes the equation exact at 25 °C and 85 °C and leaves an error elsewhere. For the NTCG163JF103FT1S (R25 = 10 kΩ, B25/85 = 3435 K), against the spec sheet's nominal resistance:

Temperature (°C)	Beta reads (°C)	Error (°C)
−40	−35.56	+4.44
−20	−17.47	+2.53
0	1.11	+1.11
25	25.00	0.00
50	49.58	−0.42
85	85.01	+0.01
100	100.50	+0.50
125	126.68	+1.68
150	153.19	+3.19

Derived from Table 3 of the spec sheet; the host tests check the 0 °C, 25 °C, 85 °C, 125 °C and −40 °C rows. Between 10 °C and 95 °C the error stays under 0.6 °C. Outside it, use the lookup table or the Steinhart-Hart equation below, or a calibration offset near the temperature that matters to you.

Using B25/50 instead (3382 K on TDK's product page for the NTCG163JF103FT1) moves the exact points to 25 °C and 50 °C and shifts the error to higher temperatures. Use the B your datasheet gives for the range you care about.

Steinhart-Hart equation¶

1/T = A + B · ln(R) + C · ln(R)³

Three coefficients fitted at three temperatures follow the curve far better than Beta. NTC::CalculateSteinhartHartCoefficients() fits them through three (temperature, resistance) points, and NTC::ConvertResistanceToTemperatureSteinhartHart() applies them. Fitted to the spec sheet at −40 °C, 25 °C and 125 °C (A = 0.8767633 × 10⁻³, B = 2.527899 × 10⁻⁴, C = 1.906681 × 10⁻⁷), the equation stays within 0.2 °C of the nominal table from −40 °C to +125 °C and within 0.41 °C at +150 °C (derived; the host tests check every row):

// Fit A, B and C through three rows of the datasheet table (-40, 25 and 125 C) ...
float a = 0.0F;
float b = 0.0F;
float c = 0.0F;
if (!NTC::CalculateSteinhartHartCoefficients(-40.0F, 188500.0F, 25.0F, 10000.0F, 125.0F, 534.0F,
                                             &a, &b, &c)) {
  return false;
}
// ... and apply them to the resistance the driver measures.
float ohms = 0.0F;
return ntc.GetResistance(&ohms) == NtcError::Success &&
       NTC::ConvertResistanceToTemperatureSteinhartHart(ohms, a, b, c, &celsius);
// For the NTCG163JF103FT1S, NTC::ConvertResistanceToTemperatureDefault(ohms, &celsius) uses this fit.

NTC::ConvertResistanceToTemperatureDefault() and NTC::ConvertTemperatureToResistanceDefault() use that fit, so they suit the NTCG163JF103FT1S only; pass your own coefficients for another part. The temperature to resistance direction solves the cubic exactly, so it inverts the equation. The driver's read path does not call Steinhart-Hart: read the resistance with GetResistance() and apply it, as above, or use LookupTable.

Lookup table¶

NtcConversionMethod::LookupTable searches a per-type table and interpolates. The three TDK types share the NTCG163JF103FT1S table in src/ntc_lookup_table.cpp: the nominal column of Table 3 of the spec sheet, in ohms, from −40 °C (188.5 kΩ) to +150 °C (313 Ω) in 5 °C steps. The resistance falls as the temperature rises, and the search brackets the reading between two rows. Between rows the driver interpolates the temperature linearly in ln(R), which follows the curve to well under 0.1 °C over one 5 °C step.

At every row the table returns the datasheet temperature exactly, so it removes the Beta equation's error at the ends of the range (+4.4 °C at −40 °C). A reading outside the table (colder than −40 °C or hotter than +150 °C) falls back to the Beta equation. Custom has no table and always uses Beta.

Other conversions¶

Function	Does
`NTC::ConvertTemperatureToResistanceBeta()`	Beta, the other way: R at a temperature
`NTC::CalculateBetaValue()`	B from two (temperature, resistance) points
`NTC::CalculateThermistorResistance()`	The divider equation
`NTC::CalculateThermistorVoltage()`	The divider, the other way: V_NODE for a resistance
`NtcThermistor<>::CelsiusToFahrenheit()` and friends	Unit conversions

// B25/85 from two datasheet points: 10.00 kOhm at 25 C, 1.451 kOhm at 85 C.
return NTC::CalculateBetaValue(25.0F, 10000.0F, 85.0F, 1451.0F, &beta);  // beta ~= 3435 K

Next¶

API guide shows the calls by task; Configuration lists every setting.