A decade resistance box can generate the resistance values required to reproduce an NTC thermistor curve for controller testing, provided its range, resolution, and measurement accuracy match the sensor circuit.
The simulation covers the NTC’s decreasing resistance-temperature response, from the sensor’s nominal resistance through its selected temperature endpoints, with the box acting as a controllable substitute rather than a thermal sensor.
NTC Simulation Scope and Resistance Behavior
An NTC thermistor is a temperature-sensitive resistor whose resistance decreases as its body temperature rises. Ceramic NTC devices are commonly produced from metal oxides, including manganese, nickel, cobalt, iron, copper, and titanium, with the mixture, proportions, sintering atmosphere, and sintering temperature determining resistivity and temperature-coefficient characteristics [S5].
The relevant simulation target is not a general-purpose temperature sweep but the resistance presented to the equipment under test. For a given NTC model, the decade box must cover the sensor’s cold and hot resistance values, including the resistance at each simulated temperature point.
A decade box is therefore suitable for functional testing of an input card, thermostat, boiler controller, transmitter, or alarm system that expects an NTC resistance. It is not a temperature chamber and does not reproduce thermal lag, self-heating, contact resistance, moisture ingress, or the behavior of a damaged thermistor. Those effects require a real sensor or a more complete analog network.
Choosing the Box for NTC Curve Work
Selection should begin with the NTC’s specified resistance-temperature curve, not with the box’s maximum resistance. The box must cover the complete resistance range required by the chosen temperature interval with useful resolution between adjacent values. A box whose nominal range is broad but whose dial steps are too coarse will introduce avoidable conversion and controller errors. [S3]
The decade resistance box is the relevant instrument for manually or remotely substituted resistance. Resistance standards and decade steps determine the smallest selectable increment and the quality of the simulation, while the NTC’s tolerance determines how closely the simulated curve can match the physical sensor. The thermistor’s resistance tolerance, conversion error, excitation current, self-heating, electronics, and thermal coupling all contribute to the final measurement error [S3].
Three selection criteria matter. First, the usable resistance range must exceed the target NTC values without being so wide that the increments become too coarse. Second, the box resistance tolerance and stability must be compatible with the controller’s required accuracy. Third, the connection method must reproduce the intended circuit, including any pull-up, pull-down, excitation source, cable resistance, and input topology.
Generating Resistance Values for the Curve

The beta equation can approximate an NTC resistance-temperature curve over a limited range. Higher-accuracy conversion uses the manufacturer’s resistance table or suitable Steinhart-Hart coefficients [S3]. A decade box should therefore use the manufacturer’s table whenever available, rather than relying on a single beta value across the complete range.
For each selected temperature, calculate or look up the corresponding NTC resistance, then set the decade box to that value. The simulated voltage is then determined by the complete measuring circuit, not by the resistance box alone. In a constant-current arrangement, voltage changes with resistance; in a divider arrangement, the controller measures the divider ratio formed by the sensor and its fixed reference component.
SPICE users have the same modeling decision in a different environment. One documented approach treats the thermistor as a resistor with tc1 and tc2 temperature coefficients [S6]. Another discussion shows why a fitted or measured model is needed when resistance must change with rising temperature [S2]. Those software methods help define the test stimulus, but the physical decade box is generally easier for checking an input channel’s resistance-to-temperature interpretation.
Verifying the Simulated Sensor Circuit
Verification starts by measuring the resistance actually present at the controller input, not only at the box terminals. Cable resistance, relay contacts, connectors, and contamination can shift the effective sensor value, particularly at the lower end of the resistance range.
For a simple divider with a known reference resistance, the relation is Vout divided by Vin equals Rsensor divided by the sum of Rsensor and Rreference. The resistance inferred by the controller can then be compared with the selected decade value. Any discrepancy can come from meter accuracy, source loading, input bias current, wiring resistance, or an incorrect NTC conversion model.
Thermal excitation must also be controlled. NTC thermistors can change resistance because of their own dissipated power, and an overly energetic test signal can move the simulated curve away from the intended table value. Keep the excitation within the sensor circuit’s normal operating conditions and allow the temperature to stabilize before recording the result.
For calibration work, the deadweight tester is not part of this test method, but the same discipline applies: compare the applied quantity with the measured response and record the residual error rather than relying on the nominal setpoint.
Accuracy Limits and Failure Modes

The principal error is mismatch between the selected box resistance and the resistance required by the actual NTC at the target temperature. Additional errors include decade-box tolerance, switch contact resistance, wiring resistance, excitation-induced self-heating, and the controller’s resistance-to-temperature conversion. The dry-block temperature calibrator becomes relevant only when the physical NTC and its process temperature must be checked under controlled heat. [S3]
Resistance tolerance alone does not describe the complete simulation uncertainty. An NTC sensor may also deviate from its stated curve because of component tolerance, temperature gradients, mechanical stress, aging, or inadequate thermal contact. A simulation based on a nominal curve therefore tests the expected nominal response, not the complete distribution of real sensor behavior [S3].
Fast transitions expose another limitation. A mechanical decade box changes resistance in discrete steps, so it cannot reproduce a continuously falling curve unless the controller test can tolerate the selected increments or the resistance is interpolated by another circuit. The step size should be small enough that each change produces a temperature interpretation consistent with the required controller resolution.
Electrical isolation also matters. A decade box is normally a passive resistive element and must be inserted into the intended circuit without creating an unintended ground path. If the controller uses an isolated input, verify that the test arrangement does not bypass that isolation. If the sensor input is powered or includes active circuitry, a passive resistance box may not be an adequate substitute.
Test Procedure for Controller Validation
A controlled test can proceed through six steps: identify the NTC model and resistance-temperature table, choose the temperature endpoints, calculate one resistance per point, confirm the decade box range and resolution, apply each resistance at the controller input, and record the interpreted temperature. The expected result should include resistance, applied voltage, interpreted temperature, and the difference from the target temperature. [S4]
Start near the nominal resistance and check the controller’s zero or reference behavior before moving toward either temperature limit. Then test several points across the range, including the endpoints. Repetition at the same point reveals switch repeatability and wiring noise, while reversing the test order can expose thermal drift or history effects in the controller electronics.
Use a multifunction process calibrator when it can source or measure the relevant electrical signal alongside the decade box. If the equipment requires a calibrated loop measurement, a loop calibrator can provide the appropriate process-signal reference, provided its electrical interface matches the sensor circuit.
For a boiler-controller example, two resistance states may suffice if the controller merely compares outdoor temperature against a thermostat setting.
The box should also be checked against the intended excitation and input range before every campaign. Record the selected resistance, measured resistance at the controller terminal, source voltage, controller display, interpreted temperature, and deviation. That record distinguishes a decade-box or wiring error from an incorrect NTC table or a controller conversion fault.