A charge controller decides on millivolts and fractions of a degree, so its measurements must be more accurate than the phenomena it watches. Poor current sensing corrupts coulomb counting and current-mode control; a badly placed or poorly conditioned thermistor turns dT/dt into noise; and a connector's contact resistance, mistaken for cell behaviour, causes false termination. This paper designs the two measurement chains every NiMH charger relies on - current sensing and NTC temperature sensing - covering component selection, connection topology, filtering and fault detection, and shows how measurement integrity propagates directly into termination reliability and charge accuracy.
Charge current is most commonly measured as the voltage across a low-value series sense resistor, chosen to balance measurement resolution against I-squared-R dissipation: large enough that the amplified signal clears amplifier offset and ADC noise, small enough not to waste power or add heat. A precision, low-temperature-coefficient alloy resistor keeps the reading stable as the charger warms, since self-heating drift would otherwise look like changing current.
The shunt sits in the power path whose current the controller commands (battery-side for true battery current), and its voltage is amplified by a current-sense amplifier or a dedicated differential ADC channel before digitisation; the resulting bandwidth must exceed the pulse/PWM content so pulse and multi-stage profiles are measured truthfully.
Contact and trace resistance must never appear in the measurement: the sense voltage is taken with a separate Kelvin pair directly at the shunt pads, and battery voltage for -delta-V is sensed at the cell terminals or pack connector, not at the regulator output, so holder contact resistance is excluded. In consumer holders with spring contacts, failing to Kelvin-sense lets contact drift masquerade as rising internal resistance or a voltage peak - the false-trip mechanism flagged in Paper 6.
Good practice uses a star ground for signal and power returns, separates the sense traces from the switching power loop (Paper 36), and guards against PWM common-mode transients that would otherwise alias into the low-frequency charge measurement.
A negative-temperature-coefficient thermistor converts cell temperature to resistance, usually in a divider excited from a stable reference; its nominal resistance (commonly 10 k at 25 C) and beta match the ADC range over the cold-to-hot charge window, and its dissipation constant must keep self-heating negligible by limiting excitation current or duty-cycling the divider. Precision-grade NTCs with tight beta tolerance are warranted because dT/dt differentiation amplifies every error.
The divider is designed so cold, hot and open/short conditions produce distinguishable voltages: an open NTC reads as an extreme that a robust design treats as 'do not fast-charge' rather than misreading as cold, and a shorted NTC likewise faults - the sensor-integrity gate required by Paper 12.
The thermistor must be thermally coupled to the cell - clipped to a cell can, bonded to the pack's hottest interior cell, or integrated in the holder contact - and thermally isolated from the charger's own switcher and sense-resistor heat, otherwise the charger measures itself. In multi-cell packs the sensor covers the geometrically hottest cell (Paper 28), and multiple NTCs are used where gradients are significant; mechanical spring force and thermally conductive coupling ensure the reading tracks the cell rather than the air gap.
Placement is validated by comparing the NTC reading to instrumented thermocouples on the cells during a worst-case charge, confirming lag and offset are small enough that the 1 C/min dT/dt decision remains accurate.
Both chains are filtered to the bandwidth of the underlying phenomenon - voltage for -delta-V is lightly filtered to preserve the millivolt peak while rejecting switching ripple; temperature is filtered over a sliding window before differentiation so slope reflects recombination rather than noise. Sampling is synchronized away from PWM edges, and the controller requires multi-sample confirmation (Paper 6/7). Calibration trims amplifier offset and NTC offset at production, turning component tolerances into measured accuracy.
The first figure diagrams the current and temperature measurement chains from physical quantity to decision; the second shows how placement error and noise degrade an otherwise clean dT/dt signal, motivating the conditioning design.
Verification injects known currents and temperatures, tests open/short sensor faults, confirms rejection of PWM ripple and contact-resistance steps, and validates -delta-V/dT/dt decisions against reference instrumentation across temperature. Weijiang provides cell thermal time constants and surface-temperature behaviour to help place and validate NTCs. The next paper turns from measurement accuracy to the energy it costs - charger efficiency, standby power and energy-efficiency regulation.
Weijiang Power designs and manufactures nickel-metal hydride cells, matched packs and charging-ready configurations for consumer, industrial, medical and mobility customers, and supports partners with charge-protocol guidance, IEC 61951-2 performance files, IEC 62133-1 safety evidence and charger co-validation. Share your cell format, charge rate, thermal envelope and cycle target and our engineers will specify a cell-and-charge combination that protects both runtime and service life. Review the range on the products page.
