The negative-delta-V method is the canonical way to terminate a fast NiMH charge, yet it rests on a phenomenon only a few millivolts high: after climbing for most of the charge, the cell's terminal voltage quietly peaks and falls by roughly 5 to 15 millivolts per cell, and a precise charger must catch that tiny reversal against a noisy background. This closing paper of the electrochemistry group explains, from first principles, why the peak exists at all - why a cell that is still receiving constant current should see its voltage decline - and why the effect is smaller, later and more temperature-sensitive in NiMH than in the older nickel-cadmium chemistry for which the method was originally devised.
Terminal voltage under charge equals the positive potential minus the negative potential plus the ohmic drop. During bulk charge all three drift upward as the electrodes traverse their reaction plateaus and concentration polarisation builds, so voltage rises. At full charge the positive is dominated by oxygen evolution, whose potential is temperature-sensitive, while the negative sits on its hydride plateau and is being heated by recombination; the cell is also warming rapidly for the reasons quantified in Paper 4.
Once recombination heating dominates, the electrodes' equilibrium and kinetic potentials fall with rising temperature faster than polarisation continues to rise, so the summed terminal voltage stops climbing, rounds into a peak, and declines slightly even though charge current is unchanged. The -delta-V is therefore a thermal-electrochemical signature: it is the voltage shadow of the recombination heat burst, not an independent electrical event.
Charging guides and charger-IC datasheets place the NiMH full-charge dip at roughly 5 to 15 millivolts per cell after the peak, and Analog Devices observes that this is only about half the magnitude seen in NiCd. NiCd's cadmium negative produces a sharper, larger end-of-charge polarization change; NiMH's hydride electrode and its smoother thermal transition yield a gentler, shallower peak that is correspondingly easier to miss.
A small signal demands careful measurement: adequate resolution, a running peak register that is never reset downward by noise, and confirmation over several consecutive samples to avoid terminating on a transient. The Northeastern solar-charger work illustrates the care required - voltage tracking must ignore moments when current itself has just changed, since a current step moves voltage for ohmic reasons unrelated to full charge.
The -delta-V only forms when recombination is vigorous enough to heat the cell decisively, which requires sufficient current. Below roughly C/3 the oxygen generation and self-heating are too gradual and the voltage curve merely flattens rather than reversing - the well-documented failure of -delta-V at low charge currents, where an analyzer or charger simply never sees a stop signal and risks indefinite overcharge. This is why slow charging relies on timers while fast charging can rely on -delta-V.
Conversely at 1C the peak is larger and arrives earlier in state-of-charge terms but is also more abrupt and accompanied by more heat and pressure, narrowing the margin between a clean detection and an overshoot into abusive overcharge. Charge current and termination threshold are therefore a matched pair, not independent settings.
Because the peak is thermally mediated, a cell that starts warm shows a smaller dip: at elevated temperature the oxygen-evolution potential is already reduced and the additional end-of-charge heating changes the potentials less, so the reversal can shrink to the point of being undetectable - the 'almost non-existent at high temperatures' condition noted in charger literature. A cold-start cell shows the opposite, larger but delayed peak. Chargers consequently pair -delta-V with dT/dt so that the thermal signal covers the warm condition in which the voltage signal fails.
The ST application-note tradition of detecting the voltage-curve inflection point rather than waiting for a full reversal is a direct response to NiMH's shallow dip: terminating at the second derivative change catches full charge earlier, limiting the very heating that would otherwise erase the signal and extending cell life.
Several events masquerade as a full-charge peak. At charge start a cold cell's voltage can dip as it warms; a current fluctuation in a solar or USB-powered charger moves voltage resistively; a poorly matched series string produces per-cell peaks at different moments that blur the pack-level signal; and contact resistance in a dirty holder adds drift. Robust firmware arms detection only after a minimum charge fraction, requires the voltage to have exceeded a threshold, tracks a clean maximum, and demands a sustained decline of a set magnitude.
The first animated figure overlays the voltage peak and the simultaneous temperature inflection, showing their common origin; the second sequences a state machine that arms, tracks the peak, confirms -delta-V over samples, and cross-checks dT/dt before switching to top-off - the architecture realised in silicon by charge-management ICs and dissected in the termination group that follows.
Understanding the peak as a thermal signature sets its correct role: -delta-V is the preferred fast-charge stop under moderate temperatures and adequate current, but it is one channel in a redundant system. A defensible specification states the current window in which -delta-V is valid, the per-cell dip threshold and confirmation time, the warm-start condition under which dT/dt takes precedence, and the timer and absolute-temperature limits that bound every other failure.
For a cell supplier this means characterising the actual peak magnitude and timing of each production grade at 0.5C and 1C across temperature, so charger designers are not using a generic 10-millivolt assumption. The electrochemistry group closes here having explained storage, efficiency, pressure, heat and the voltage peak; the next group takes these observables and builds the complete termination architecture of a production NiMH charger.
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.
