Contactor Weld Detection - Battery Design

A contactor that has welded shut will still de-energise its coil normally when commanded open — there is nothing electrically wrong with the coil circuit itself — but the contact tips stay fused together, so the circuit never actually opens. If the weld is strong enough, the return spring cannot break it, and the armature is mechanically prevented from travelling back to the open position at all. This page looks at why that fusing happens and how a battery management system can check for it, rather than simply trusting that a command to open was obeyed. It sits alongside Contactor Arcing, which covers the arc that forms as a healthy contactor interrupts current, and the broader Contactors overview.

Why Contacts Weld

A closed contactor carries current through a comparatively small contact area, and that area heats up in proportion to the current squared. Under normal load this heating is manageable, but a large enough current spike — an inrush surge into an uncharged capacitor bank, a short circuit, or a fault current that arrives faster than upstream fusing can clear it — can heat the contact surfaces to their melting point for a brief instant. If the contacts are pressed together or bouncing against each other while that molten metal is present, they can solidify fused into a single piece once the current subsides. This is a different failure mode from the gradual pitting and material transfer that ordinary arcing causes over many switching cycles: welding can happen in a single high-current event, and it happens without any fault in the coil, spring, or actuation mechanism — those parts try to do exactly what they are supposed to, they simply cannot overcome the weld holding the contacts together.

Why It Cannot Be Detected From the Coil Side

On the coil side, nothing looks wrong: the coil drive de-energises and coil current drops to zero exactly as commanded, because that part of the contactor genuinely has not failed. What has failed is downstream of the coil — the armature is mechanically stuck by the weld rather than free to return — and most contactors give the control system no independent way to see that directly. Safety-rated contactors sometimes use force-guided (mirror) auxiliary contacts, built and certified under IEC 60947-4-1 Annex F so that a welded main contact physically blocks the linked auxiliary contact from indicating ‘closed’ as well, which does give reliable position feedback. An ordinary auxiliary contact is not built or certified to that standard, so it cannot be relied on the same way. Confirming a weld reliably means measuring the electrical state of the power circuit itself, rather than inferring it from the coil or from position feedback that was never certified to catch this specific fault.

Detection Methods

Weld checks generally work by commanding a contactor open and then looking for a voltage that should not be there if the contact had genuinely separated:

Voltage-across-the-contact check: with the contactor commanded open, the system measures the voltage across the two contact terminals. A healthy open contact should show a substantial voltage across it (up to the full bus or supply voltage, depending on what remains connected on the far side); a welded contact instead reads close to zero, because current is still able to flow straight through it.
Bus-to-source comparison: on the charging path, a similar check compares the high-voltage bus voltage against the charger or external supply voltage after the charge contactors have both been commanded open. If the two voltages track each other rather than diverging, current is still passing between them somewhere it should not be able to — evidence of a weld on one of the contactors in that path.
Sequenced isolation testing: where more than one contactor sits in the same current path, opening them one at a time (rather than all together) and re-checking the voltage after each step localises which specific contactor has welded, rather than only confirming that a weld exists somewhere in the path.

These checks are typically run at points where the system already expects the contactors to be open and no drive torque is needed — for example immediately after a charge session ends, or as part of the shutdown sequence before the vehicle is left unattended — rather than continuously in normal driving, since deliberately opening a contactor under load purely to test it would expose it to the same arcing stress the test is trying to guard against.

Safety Implications

A welded contactor defeats the isolation strategy a battery pack relies on: if the main or precharge contactors cannot actually be relied upon to open, neither a driver-initiated shutdown nor a fault response can guarantee the high-voltage circuit is broken. This is why weld detection is treated as a distinct check from ordinary contactor health monitoring such as contact-resistance tracking (see Contactors for how gradual resistance rise is tracked) — a weld can appear suddenly after a single high-current event, rather than developing gradually, so relying only on trend-based degradation monitoring would miss it.

Note: weld-check strategies and thresholds vary by vehicle and BMS supplier implementation; treat the methods above as the general principle rather than a specific product’s exact algorithm.