Why Does My Interface Relay Fail? Contact and Coil Faults

Why does an interface relay fail? Three things wear out at different rates: the mechanical contacts, the coil circuit, and the socket connection, and each fails for a different physical reason — contact welding from inrush current, coil burnout from reverse polarity or overvoltage, or a loose push-in terminal that mimics a dead relay. A misdiagnosed relay gets swapped for a new one that fails again within days because the root cause (an undersized contact for the load, a missing freewheel diode, an unretorqued screw terminal) was never fixed. This article covers contact welding, coil faults, chatter and bounce, socket-related false failures, EMR-versus-SSR failure modes, and a field sequence for isolating the fault before you order a replacement.

What Fails First: Contacts, Coil or Socket?

In a panel that has run for a few years, contact wear is the most common failure mode, followed by coil open-circuit, followed by socket or terminal faults that present as a "dead" relay but are not the relay at all. The three failure paths do not share symptoms. A welded contact keeps the load energized after the coil de-energizes. A burnt coil draws no current and the relay never picks up. A loose socket terminal produces an intermittent fault that only shows up when the panel vibrates or heats up. Knowing which category you are in before you pull the relay saves a wasted swap.

Contact Welding: Inrush, DC Loads and Missing Suppression

Contact welding happens when the current at the moment of closing, or the arc energy at the moment of opening, exceeds what the contact material can survive without local melting. Two loads cause most of it in a panel: a contactor coil with a high inrush spike, and a DC inductive load switched without a freewheel diode or snubber across it. A relay rated 6 A AC1 resistive can weld shut on a load with a 10-20x inrush transient that lasts a few milliseconds, because the rated current describes steady-state switching, not the inrush pulse.

DC loads are the harder case. There is no natural current zero-crossing to help extinguish the arc, so breaking capacity on DC falls off sharply as voltage rises — a contact good for 6 A at 24 VDC may only handle a fraction of that at 110 VDC. Skip the diode and the transistor dies; skip the diode on a heavier DC coil and the relay contact welds instead, because the collapsing field has to discharge somewhere and an unsuppressed arc across the contact is where it goes.

That stored energy has to go somewhere in the microseconds after the contact opens. With a freewheel diode or RC snubber across the load, it dissipates as heat in the suppression element. Without one, it dissipates as an arc across the separating contacts, and repeated arcing pits and eventually welds the contact surface. This is why coil-suppression choice belongs to the load side of the relay, not just the relay's own coil — see the coil suppression guide for diode, RC and varistor selection by load type.

Coil Failure: Overvoltage, Reverse Polarity and Open Windings

A coil failure looks simple from outside: apply rated voltage, the relay does not pick up, LED (if present) does not light. The cause is rarely the coil wire itself failing spontaneously — coil power is a few hundred mW to about 1 W, well within the winding's thermal design. What actually burns a coil out is sustained overvoltage (a 24 VDC coil left on a 48 V rail by a wiring error), reverse polarity on a polarized DC coil driving current through a path the design did not intend, or repeated coil-suppression failure feeding transient overvoltage back into the winding on every switch cycle.

Pull-in voltage is roughly 80% of rated coil voltage and drop-out is roughly 10-30%, so a coil running at the edge of its rated band — a "24 VDC" coil actually seeing 18-19 V from a sagging supply — chatters instead of failing outright, which is a different symptom and a different fix (check the supply, not the relay). Full details on coil voltage bands and what happens outside them are in the relay coil voltage guide.

Chatter, Bounce and Contact Wear From an Undersized Relay

Relay contacts bounce for a few milliseconds on every closure — this is mechanical, not a defect — and each bounce event is a tiny make-break-make cycle that arcs and wears the contact surface exactly like a real switching operation. A relay running at or near its rated current sees this wear accumulate faster than one running at half load. Chatter is a different, slower phenomenon: repeated pick-up and drop-out caused by a coil voltage sitting near the drop-out threshold, often from voltage drop on a long control-wire run or a marginal power supply under load.

What we see in the field: a relay swapped three times for "random failures" that turns out to be undersized for the actual inrush of the load it drives — the fix is not a better brand of relay, it is a relay (or contactor) sized for the transient, or a soft-start/snubber on the load side. Contact configuration and current rating by contact set (1 CO through 4 CO) are covered in the contact configuration guide.

Socket and Terminal Faults That Look Like Relay Failure

A meaningful share of "bad relay" call-outs are not the relay at all. Push-in terminals that were never fully seated during installation, screw terminals that back out under vibration, and retaining clips that do not seat the relay fully into the socket all produce an intermittent open circuit that reads exactly like a dead relay on a static continuity check but clears itself when the panel is tapped or reseated. The socket carries its own current and voltage rating, separate from the relay plugged into it, and a socket rated below the relay's contact rating becomes the actual failure point under sustained load, not the relay.

Before condemning a relay that tests intermittently, reseat it fully, check the retaining clip is latched, and re-torque or re-insert the field wiring at the socket terminals. This is a five-minute check against a relay swap that may not fix anything if the socket connection was the real fault.

EMR vs SSR: Different Parts Wear Out

Electromechanical relays fail through the mechanisms above — contact wear, welding, coil burnout, mechanical fatigue of the armature spring after enough operations. Solid-state interface relays have no moving contacts, so they do not weld or bounce, but they fail differently: leakage current through the opto-isolator output stage even in the "off" state, thermal failure of the output semiconductor if heat-sinking was undersized for the switched current, and complete failure short-circuited (load stays on) rather than open, which is the opposite failure mode from a welded EMR contact in terms of what it does to the downstream equipment but has the same practical symptom — the load will not turn off. Match relay type to switching frequency and load before failure analysis, not after; see the EMR vs SSR comparison for where each is the right pick.

A Field Troubleshooting Sequence

Work in this order rather than jumping straight to a relay swap. First, confirm what actually failed: load stuck on (contact weld or SSR short) or load never energizes (coil, socket or upstream signal). Second, if the load never energizes, measure actual voltage at the coil terminals under load, not just at the PLC output card, since voltage drop along a long control run or a socket with a failing terminal can starve the coil even when the PLC output is healthy. Third, check the socket retaining clip and terminal torque before touching the relay itself. Fourth, if it is a contact fault, look at what the relay is switching — an inductive DC load without suppression, or a lamp/capacitive inrush without a limiting resistor, points straight at the cause rather than the part. This sequence maps onto how a PLC interface relay sits between the control side and the load side — isolating which side of that boundary the fault is on is most of the diagnosis.

Conclusion

Contact welding, coil burnout, chatter, and socket-related false failures each have a distinct signature and a distinct fix. Diagnosing which one you have before pulling the relay — load stuck on vs. never energizing, static continuity fault vs. intermittent under vibration — turns a repeat failure into a one-time fix instead of a recurring parts swap.