The Relay That Welded Its Contacts: Sizing a Relay for an Inductive Load
A 24 V DC relay switching a solenoid valve coil. The relay closed, the valve clicked, the relay contacts welded shut. The valve stayed energized. The machine kept cycling after the stop button was pressed. Emergency stop cut the main power, but the relay contacts were stuck. We pulled the relay out. The contacts were fused — a little bead of copper between the terminals. This is what happens when a relay is undersized for an inductive load.
Why inductive loads kill relays
When a relay opens, the current through the contact tries to stop. For a resistive load (a lamp, a heater), the current stops cleanly. For an inductive load (a solenoid coil, a motor winding), the inductor fights the current change. The coil generates a back-EMF voltage — up to hundreds of volts — across the opening contacts. The voltage ionizes the air between the contacts, creating an arc. The arc burns the contact surface. Over many cycles, the contacts pit. Eventually, they weld shut.
The solenoid coil on our valve was 10 W, 24 V DC. The steady current was 0.42 A. The relay was rated for 5 A resistive. That seems fine. But at the moment of opening, the inductive kick created a 300 V arc. The relay’s contact rating for inductive loads is much lower — typically 10–20% of the resistive rating. A 5 A resistive relay is about 0.5 A inductive. Our 0.42 A was at the edge. After 500,000 cycles, the contacts welded.
The relay rating table
| Load type | Contact derating | Typical example |
|---|---|---|
| Resistive | 100% of rated | Heater, lamp |
| Inductive (DC) | 10–20% of rated | Solenoid, relay coil |
| Inductive (AC) | 30–50% of rated | Motor, transformer |
| Capacitive (inrush) | 20–30% of rated | Capacitor bank, LED driver |
| Motor (locked rotor) | 5–10% of rated | Starting a motor |
The fix: three options
Option 1: Bigger relay. Use a relay rated for 5 A inductive (25 A resistive). This overkill works, but the relay is physically larger and costs more. Not the elegant solution.
Option 2: Snubber / flywheel diode. Add a diode across the solenoid coil. When the relay opens, the diode provides a path for the inductive current. The back-EMF is clamped to about 0.7 V. No arc. The relay contacts see no inductive kick. This is the standard fix for DC inductive loads. A 1N4007 diode, reverse-biased across the coil terminals. Cost: $0.10.
Option 3: Solid-state relay (SSR). An SSR has no moving contacts. It switches with a semiconductor. No arc, no welding. But it has a voltage drop (1 V across the output) and generates heat. For a 0.4 A load, a small SSR works. For higher currents, a heat sink is needed.
We used the diode. It’s cheap, it’s passive, and it solves the root cause (the inductive kick). The relay didn’t weld again.
The snubber for AC loads
For AC inductive loads (contactors, motors), a diode doesn’t work (AC goes both ways). Use an RC snubber: a resistor and capacitor in series across the contacts. The snubber absorbs the inductive kick and suppresses the arc. Typical values: 100 Ω resistor + 0.1 μF capacitor. For higher-power loads, use a varistor (metal oxide varistor) across the coil.
What we spec now
- Diode (flywheel) across every DC solenoid/relay coil. Always.
- RC snubber across every AC contactor/motor contact.
- Relay contacts rated for at least 2× the actual inductive current (not resistive).
- For high-cycle applications (10+ cycles/min), use an SSR instead of a mechanical relay.
The rule: A relay switches current. If the load is inductive, the relay’s contacts see an arc every time they open. Size the relay for the inductive rating (10–20% of resistive), not the steady current. Or just add a diode and skip the problem entirely. The relay that welded was doing a job it was never rated for.
