A safety interlock switch is the door-position sensor in a machine safety circuit, while a guard-locking variant adds a solenoid or spring-driven bolt that physically keeps the door shut until the control system grants release [S3][S4].
Non-locking designs only report open/closed status to the safety relay or controller; locking designs combine that reporting function with a mechanical holding force, typically rated from a few hundred newtons up to several kilonewtons, that resists door opening until a release signal is applied [S2][S7].
Functional Difference Between Locking and Non-Locking Types
On a non-locking interlock, a cam-driven normally-closed contact opens the safety circuit the moment the actuator leaves the switch body, so any door opening forces a stop command but the door itself can be pulled at any time [S3].
On a guard-locking interlock, a spring (spring-lock variant) or a solenoid (solenoid-lock variant) keeps a tongue or bolt engaged with the actuator even if a person pulls on the door; the safety circuit then stays closed until the controller de-energises the solenoid or the spring is mechanically released, depending on the variant [S2][S4]. Bernstein's wiki states plainly that a guard-locking switch "combines the safety lock with an additional interlocking function" and "prevents operation whilst the door is open" rather than merely detecting it [S4].
This means non-locking devices are pure position sensors in the locking assembly chain, whereas guard-locking devices add a process-interlock layer, stopping the machine is not enough; the door must also be physically restrained until the hazard is verifiably gone.
When Guard Locking Is Required by the Risk Assessment
Guard locking is specified when the hazard does not disappear immediately on stop command: stored hydraulic or pneumatic energy, flywheel coast-down, hot tooling above 70 to 80 °C, robotic arms that can resume motion on a peripheral fault, or any access during a partial-still-running state such as limited-speed maintenance [S2][S3].
IDEC's safety guide ties the requirement directly to stop category and access timing: if personnel can reach the hazard before it has reached a safe state, a spring-lock or solenoid-lock interlock must be used, and the door must remain locked until the safety output confirms a safe condition [S3]. IDEM's product literature reinforces that "guard locking interlocks both monitor and lock, preventing access while machinery operates or stored energy presents a [risk]" [S2].
Conversely, a non-locking interlock is sufficient where stop time is fast (Category 0 or Category 1 with verified zero motion), residual energy is negligible, and a stop command reliably removes the hazard before an operator can physically reach the danger zone [S3][S8].
Spring Lock vs Solenoid Lock: Two Sub-Types of Guard Locking
Spring-lock (mechanical lock) interlocks keep the door bolted by spring force and only release when the solenoid is energised, which makes them fail-safe on power loss: if the cable is cut or the safety relay trips, the spring holds the door shut and the operator cannot enter a potentially energised cell [S3].
Solenoid-lock (electrical lock) interlocks do the opposite; the bolt releases only when the solenoid is energised, so a power loss immediately frees the door, which is the preferred choice for evacuation paths and ergonomic access where trapped personnel must be able to exit without controller intervention [S3].
Choosing between them is therefore a safety-direction decision, not a feature preference: spring-lock for hazards that must stay locked on failure, solenoid-lock for hazards that must unlock on failure, with the safety certification chain (typically PLe / SIL 3 on the monitoring contacts) unchanged in both cases [S3][S7].
Selection Criteria: Operating, Mounting, and Integration Limits
Hold force is the headline number on a guard-locking datasheet: IDEM and Reynders specify ranges from roughly 600 N for compact plastic-bodied units to 2,500 N and above for stainless-steel hygienic or heavy-duty models, with solenoid voltage commonly 24 V DC and a small handful of 110-230 V AC variants [S2][S7].
Contact configuration on the monitoring side is typically 2 NC + 1 NO or 3 NC, all positively driven, with variants offering OSSD (output signal switching device) semiconductor outputs for direct tie-in to a safety controller; AutomationDirect's range explicitly covers both a solenoid-locking tongue switch and a non-contact magnetic-locking RFID switch, illustrating the two physical-interface options in one product line [S6].
Environmental and mechanical limits matter as much as electrical ones: IP67 or IP69K is common for washdown, stainless 316 bodies are used on food and pharmaceutical skids, mechanical life is rated in the 1 million to 10 million cycle range, and a fire safety or ATEX/IECEx variant is required for explosive-atmosphere cabinets [S2][S7]. Mounting tolerance, typically a few millimetres of misalignment allowance on the actuator entry, is the practical reason RFID-coded non-contact variants are gaining share on vibrating or flexing guards [S6].
Decision Matrix: Locking vs Non-Locking by Use Case
For a small parts-bin guard on a packaging conveyor with sub-100 ms stop time and no stored energy, a non-locking hinge or tongue interlock at 1 NC or 2 NC is the lowest-cost correct answer; the door can be opened at any time and the safety relay will still drop out [S3][S8].
For a robot cell where coast-down plus capacitor-residual energy keeps the arm moving for 1-3 s after stop, a spring-lock interlock with 1,000-2,500 N hold force and PLe-rated contacts is the right call, because the spring must hold the door even if the safety contactor welds; a non-locking switch would let an operator pull the door into a still-coasting arm [S2][S3].
For a hydraulic press with stored pressure in the accumulator, a solenoid-lock interlock wired so the safety output energises the release only after the bleed-down timer has expired is the standard architecture; the bolt stays locked while pressure is above the safe threshold and unlocks only on confirmed safe state [S3][S4].
For a safety barrier door on a washdown skid in food or pharma, a stainless 316 spring-lock unit with IP69K and high coding-level RFID actuator is preferred, since the high coding level prevents defeat with a generic magnet or tool [S2][S6].
For maintenance access during limited-speed operation (safe speed monitoring), only a guard-locking switch with a separate lock-monitoring feedback contact, not a plain interlock, is accepted by most risk assessments, because the door must be physically held until the drive is in the safe-speed state [S2][S3].
Wiring, Diagnostics, and Common Failure Modes
Ferndale's 2025 wiring guide stresses that safety interlock contacts must be wired to a safety relay or safety controller input that performs cross-fault monitoring; tying a guard interlock into a standard PLC input defeats the dual-channel supervision that makes the interlock "safety-rated" [S8].
Common failure modes on non-locking types are actuator breakage, cam wear causing the NC contact to stick, and defeat by a spare actuator left near the switch; the coded-RFID variant directly addresses the defeat mode by requiring a uniquely coded actuator [S6].
Standards Mapping and Sourcing Notes
The functional requirements sit inside ISO 14119 for interlocking devices associated with guards, which the supplier pages reference implicitly when they describe two categories (with and without locking) and the defeat-prevention requirements that drive coded actuators [S3][S6].
For risk-graph and Performance Level calculations, the interlock feeds PL d or PL e depending on category and diagnostic coverage, with typical architectures 4 (dual-channel with monitoring) for the interlock contacts and Cat. 1 or 3 for the locking function depending on whether the lock is a safety-rated output or a process interlock [S3][S4].
On sourcing, the 2025 vendor landscape (Reynders, IDEM, Bernstein, IDEC, Keyence, AutomationDirect, Stronghold) all carry both technologies in the same product family, so the gating decision is the risk assessment outcome, not part availability; the lead-time split is roughly 1-2 weeks for the catalogue non-locking units versus 3-6 weeks for the stainless or coded-RFID guard-locking variants [S5][S6][S7].
Trackable signals worth watching: wider adoption of OSSD-only locking switches replacing electromechanical contacts in new-build lines, and the gradual shift from spring-lock to solenoid-lock on collaborative-robot cells as power-loss evacuation becomes a documented requirement in the cell risk assessment.
See also our earlier report, Water level hose vs laser level for work around obstructions.