OSSD test pulses are short, periodic drops of the 24 V output to 0 V, typically 100-300 µs wide, used by the sensor itself to detect cross-faults, shorts, and open wires in the field wiring [S5]. On a compliant safety input those pulses are filtered out and do not register as a turn-off event, which is why a safety relay sees a stable high while a standard PLC input sees a stream of glitches.
For practical wiring the rule is binary: the OSSD pair from a light curtain, laser scanner, or safety mat terminates at a safety-rated input (safety relay, safety PLC, or safety I/O module). Routing that same pair into a generic DI card breaks the safety argument because the PLC has no defined behavior for the pulses and will interpret them as a fault or as a cycling input [S3].
What an OSSD actually outputs
An Output Signal Switching Device is the electronic output of an ESPE such as a Type 2 or Type 4 light curtain, and it always comes as a redundant pair (OSSD1, OSSD2) sourced from the sensor's own internal safety logic [S3]. The pair carries 24 V signals whose test pulses are intentionally out of phase, so a cross-wire between the two conductors produces a recognizable collision pattern the sensor can detect [S3].
The pulse envelope is narrow on purpose: Kollmorgen's KAS safety controller documents a pulse width T1 between 150 µs and 1 ms and a repetition period T2 between 100 ms and 1 s on its safe digital inputs, with pulses declared non-overlapping [S4]. On the output side of the same controller, the test pulse window tightens to T1 = 450-550 µs and T2 = 810-990 ms, again non-overlapping across the four SOUT channels [S4]. Sensors from other vendors stay inside this same microsecond band, which is the range the safety input filter is designed to swallow.
How the safety relay input filters the pulses
Safety input stages use a low-pass or pulse-width discriminator that ignores any high-going-to-low transition shorter than the filter window, so a 150 µs OSSD dip is treated as a still-asserted input [S1][S2]. The same source is explicit: the safety-rated input at the other end does not generally evaluate the OSSD pulses because they are filtered out [S1][S2]. The Universal Robots safeguard input carries the same rule in plain text, with pulse lengths under 3 ms allowed through the filter so OSSD equipment can be wired directly [S7].
Kollmorgen formalizes the tolerance as a safe band: 150 µs to 1 ms wide and at least 100 ms apart, all pulses inside that envelope are transparent to the safety function, while pulses outside it will still trip the safety function rather than create a dangerous situation [S4]. Below the band, the relay is intentionally deaf; outside the band, the relay latches into its I/O failure state, typically with a fault reaction time of 2 ms or less [S4].
Why a standard PLC input is the wrong destination
A conventional 24 V digital input on a non-safety PLC has no defined OSSD filter, so it will register each test pulse as a logic-low transition and chatter the bit at the pulse repetition rate [S3][S6]. The DigiKey field guide is blunt: handling the OSSD signal pair as a discrete control wire is a categorization error leading to misapplied safety systems and extended downtime, and the signals must never be wired directly to a standard PLC [S3].
The PLCtalk thread reaches the same conclusion from the integrator side: because most OSSD outputs perform pulse testing by briefly dropping the voltage to zero, the PLC inputs must be capable of being conditioned, and a stock DI module is not [S6]. The failure mode is the same whether the pulses get registered as a stuck-on signal, a stuck-off signal, or a rapid toggle: in every case the safety stop command is no longer deterministic, which voids the SIL or PL claim on the loop.
Selection criteria for the receiving device
The first criterion is explicit OSSD compatibility, not just "24 V input". IEC 61496 is the governing standard for electro-sensitive protective equipment and is the document most safety controller manuals cite when they declare an input OSSD-compatible [S4]. The second criterion is a documented pulse filter window, with the vendor publishing both the pulse width it ignores and the maximum pulse width it still rejects as a fault, since those two numbers bound the entire integration.
The third criterion is the redundancy path: OSSD comes as a pair, so the receiving device must evaluate both wires, not just one, and must have a defined reaction if the two signals ever disagree [S3]. The fourth criterion is the safety rating on the input itself: at least PL d / Cat 3 under ISO 13849-1 or SIL 2 / SIL 3 under IEC 61508, matched to the SIL or PL claim of the sensor, otherwise the loop rating collapses to the weaker of the two devices. The fifth criterion, often missed, is diagnostic output: a safety relay or safety I/O module that names the specific OSSD channel that has dropped or shorted, which Kollmorgen implements through its I/O failure state messages [S4].
OSSD input vs test-pulse input vs standard DI
The three input classes diverge in exactly the place the field technician cares about. An OSSD-rated safety input filters pulses in the 150 µs to 1 ms band and evaluates the OSSD pair as redundant, so a single 24 V dip is invisible and a sustained drop is the stop command [S1][S2][S4]. A test-pulse safety input, used for dry contacts such as an E-stop or a guard-door limit switch, generates its own clocked 24 V from the safety card and looks for the return pattern; if the contact is shorted to 24 V or the wire is broken, the expected pulse train is corrupted and the input faults [S1][S2]. A standard PLC DI does neither, sees every test pulse as a logic change, and offers no defined safe state, which is why it cannot anchor a safety function [S3][S6].
The cost dimension lines up the same way: OSSD-capable safety I/O and safety relays sit in the higher price band and carry the certification paperwork, while test-pulse-capable safety inputs are often the same hardware running in a different parameter set, and a stock DI module is the cheapest of the three but disqualified from the safety stop path by definition.
Real wiring patterns that work
The most common pattern is a Type 4 safety light curtain with dual OSSDs landing on a safety relay or a safety input module, with the relay's safety contacts driving the contactor or the STO input of the drive [S3][S4]. A second pattern uses a safety laser scanner or safety mat whose OSSD pair feeds the safe digital input of a Kollmorgen AKD2S drive (SIN1-SIN4), with the drive itself configured to ignore the documented test-pulse envelope [S4]. A third pattern routes the OSSD pair to a robot controller's safety input, where the UR safeguard input's 3 ms pulse filter is wide enough to accept vendor-standard OSSD pulses without nuisance trips [S7].
A fourth pattern, frequently misused, is a light curtain feeding a standard PLC DI with a 24 V to 0 V signal that the integrator tries to "debounce in software": this is the failure case called out explicitly in the DigiKey field guide and on the PLCtalk forum, and it must be replaced with a proper safety input [S3][S6]. In machine retrofit work the cheapest fix is usually a DIN-rail safety relay with OSSD-compatible inputs, then a short jumper from the relay's safety contacts to the existing PLC DI for status-only monitoring, not for the stop command.
Failure modes the filter does and does not catch
The pulse filter is sized to ignore normal sensor self-test pulses, so it will not nuisance-trip on a healthy OSSD signal, but it will not rescue a wiring error either [S1][S2][S4]. A cross-wire between OSSD1 and OSSD2 changes the pulse phase relationship, which the sensor's own internal logic catches and forces both outputs off [S3]. A short from either OSSD to 24 V or to 0 V is also caught by the sensor, with both OSSDs driven to the safe state and the safety relay input going low for real, which is the intended stop path [S3].
The narrower failure modes are where the integration rule matters. If a single OSSD wire is open, the safety input loses one of its two redundant channels and the safety relay or safety module interprets that as a fault and trips, which is the correct fail-safe behavior [S3]. If the pulse width drifts outside the documented band, the Kollmorgen controller trips the safety function within 2 ms rather than letting the out-of-spec pulse through, with three overlapping pulses on different channels in a row also triggering the I/O failure state [S4]. If the receiving device is a non-safety PLC, none of those rules apply, and the same physical fault can present as a chattering bit, a stuck bit, or a missed stop, which is the entire reason the standard PLC destination is forbidden.
Standards and sourcing
The OSSD signaling concept is defined under IEC 61496 for electro-sensitive protective equipment, and a safety controller that claims OSSD compatibility typically cites that standard explicitly in its manual [S4]. The safety integrity of the receiving input is covered by ISO 13849-1 for performance level and category, or by IEC 61508 / IEC 62061 for SIL, depending on the machine builder's risk assessment. Kollmorgen's published timing bands (T1 = 150 µs to 1 ms, T2 = 100 ms to 1 s on inputs; T1 = 450-550 µs, T2 = 810-990 ms on outputs) and the UR safeguard input's 3 ms filter window are useful reference numbers when validating a third-party sensor against a known safety controller [S4][S7].
For new designs, the safest sourcing path is to match the OSSD sensor and the safety input from the same vendor's compatibility list, which removes the pulse-width and channel-redundancy arguments in one step. For brownfield retrofits, the next step is to verify on the bench with an oscilloscope that the chosen safety relay's input does not register the sensor's OSSD pulses as dropouts, which is the test the Beckhoff and automation.com articles walk through [S1][S2]. The signal diagnostics to capture are pulse width, pulse period, and the relative phase of OSSD1 versus OSSD2, with the comparison done against the controller's documented input filter window before the safety loop is commissioned.
Trackable signals to watch over the next quarter: vendor datasheets publishing the OSSD pulse envelope with a tolerance band rather than a single number, more safety I/O modules declaring a configurable pulse-filter width, and updates to the IEC 61496 family that may tighten the allowed test-pulse band. Each of those will shift which sensor-relay pairings qualify as drop-in compatible, and each is worth checking against the installed bill of materials before the next safety retrofit.
The underlying component specifications are covered under safety relay, construction machinery and equipment, and lamps and light fittings.
This topic is covered further in Quick-Lift vs Standard Pump on Manual Pallet Jacks: Mechanism, Strokes, and Sourcing.