Safety Light Curtain Selection: Type, Resolution, Distance

Safety Light Curtain: What You Actually Choose

Short version. A safety light curtain comes down to three choices: resolution (14 mm means finger protection, 30 mm means hand protection), protective field height (300 to 2,100 mm in 150 mm steps on the Sick deTec4 Core), and the mounting distance from the hazardous point. You do not choose that distance, you calculate it per ISO 13855:2010, and the formula needs the stopping time of your machine. You measure that. And one more thing that drops out of the selection most often: the curtain switches nothing off by itself, its OSSDs still have to be processed safely by the machine's safety relay or safety controller (pp. 10 and 22 of the instructions). Type 2 or Type 4 per IEC 61496 is not a fourth matter of preference, it follows from the Performance Level your machine's risk assessment requires.

What sits in our safety device section: six Sick deTec4 Core units, the receiver of a single-beam Sick L21 barrier (with no matching sender in the catalogue), nine safety laser scanners and one Omron door interlock switch. Below we work through how those six units form two working pairs and why two of them do nothing at all on their own — and by the end the same has to be said about the single-beam device.

The boundary of responsibility. We sell the devices and quote figures from the manufacturer's documents with page numbers, while the risk assessment, the required PL, the design of the safety circuit, the installation and the validation belong to whoever is responsible for the machine. This article certifies nothing and replaces neither a risk assessment nor acceptance testing.

Type 2 versus Type 4: the Difference Is Printed in the Datasheet

If the risk assessment calls for PL e, a Type 2 single-beam device will not do, and the two datasheets show that without any interpretation. We stock both classes, so we can put the figures side by side.

Here is that condition word for word from the L21E datasheet: "Only in combination with an appropriate test device, e.g., Flexi Classic or Flexi Soft". That footnote sits on five rows: Type 2, SIL 1, Category 2, PL c and PFHd. The TM row (20 years) carries no footnote at all, and the test rate and maximum demand rate have footnotes of their own. Without an external test device the figures in that table simply do not apply. The same datasheet adds that at least 100 internal or external tests must occur between two demands on a safety-related response, and that the maximum demand rate is 60 per minute.

Flexi Soft, which the datasheet names as an example, we stock under PLCs: the FX3-CPU000000 controller and the FX3-XTIO84002 module. We state that as a catalogue fact, not as a recipe: whether a particular combination delivers the PL you need on your machine is settled by your circuit calculation and your validation, not by our article and not by two part numbers sharing a basket.

What follows is our own practice, not a requirement of a standard or a datasheet. We fit an L21S + L21E pair when one level in a narrow passage needs monitoring and the machine's risk assessment accepts PL c with a test device. We fit a curtain when the guarded opening is one a person reaches into and the risk assessment calls for PL e. Which PL you actually need is settled by the risk assessment, not by us. The PFHd gap between our two devices is roughly 270 times (that is our division of 1.0 × 10⁻⁶ by 3.7 × 10⁻⁹, not a datasheet figure).

Resolution: 14 mm for Fingers, 30 mm for Hands

The Sick operating instructions, page 11, leave no room for interpretation: resolution is the size of the smallest object the curtain detects in the protective field, and it equals the diameter of the test rod belonging to that curtain. Then come exactly two sentences: 14 mm provides finger protection, 30 mm provides hand protection. That is all the document says about it. There is no ready-made rule saying "if a finger fits, take 14 mm", and we are not inventing one: which resolution your case needs is settled by the machine's risk assessment.

The test rod ships with the receiver, together with the safety note and the mounting instructions, while the sender ships on its own (p. 65). And separately, a DANGER notice on page 26: use only the included test rod, in the diameter stated on the type label of that curtain, and do not use rods of a similar or identical diameter from other curtains. Remember that, it matters two sections from here.

Resolution drives three things at once, which is the main reason you do not pick it "with margin":

• What gets detected. A 30 mm curtain does not see a finger. That is not a defect, it is a rated property.
• Mounting distance. Resolution enters the distance formula as the term 8 × (d − 14). For 30 mm that adds 128 mm to the distance; for 14 mm it adds nothing.
• Response time. Here 14 mm loses: at 1,800 mm field height it is 20 ms against 13 ms for the 30 mm version.

Field width depends on resolution too, and in the opposite direction. Per the general data table in the instructions, the 14 mm version works over a minimum guaranteed 0 to 8 m and a typical 0 to 10 m, while the 30 mm version manages 0 to 12 m and 0 to 15 m respectively. The sender emits invisible near-infrared at 850 nm.

Safety Distance: the Procedure from the Manual, Not from Memory

The distance between the curtain and the hazardous point exists so that a person cannot reach the hazardous point before the machine leaves its dangerous state. The deTec4 Core instructions give the calculation per ISO 13855:2010 for an orthogonal (right-angled) approach to the field, and warn plainly that a different geometry (a field parallel to the direction of approach, at an angle, or an indirect approach) calls for a different calculation. The same page 17 states the basis of the calculation itself: it rests on the international or national standards and the statutory requirements applicable at the place where the machine is installed. For the USA, within the scope of OSHA and ANSI, the instructions name different rules specifically: CFR 29 Part 1910.217 and ANSI B11.19. And any deviation from ISO 13855:2010 calls for an individual effects analysis.

The procedure, as the instructions print it:

1. Calculate S = (K × T) + 8 × (d − 14 mm), where S is the minimum distance in mm, K is the approach speed of a person or body part in mm/s (for example 2,000 mm/s), T is the machine stopping time plus the curtain's response time after the light path is interrupted, in seconds, and d is the resolution in mm.
2. If S ≤ 500 mm, that value is the minimum distance.
3. If S > 500 mm, recalculate with an approach speed of 1,600 mm/s.
4. If the new S > 500 mm, use the new value.
5. If the new S ≤ 500 mm, use 500 mm.

Sick's own worked example in the same document: machine stopping time 290 ms, curtain response time 30 ms, resolution 14 mm. T = 0.32 s, S = 2,000 × 0.32 + 8 × (14 − 14) = 640 mm, which exceeds 500, so the recalculation at 1,600 gives 512 mm.

Now our own calculation. Two complete pairs assemble from our items: 750 mm / 30 mm (the `C4C-SA07530A10000` sender plus the `C4C-EA07530A10000` receiver) and 1,800 mm / 30 mm (`C4C-SA18030A10000` plus `C4C-EA18030A10000`) — both are calculated below. The 1,800 mm / 14 mm rows are calculated for the `C4C-EA18010A10000` receiver, which has no matching sender in the catalogue (Sick 1211486), and they are here not as an offer but to show how resolution drags the distance along with it. The response times come from Table 16 of the instructions (p. 62), the resolutions from the same document, and the machine stopping time is an assumption we made for the example. On your machine it is a different number and it has to be measured. Sick, incidentally, offers a stopping and run-down time measurement service in many countries — that is stated in the same instructions on page 17.

That table shows something a price list cannot. The 8 × (d − 14) term pushes a 30 mm curtain 128 mm further from the hazardous point than a 14 mm one at the same stopping time. The 14 mm version hands part of that gain back through its longer response time. On a fast machine the difference works out around 60 mm, on a slow one around 114 mm. A caveat about the comparison itself: we used different heights (750 against 1,800 mm), and the response time in Table 16 depends on height as well (at 1,800 mm it is 20 ms at 14 mm resolution against 13 ms at 30 mm). At a single height of 1,800 mm the gap due to resolution does not go away: on a fast machine it is the same 60 mm, and on a slow one slightly larger, about 117 mm (948.8 mm at 30 mm against 832 mm at 14 mm). Mixed heights, if anything, understate that gap; to see the effect of resolution alone, hold the height constant. That does not mean resolution is chosen to fit the space available: the required resolution comes from the risk assessment, and if there is not enough room for the calculated distance, what has to change is the layout or the machine stopping time.

One more clause from the same document that people like to skip. Per ISO 13855:2010 the ESPE must not be defeatable. If access to the hazardous area by reaching over the protective field cannot be prevented, the field height and the distance are determined by comparing two calculations, one based on detecting limbs and one based on reaching over, and the greater value is the one you use. And in note 3 to the figure on page 18 there is a separate layout requirement: depending on the application and the distance, persons must be prevented from standing behind the protective device — otherwise someone ends up inside the hazardous area where the field no longer sees them.

Stainless and Guards: Reflective Surfaces

Beams from the sender can bounce off a shiny surface and travel around an object, so that the object goes undetected. The instructions give a separate minimum distance a from the protective field for this, and it has to be held on every side of the field: horizontally, vertically, diagonally and at the end of the curtain. The same area has to stay free of dust, fog and smoke.

The formula from the table in the instructions (for automatic calibration of the field width) depends on the distance D between sender and receiver: for D ≤ 3 m the minimum distance is a = 131 mm, and for D > 3 m, a = tan(2.5°) × 1,000 mm/m × D, which is 43.66 × D in millimetres with D in metres. For a 6 m wide field that is roughly 262 mm of clear space on each side. And now the multiplier, which applies to five of our six C4C units. Section 14.3 of the instructions (p. 66) states that the weld spark guard reduces the system's scanning range by 15%, and separately that for a device with a resolution of 30 mm fitted with that guard the calculated minimum distance from reflective surfaces must be doubled. So instead of 131 mm at D ≤ 3 m you keep 262 mm, and for a 6 m field, instead of our calculated 262 mm, about 524 mm. Five of our six units have exactly that 30 mm resolution.

In practice this means a stainless chute, an aluminium table or a freshly painted guard 10 cm from the field is not a layout detail. It is the reason the curtain stops seeing a hand.

What We Actually Have: Six Units and Two Working Pairs

A light curtain is always two devices: a sender and a receiver of the same height and the same resolution. In the type code the sender is SA and the receiver is EA. The rest of the code reads simply: the three digits after SA or EA are the field height divided by ten, and the next two digits are the resolution code (10 for the 14 mm version, 30 for the 30 mm one). The part numbers below come from the ordering tables in the instructions; they are not inferred from the code pattern.

Now the part that matters, stated plainly because a product page does not show it. Those six items form two complete pairs: 750 mm at 30 mm, and 1,800 mm at 30 mm. Two of them, EA12030A10000 and EA18010A10000, are receivers with no matching sender in the catalogue. Ordering a receiver on its own only makes sense if you already have a sender of the same height and resolution, as a replacement for a damaged one. As a standalone device a receiver does nothing. That is the state of the catalogue as this was written, not a statement about availability; check the product pages or ask us.

If what you need is the 1,200 mm height or 14 mm resolution at 1,800 mm, the matching sender for those two receivers (1211501 and 1211486 in the ordering tables) is not in the catalogue right now. Write to us and we will tell you whether it can be brought in.

On the order content: a working pair is a sender plus a receiver, and the brackets are separate part numbers. The instructions list them on page 66 (QuickFix, FlexFix, Compact FlexFix, flat mount, swivel mount), and without brackets there is physically nothing to mount the curtain on. From that list our catalogue holds the BEF-1SHABPKU4, a kit of four FlexFix brackets, Sick part no. 2066614 from the same Table 23. For the rest, write to us and we will check what we can bring in.

One more detail from the scope of delivery: the test rod travels with the receiver. If you buy only a sender as a replacement, there will be no rod in the box, and the rod is exactly what you check the curtain with, in the diameter stated on the type label.

Electrical Side: the Curtain Switches Nothing Off

This is the heart of the whole section, and it is stated as early as page 10: when one or more light beams are completely interrupted, the curtain reports that on the OSSDs through a signal change, and it is the machine or its control system that has to evaluate those signals safely and end the dangerous state, for example via a safety relay or a safety controller.

Page 22 expands this into a list of requirements carrying DANGER status, meaning that without them the device does not fulfil its protective function:

• The output signals have to be analyzed by downstream controllers so that the dangerous state of the machine is ended safely; depending on the safety concept that is a safety relay or a safety controller.
• It must be possible to influence the machine control electrically.
• The electrical control of the machine must meet the requirements of IEC 60204-1.
• The device has no integrated restart interlock. Depending on the applicable national regulations or the required reliability of the safety function, it has to be implemented in the external control.
• When a safety controller is used, differing signal levels on the two OSSDs have to be detected — and the instructions put that conditionally: "depending on applicable national regulations or required reliability of the safety function". The maximum discrepancy time the controller tolerates is selected according to the application.
• OSSD1 and OSSD2 must not be connected to each other, and the machine controller has to process their signals separately. The same requirement is repeated on page 44 as a DANGER notice ahead of the electrical installation — and a third DANGER sits right beside it, the one people forget: prevent the formation of a potential difference between the load and the protective device.

Now the outputs themselves. Two PNP OSSDs, short-circuit protected and cross-circuit monitored against each other, each carrying up to 300 mA with up to 2 mA of leakage current (p. 61). If the device detects a fault, at least one OSSD goes to the OFF state (p. 60).

Two points that usually explain a failed first start-up:

• The outputs are tested cyclically with brief LOW pulses, and there is a device-revision trap here. The test pulse width of 130 / 150 / 300 µs (min/typ/max) applies only to devices marked "(Rev. 1)" in the "Ident No." entry on the type label. For devices without that marking the pulse width is typically 300 µs and at most 350 µs. The rate of 3 / 5 / 10 s⁻¹ carries no such condition. So read the label before choosing the input and design for the worse figure, 350 µs rather than 300. The same marking splits one more row, and there the gap is more than a factor of 70 (73, to be exact — our division): the permissible load capacity on the OSSDs is 2.2 µF on devices marked "(Rev. 1)" and only 30 nF on devices without it (p. 61, footnote 4). A long shielded cable to a safety relay reaches that capacitance easily. The instructions ask you to make sure the test pulses do not shut the downstream controller down; type 1 and type 3 digital inputs per IEC 61131-2 are compatible with them in terms of voltage levels — the IEC 61131-2 footnote sits on the HIGH and LOW switching voltage rows of that table specifically (p. 61) — and that is all this compatibility means. It does not permit wiring the OSSDs into an ordinary, non-safety PLC input, because the page 22 requirement for safe evaluation of the signals still stands.
• Cable resistance between OSSD and load must stay at or below 2.5 Ω. Otherwise a cross circuit between the outputs may not be detected reliably, which touches the safety function itself. The instructions point to IEC 60204-1 here as well (p. 61). In terms of length, Table 18 on page 63 puts it like this: for a 0.34 mm² copper conductor, up to 85 m on the sender cable and up to 15 m on the receiver cable, with separate cables running to the cabinet holding the safety relay or safety controller.

And about the test rod check, which falls to the installer. The instructions attach three DANGER notices to it (p. 26): the dangerous state of the machine has to be switched off and stay switched off during the check, the curtain outputs must have no effect on the machine during it, and above all, if the OSSD LED on the receiver lights up green even briefly while the rod passes through, work on the machine must stop and the mounting and electrical installation have to be checked by qualified safety personnel. Normally the LED stays red continuously (pp. 26-27).

For the supply, the instructions require SELV or PELV per IEC 60204-1, a fuse rated no more than 4 A in the 24 V circuit, and a supply able to buffer a brief 20 ms power failure (also per IEC 60204-1). The voltage range is 19.2 to 28.8 V at a 24 V nominal. Operating temperature is −30 to +55 °C, with a separate condition that the temperature difference between sender and receiver must not exceed 25 K and that the cable must not be flexibly mounted below −25 °C. Enclosure rating is IP65 and IP67, the housing is a 31 × 34 mm extruded aluminium profile, and the front screen is polycarbonate with a scratch-resistant coating.

If you need to get the logic of PNP outputs and controller inputs straight in general, not only in a safety context, we have a separate piece on it: connecting a sensor to a PLC, NPN versus PNP. The device that will process the OSSDs is selected on three counts: the number of channels (one curtain needs two inputs), the ability to analyze both OSSDs separately, and an adjustable discrepancy time between them. The Sick Flexi Soft safety modules that the L21E datasheet itself names as an example of a test device sit in our PLC section, and here it matters not to order the wrong one: the safe inputs that OSSDs are wired into are on the I/O module, the FX3-XTIO84002 (its type code carries the 84: eight safe inputs and four safe outputs), while the main module, the FX3-CPU000000, has no safe inputs at all (Table 3 on p. 19 of the Flexi Soft instructions: the FX3-CPU0 row carries dashes in both the "Inputs" and "Outputs" columns, the FX3-XTIO row carries 8 and 4). A CPU on its own does not accept a curtain. And one more detail from the same instructions that no price list shows: the main module only works together with the FX3-MPL0 system plug (p. 20), which supplies the system and stores the configuration. For this CPU that is the FX3-MPL000001, Sick part no. 1043700 (p. 176; the variant with protective coating is the FX3-MPL000011, 1050619, same row). The plug is not in our catalogue — write to us and we will tell you whether it can be sourced. Which combination you need is decided by your safety circuit design, not by our article.

When a Curtain Is the Wrong Device

A curtain covers a vertical plane. Three jobs exist where it either does not work or costs more than the right answer, and we carry a device class for each.

• A horizontal zone around a machine or under a conveyor. This needs a safety laser scanner with a software-defined contour. We have nine: the S3000 series (for example S30A-4011BA), microScan3 and nanoScan3. We did not open the datasheets of those nine scanners, so there is not a single figure for them here and we are not inventing any: field angle, radius and response time are on the product pages. If it is easier, send us the dimensions of the zone to be guarded and the measured machine stopping time, and we will open the datasheets against that zone.
• A door or a guard opened by hand. That is a job for a safety door interlock switch, not a curtain. We stock the Omron D4NS-5AF: 1NC+1NO slow-action contacts (letter `A` in the model designation) and two Pg13.5 conduit entries (the `5` in the designation marks the 2-conduit version — Omron catalogue C128, pp. 1-2). But the same story applies here as to the curtains: the operation key for this switch is sold separately — the D4NS datasheet says so twice, verbatim "The Operation Key is sold separately", and the keys are a separate D4DS-K series. We do not stock keys, and without one the switch does not monitor the door. Write to us and we will tell you whether a key can be brought in for your door design. The mechanics of limit switches in general are covered in a separate piece.
• One level in a narrow passage over a long distance. That calls for an L21 single-beam barrier, and it too consists of two devices: an L21S sender and an L21E receiver. Our catalogue holds the receiver only, the L21E-21MA1A (Sick part no. 6034875); the matching L21S-21MA1A sender (6034874) is not there. Like a curtain receiver, it does nothing on its own. Its datasheet figures: 16 m, a response time of 0.2 ms plus the test device response time, Type 2, Category 2, PL c, SIL 1 — and all five of those values are valid only in combination with a suitable external test device. Write to us and we will tell you whether the sender can be brought in.

Separately: if your job is not protecting a person but simply detecting an object, a curtain is not needed at all and there is nothing to pay Type 4 for. Take an ordinary photoelectric sensor. We covered the types in our guide to photoelectric sensors, and the general task-based selection is worked through in choosing sensors for automation.

What We Ask Before an Order

Four things, and without them curtain selection is guesswork: the height of the opening to be covered; what has to be detected (finger or hand); the distance between sender and receiver; and the measured stopping time of the machine. The fourth question is the one that usually stops a customer, and that is fine: without it the mounting distance cannot be calculated, and without the distance the rest of the choice means nothing.

To see what we currently have in this class, look at the safety device section; all Sick optics are collected in the brand filter, and the full sensor catalogue sits in the sensors section. Send us the opening dimensions and the measured stopping time through contacts and we will show the substitution into the manufacturer's formula with page numbers, so it is visible where every figure came from. That is arithmetic on datasheet data, not a safety engineering service: the calculation itself, the choice of PL and the validation are carried out by whoever is responsible for the machine.

Sources

• SICK, deTec4 Core Safety Light Curtain, Operating Instructions, document 8014253/1S7O/2025-04-08: pp. 10 and 22 (safe evaluation of the OSSDs, IEC 60204-1, no integrated restart interlock, discrepancy time), p. 11 (resolution, scanning range), pp. 17-19 (minimum distance per ISO 13855:2010, place-of-installation requirements, the stopping time measurement service, reaching over the protective field, the note on standing behind the device), pp. 19-20 (reflective surfaces), pp. 26-27 (test rod check), p. 44 (separate OSSD wiring), pp. 60-62 (Type 4, Category 4, PL e, SIL 3, PFHd, OSSDs, response time, ambient data), p. 63 (Table 18, cable lengths), p. 65 (ordering tables, scope of delivery), p. 66 (brackets Table 23, weld spark guard, doubling of the distance from reflective surfaces at 30 mm resolution).
• SICK, the system datasheet for the L21S-21MA1A + L21E-21MA1A pair (`dataSheet_L21P-21MA1A_p76845_en.pdf`), p. 2, "System part" rows: receiver L21E-21MA1A — 6034875, sender L21S-21MA1A — 6034874. Plus the separate receiver datasheet for the L21E-21MA1A, part number 6034875, PDF generated 2026-09-17 — that is the file generation timestamp, not a document revision number: pp. 2-3 (Type 2, Category 2, PL c, SIL 1, PFHd, the external test device condition, range, electrical data). By contrast, the deTec4 Core mark 8014253/1S7O/2025-04-08 is a genuine revision.
• OMRON, D4NS / D4NS-SK Safety Door Switch, Data Sheet (English, assets.omron.eu): the model designation (letter A = 1NC/1NO slow-action) and the twice-repeated "The Operation Key is sold separately", key series D4DS-K, and the model-code decoding (letter `A` = 1NC/1NO slow-action).
• OMRON, catalogue C128-E2-03B, Safety Door Switch D4NS/D4NS-SK: p. 1 — the model-code legend, "5: Pg13.5 (2-conduit)"; p. 2 — the model list headed "Switches (Operation Keys are sold separately.)", row `D4NS-5AF` (2-conduit, slow-action, 1NC/1NO, Pg13.5). The verbatim sentence "The Operation Key is sold separately." sits in the note on p. 1.
• SICK, Flexi Soft Modular Safety Controller, Hardware, Operating Instructions, 8012478/1YG3/2026-10-01: p. 19 (Table 3, module overview: FX3-CPU0 with no inputs or outputs, FX3-XTIO with 8 inputs / 4 outputs), p. 20 ("The main module can only be operated together with the FX3-MPL0 system plug"), p. 24 (Table 4, system plug variants; FX3-XTIO with 8 safety inputs and 4 safety outputs), pp. 176-177 (part numbers: FX3-MPL000001 — 1043700, FX3-CPU000000 — 1043783, FX3-XTIO84002 — 1044125).