Machine safety systems are the part of an automated cell that automation technicians are most likely to be blamed for, and the part where shortcuts do the most damage. A light curtain that nuisance-trips gets jumpered. A door switch that keeps faulting gets taped down. Then someone reaches in.
This guide explains how the main pieces work together: the risk assessment that decides what is needed, the devices that detect people (light curtains, interlocked guards, safety mats), and the safety controller that decides when to stop the machine. It also covers the OSHA rules behind them and the ISO 13849 terms you will see on drawings and in validation paperwork. It is written for technicians who maintain these systems, not as a design manual.
Why Machine Safety Systems Exist: The OSHA Baseline
The starting point in US general industry is 29 CFR 1910.212. It requires employers to use one or more guarding methods to protect operators and others from hazards such as the point of operation, nip points where parts come together, rotating parts and flying chips or sparks. The examples it names include barrier guards, two-hand tripping devices and electronic safety devices. The guard must keep the operator's body out of the danger zone during the operating cycle, and the guard itself cannot create a new hazard.
OSHA's machine guarding page sums up the principle: any machine part, function or process that can injure someone must be safeguarded. OSHA also runs a National Emphasis Program on amputation hazards in manufacturing, renewed by an instruction dated June 27, 2025. If you work in a plant with presses, robots or packaging equipment, that program is part of why inspectors look closely at guarding.
Specific machine types have their own rules. Mechanical power presses fall under 1910.217, which has the most detailed US requirements for presence-sensing devices. Beyond OSHA, the agency lists consensus standards such as ANSI B11.19 (performance criteria for safeguarding) and ANSI B11.TR3 (risk assessment). Those are not OSHA regulations themselves, but designers and auditors rely on them heavily.
Risk Assessment: Where Every Safety System Starts
Before anyone picks a light curtain or a safety controller, someone should have done a risk assessment. In plain terms, it asks:
- What can hurt someone on this machine, and during which tasks (running, setup, clearing jams, cleaning, maintenance)?
- How bad would the injury be, how often is someone exposed, and could they avoid it?
- What measures reduce that risk: design changes first, then guarding and safety devices, then warnings, training and procedures?
- How reliable must each safety function be, given that risk?
That last question is where ISO 13849 comes in. The output of the assessment is a list of safety functions, such as "opening the access door stops the robot" or "breaking the light curtain stops the press stroke," each with a required performance level.
As a technician, you probably will not lead the assessment, but you should know it exists and where it is kept. When someone asks you to move a guard, add an access point or change how a safety device is wired, the assessment is what tells you whether that change is acceptable. If there is no assessment, that itself is something to raise.
ISO 13849 Performance Levels and Categories
ISO 13849-1, now in its fourth edition from 2023, is the standard most often used to rate how reliable a safety-related control function is. You will see two terms on drawings and validation reports:
- Performance level (PL a through PL e) is a band of probability of dangerous failure per hour. PL e is the most demanding.
- Category (B, 1, 2, 3, 4) describes the architecture: single channel or redundant, and how faults are detected.
| Category | Structure | Diagnostics | Highest PL possible |
|---|---|---|---|
| B | Single channel | None | PL b |
| 1 | Single channel, well-tried components | None | PL c |
| 2 | Single channel with periodic testing | Low to medium | PL d |
| 3 | Redundant, cross-monitored | Low to medium | PL e |
| 4 | Redundant, cross-monitored, high-reliability parts | High (99% or more) | PL e |
The practical meaning for a technician: in Category 3 and 4 circuits, no single component failure may cause loss of the safety function. That is why you see dual-channel wiring, with two contacts on an e-stop or door switch running to two separate safety inputs. If one channel welds or opens and the other does not, the safety controller sees the mismatch and locks out until the fault is fixed. Category 4 adds the requirement that faults cannot pile up undetected between diagnostic checks.
How a Safety PLC Differs From a Standard PLC
A safety PLC or safety controller runs the logic for safety functions: e-stops, guard doors, light curtains, safe speed and safe torque off on drives. Examples include Rockwell's GuardLogix line and safety-rated CPUs from other major vendors. The differences from a standard PLC matter when you are troubleshooting:
- Redundant processing and self-testing. Safety controllers check their own hardware and their inputs and outputs continuously. A fault that would go unnoticed on a standard PLC causes a safe shutdown on a safety controller.
- Safety-rated I/O. Inputs often send test pulses down the field wiring to detect shorts between channels or to another voltage. That is why safety input wiring is sensitive to how it is landed.
- Locked and signed logic. Safety programs are usually protected with a password or signature so that changes are deliberate and traceable. Standard logic can often be edited online; safety logic generally should not be touched without authorization and revalidation.
- Separate safety network traffic. Safety data on industrial networks uses protocols built to detect lost, delayed or corrupted messages.
A standard PLC can monitor safety status for the HMI and reset logic, but it should not be what actually stops the hazard. If you see a hazardous motion being stopped only by standard PLC logic, flag it. Our PLC programming basics piece covers standard logic; safety logic is a separate discipline with its own rules.
Light Curtains and the Safety Distance Rule
A safety light curtain is a presence-sensing device: an emitter and receiver create a field of beams, and breaking the field sends a stop signal. The key installation question is distance. If the curtain is too close to the hazard, a hand can get through it and into the danger zone before the machine stops.
For mechanical power presses, OSHA 1910.217 sets the minimum safety distance with a simple formula: distance in inches equals 63 inches per second (a hand-speed constant) multiplied by the press stopping time in seconds. A press that takes 0.2 seconds to stop needs the sensing field at least 12.6 inches from the point of operation. The same rule also says:
- The device must stop or prevent the stroke if any part of the operator's body is in the sensing field.
- It may not be used on presses with full-revolution clutches.
- It may not be used to start the stroke, except under the separate presence-sensing device initiation rules.
- A failure must not prevent normal stopping, must block the next stroke until fixed, and must be indicated.
- Every other route into the point of operation needs a guard.
Other machine types use different formulas from consensus standards, often with added distances based on the curtain's resolution. The principle is the same everywhere: stopping time drives the distance. That is the part technicians affect most, because worn brakes, changed drive deceleration settings or heavier tooling all lengthen stopping time. A curtain that was correctly placed at installation can end up too close without anyone moving it.
Muting and blanking
Muting temporarily suspends the curtain during a safe part of the cycle, such as a press upstroke or a pallet passing through on a conveyor. Blanking ignores specific beams so a fixed object or moving part can pass through. Both are legitimate when designed and validated. Both are also common ways systems get quietly defeated. If muting sensors are misaligned or a blanking setting has been widened, the curtain may no longer protect what it was installed to protect.
Safety Systems Do Not Replace Lockout
A light curtain or door switch protects people during normal operation. It is not energy isolation. OSHA's lockout rule, 1910.147, applies to servicing during production when you have to remove or bypass a guard or put any part of your body into the point of operation or another danger zone. There is a narrow exception for minor, routine, repetitive adjustments that are integral to production, and only when alternative measures give effective protection.
In practice, clearing a jam with the e-stop pressed is not the same as locking out, and opening a guarded door does not isolate stored energy in air cylinders, springs, gravity loads or capacitors. When in doubt, lock it out.
A Technician's Checklist Before and After Touching Safety
Use this whenever you work on a safety device, its wiring or anything that affects stopping time:
- Confirm you are authorized to work on the safety system, and who must sign off.
- Find the drawings and the risk assessment or safety requirements for the function you are touching.
- Replace failed devices with the same part or an approved equivalent with the same safety rating.
- Never leave a jumper, override key or bypass in place. If a temporary bypass is unavoidable for troubleshooting, follow the plant's written procedure and controls.
- After any change to brakes, drives, tooling or the device itself, have stopping time re-measured where the design depends on it.
- Function-test every affected safety function: each e-stop, each door, each beam of the curtain, each muting condition.
- Verify dual-channel inputs fault correctly when one channel is opened.
- Document what you changed, what you tested and the results.
This is also a career skill. Technicians who can troubleshoot safety circuits without defeating them are trusted with more of the plant. ISA's technician credentials touch on this area; see our CCST guide for that path.
Bottom Line
Machine safety systems are layered: a risk assessment decides what protection is needed, devices like light curtains and interlocked guards detect people, and a safety controller with redundant, self-checking logic stops the hazard. OSHA sets the baseline, and standards like ISO 13849 rate how reliable each function must be. Your role as an automation technician is narrow but critical. Keep the system working the way it was designed, treat every fault as a real failure rather than a nuisance, re-verify after changes and never leave a bypass behind.