13
PLC Installation Practices, Editing, and Troubleshooting
Parts I–III taught you to write any PLC program you’d ever need. Part IV is a different conversation: how do you actually install the hardware in a real factory, get the program safely into the field, keep it running for the next ten years, and find the fault when something inevitably breaks? This chapter is your bridge from “writing logic” to “running plants” — covering the noise-rejecting layout of a control panel, the discipline of online editing, the rhythm of preventive maintenance, and the input-to-output troubleshooting trail that finds every broken link. The four case studies are real workshop scenarios from working PLC technicians.
What you’ll be able to do after this chapter
Your goals for this chapter:
- Specify a NEMA-rated enclosure appropriate to the environment, and lay out its interior to separate power from signal cables.
- Identify common electrical noise sources (VFDs, contactor coils, welders) and apply correct mitigations.
- Diagnose leaky inputs and outputs caused by long cable runs and solid-state leakage current.
- Apply correct grounding practice — single-point ground, separate earth/chassis/signal grounds, and avoid ground loops.
- Protect a system from voltage variations and surges using MOVs, surge suppressors, isolation transformers, and UPS units.
- Perform safe online editing and commissioning on a running production system, with the discipline that prevents accidents.
- Use online monitoring and force tables to investigate program behaviour without stopping the plant.
- Run a preventive maintenance schedule that catches issues before they cause downtime.
- Troubleshoot a non-responsive output by walking the signal path: input → ladder → output → wiring → field.
- Navigate the standard programming software ecosystem — RSLinx, RSLogix 500, RSLogix/Studio 5000 — at a working level.
Key Concepts & Terms
PLC Enclosures
The enclosure is the first thing the field engineer specifies and the last thing the PLC programmer thinks about — but its design decisions affect every input, every output, and every byte of memory’s reliability for the system’s lifetime. Enclosures protect the PLC from dust, moisture, vibration, mechanical impact, and (just as importantly) protect the surrounding environment from the PLC in case of a fault.
NEMA ratings — pick the environment first
In North America, enclosure protection is specified by NEMA ratings; the rest of the world uses the equivalent IEC IP codes. The four most common NEMA classes for PLC enclosures are:
| NEMA | Equivalent IP | Protects against | Typical use |
|---|---|---|---|
| NEMA 1 | IP20 | Falling dirt, accidental human contact | Indoor offices, dry electrical rooms |
| NEMA 12 | IP54 | Dust, dripping non-corrosive liquid, oil | Indoor industrial — most factory floors |
| NEMA 4 | IP66 | Dust, hose-directed water, ice formation | Outdoor and washdown areas |
| NEMA 4X | IP66 | Same as NEMA 4 plus corrosion (stainless steel) | Food processing, marine, chemical plants |
Layout — the rules that affect electrical reliability
How components are physically arranged inside the enclosure is not just an aesthetic choice. Two principles dominate good layout:
- Separate power from signal. Line-voltage cables (motor feeders, contactor coils, 230 V control) run on one side of the enclosure. Low-voltage signal cables (24 V DC inputs, 4–20 mA loops, RS-485 networks) run on the opposite side. Where they must cross, they cross at right angles — never run parallel for any significant distance.
- Heat rises — put the PLC at the top. The PLC processor and I/O cards generate the least heat in the panel; VFDs and motor contactors generate the most. Mounting the PLC near the top of the enclosure, above the heat-generating equipment, exposes it to the coolest air. Keep at least 5 cm of clearance above and below each I/O rack for natural convection.
Figure 13.1 — Best-practice PLC enclosure layoutPower equipment (red, left) is physically separated from PLC and signal modules (blue, right). The PLC sits high in the enclosure where air is coolest. A single ground bus along the bottom collects all earth grounds. A door interlock disconnects power when the cabinet is opened.
Door interlocks — safety, not optional
A panel that contains line-voltage components must have a door interlock that automatically disconnects power when the door is opened. This is a safety code requirement in most jurisdictions (NFPA 70E, IEC 60204), not a nice-to-have. The interlock is typically a key-locked rotary disconnect at the side of the panel, or a captive-contact mechanism that physically prevents the door from opening while live.
Electrical Noise
Every electrically-driven device — every motor, every welder, every arc-flash, every fluorescent ballast — radiates electrical noise. In a clean lab environment this rarely matters. On a factory floor, noise is the single most common cause of “ghost” PLC failures: inputs that flicker, communication that drops, false alarms that wake operators at 3 a.m. Understanding noise is half of installation engineering.
Where noise comes from
- Variable Frequency Drives (VFDs). The single biggest noise generator in modern factories. The high-frequency switching of the drive’s IGBTs (typically 4–16 kHz) couples noise into nearby cables, both as conducted noise on the AC supply and as radiated noise into adjacent wiring.
- Contactor and relay coils. When a coil de-energises, the collapsing magnetic field generates a high-voltage spike (back-EMF) that can reach hundreds of volts on a 24 V coil, hundreds of thousands on a 480 V coil. This spike couples into nearby signal wires.
- Arc welders, induction heaters, RF heaters. Massive radiated EMI across a wide frequency band. A welder operating 10 m from a panel can be enough to upset a poorly-installed PLC.
- Switching power supplies. Modern DC power supplies are themselves switching converters and emit noise at their switching frequency.
Symptoms of a noise problem
- Inputs randomly flicker ON when the field switch is open — and the flickering correlates with some factory event (a motor starting, a welder striking).
- Communication errors on networks (Modbus, DeviceNet, Ethernet/IP) that come and go.
- Analog readings that are noisy or jump randomly — but only when certain equipment runs.
- Sporadic processor faults with no obvious cause, often clustered in time around shift changes or specific machine cycles.
Mitigation — the standard toolkit
- Physical separation. Noise dies with distance. 30 cm minimum between power and signal cables; 50 cm or more between VFD cables and signal wiring.
- Shielded cable for signals. A grounded foil or braid shield around the signal conductors blocks most radiated noise. Ground the shield at one end only (typically the panel end) — grounding both ends creates a ground loop that itself becomes a noise pickup.
- Twisted pair conductors for differential signals (4–20 mA, RS-485). Adjacent twists pick up equal noise that cancels at the receiver.
- Ferrite cores on cables passing through the panel wall to absorb high-frequency conducted noise.
- RC snubbers across contactor coils to absorb the back-EMF spike at the source. A 100 Ω resistor in series with a 0.1 µF capacitor across an AC contactor coil costs about $2 and prevents thousands in mystery faults.
Leaky Inputs and Outputs
A “leaky” input or output is one that’s slightly on when it should be fully off — not energised enough to actually drive a load, but enough to register as ON to a sensitive logic threshold. Two physical mechanisms cause leakage:
- Solid-state output leakage current. TRIAC and transistor outputs in their “off” state still leak a small current — typically 1.5–10 mA. Through a high-impedance load, that small current can develop most of the supply voltage. This shows up most often when chaining a solid-state output to a sensitive solid-state input through a long cable.
- Cable capacitance with leakage from sensors. A 24 V DC NPN sensor has internal leakage through its semiconductor switching transistor (typically 0.1–0.5 mA). On a long cable run with significant distributed capacitance, that leakage can charge the cable to a significant fraction of supply voltage even when the sensor’s switch is open.
The diagnostic test — measure with a meter
Suspect a leaky input? Open the field switch, leave the PLC powered, and measure the voltage at the input terminal with a digital multimeter (high-impedance, 10 MΩ).
| Voltage at input terminal (switch open) | Diagnosis |
|---|---|
| 0.0 V (or close) | Healthy off-state. No leakage. |
| Up to 25 % of rated voltage | Borderline. Modern inputs filter this out, but older inputs may misread. |
| 25–50 % of rated voltage | Leaky off-state. Most inputs will misread this as ON. Mitigation needed. |
| 50 %+ of rated voltage | Severe leakage. Input definitely sees ON when switch is open. Mitigation mandatory. |
The bleeder resistor solution
A bleeder resistor connected across the input (between the input terminal and the input common) gives the leakage current a path to ground without forcing it through the input’s high-impedance sensing circuit. Sized correctly, the bleeder pulls the off-state voltage down to a clearly-OFF value (typically < 5 V on a 24 V system) without drawing enough current to overheat itself or load the source. Typical values: 5 kΩ to 10 kΩ for 24 V DC inputs, 22 kΩ for 120 V AC inputs.
A small upfront calculation prevents big field surprises
For a 24 V DC input with a sensor leaking 0.5 mA: a 10 kΩ bleeder pulls the open-circuit voltage to 0.5 mA × 10 kΩ = 5 V — comfortably below the input’s “on” threshold (typically 10 V). The bleeder draws (24 V)² / 10 kΩ = 0.058 W when the switch is closed — a 1/8 W resistor handles it easily. This kind of calculation is often the difference between a system that works in the lab but fails in the field, and one that works everywhere.
We’ll work through a real long-cable leaky-input scenario in detail in Case Study 2.
Grounding
“Ground” sounds simple — a wire connected to the earth — but it’s the topic that catches more installation engineers off guard than any other. The trouble is that there isn’t one ground; there are at least three, and confusing them creates the famous “ground loop” problem.
Three grounds, three jobs
- Earth ground — a literal connection to a copper rod driven into the soil. Job: discharge static, provide a fault path for safety (so that a short-to-chassis trips a breaker rather than electrifying the chassis).
- Chassis ground — the metal frame of the equipment, the panel walls, the I/O rack. Tied to earth through a low-impedance bonding conductor. Job: protect against shock if a live wire touches the metal.
- Signal ground (or “0 V”) — the reference for low-voltage logic signals. Often kept separate from chassis ground except at one carefully-chosen point. Job: provide a clean reference voltage for the signal sensing circuitry.
Single-point (“star”) grounding
The most reliable grounding topology is the star: every ground conductor — every signal common, every shield, every chassis bond — runs as its own dedicated wire back to a single ground bus bar in the panel, and the bus bar is bonded to earth at exactly one point. With this topology, no two ground conductors share a path, so no two grounds can develop a voltage difference between them.
Ground loops — the failure mode
A ground loop exists when a signal has two parallel paths to ground at different potentials. Picture an analog sensor whose enclosure is bonded to earth at the field, and whose 4–20 mA loop also connects to the panel’s signal common (which is bonded to earth at the panel). If the field earth and the panel earth are at slightly different potentials — they almost always are — a circulating current flows through the loop, and that circulating current shows up as noise on the analog signal. The fix is to break the loop at one end: float the signal common at the field, or use an isolated transmitter, or ground the cable shield at one end only.
A field engineer’s rule of thumb
If you measure between two “ground” wires that are both supposedly earth-grounded and you read more than about 50 mV AC between them, those grounds are not the same ground. They’re two separate grounds with a voltage difference, and any signal cable that touches both is a ground loop in waiting. The two clean-up moves are: (a) bond them together with a thick copper conductor (turning two grounds into one), or (b) isolate the signal that touches both (with an isolation transformer or opto-isolator). Don’t ignore the reading — it’s telling you something.
Voltage Variations and Surges
The supply voltage that arrives at your panel isn’t a perfect 230 V (or 120 V, or 480 V) sine wave. It’s a slightly distorted, occasionally-disturbed approximation. Understanding the four major disturbances and what each one does to a PLC is essential.
Four types of voltage disturbance
| Disturbance | What it is | Typical cause | Effect on PLC |
|---|---|---|---|
| Sag (dip) | Voltage drops to 70–90 % for a few cycles up to seconds | Large motor starts on the same supply | Possible processor reset or comms fault |
| Swell | Voltage rises to 110–130 % briefly | Sudden large load disconnect | Can damage power supply input filters |
| Transient (surge / spike) | Microsecond-scale voltage excursion to thousands of V | Lightning, contactor switching, capacitor banks | Damages I/O modules and processors |
| Harmonic distortion | Persistent non-sinusoidal waveform | VFDs, switching power supplies on the supply | Overheats transformers; rarely affects PLC directly |
Protection — three layers
- Surge suppressor (SPD). Installed at the panel’s power input. Uses MOVs (metal-oxide varistors) or gas-discharge tubes to clamp transient voltages to safe levels. Essential — and a sacrificial component that needs replacement after a major surge.
- Isolation transformer. 1:1 transformer between supply and panel. Breaks ground loops, attenuates common-mode noise, and provides some surge attenuation. Common in critical or sensitive installations.
- Uninterruptible Power Supply (UPS). Battery-backed inverter that keeps the PLC running through sags and short outages. Sized for the PLC plus its essential I/O, not the entire plant. A 10-minute UPS is sufficient for most PLC applications — long enough to ride through utility events and signal a safe shutdown if needed.
A common mistake — protecting the PLC but not its inputs
Surge protection has to extend to every wire entering the panel.
- An SPD on the AC supply protects the PLC’s power supply but does nothing for a 4–20 mA cable that’s been routed through a 100 m outdoor conduit during a thunderstorm.
- Each long signal cable entering the panel — especially analog inputs from outdoor sensors — needs its own dedicated surge suppressor, sized for the signal’s voltage range and impedance.
- The cost of a per-channel surge suppressor (~$30) is trivial next to the cost of replacing a fried analog module after a single nearby lightning strike.
Program Editing and Commissioning
Writing a program in the office and downloading it to a brand-new PLC on a bench is the easy half of the job. The harder half is deploying that program — or modifying it later — to a system that’s already in production, where mistakes have real consequences.
The commissioning sequence
For any new installation or major change, the disciplined commissioning sequence runs through three phases:
- Bench test. Program runs on a desk PLC with simulator inputs and lamp outputs. Every rung tested against expected behaviour. Edge cases (E-stop, power loss, sensor failure) deliberately triggered. Nothing leaves the bench until the test report is signed.
- Controlled startup. Program loaded into the field PLC, but actuators (motors, valves) are physically disconnected or air-locked. Inputs come from real sensors; outputs are observed but don’t actually move equipment. Operators verify each rung’s intent before authorising real motion.
- Full operation. Actuators reconnected. First runs are at reduced speed, with operators watching for unexpected behaviour. Full rated speed only after a clean run at reduced.
Online vs. offline editing
There are two modes for changing a program in a deployed system:
- Offline edit. Stop the PLC, edit the program on the laptop, download the new program, restart the PLC. Safe but disruptive — the line is down for the duration of the edit and download. Use this for any structural change (new I/O, new program structure, new tag).
- Online edit. Connect to the running PLC, modify a rung “live” while the program continues to execute, then commit the edit so the modification takes effect. Fast and non-disruptive — but also dangerous, because a typo in an online edit can crash the running plant. Use this only for small, well-understood changes (a setpoint adjustment, a timing tweak).
The pending → test → accept lifecycle
Modern programming software protects you from yourself with a multi-stage online edit workflow:
- Pending. Edit made on screen but not yet downloaded. Visible only to you on the laptop. Cancellable at zero cost.
- Test. Edit downloaded into the PLC’s working memory. Running alongside the original. The plant sees the edit’s behaviour, but the original code is still on disk. Cancel here and the test edit is discarded; the original keeps running.
- Accept (assemble). The edit is committed. The original is overwritten. From this point, only the new code runs.
The discipline is to verify thoroughly during the test stage and only accept once you’re sure. Case Study 3 walks through this entire workflow on a real plant.
Programming and Monitoring
Writing rungs is one part of the programmer’s day; watching them run is the other. The PLC programming environment is also a real-time monitoring console — and a strong programmer is fluent in both.
Online monitoring
When the laptop is connected to a running PLC, the program editor switches to a live mode where every contact, coil, and value updates in real time. True contacts are highlighted; false contacts are dim. Active rungs glow; inactive rungs are grey. Watching a program run live is the fastest way to understand its behaviour and to confirm that it’s doing what you intended.
Force tables
A force overrides the actual state of an input or output, telling the program “pretend this bit is ON regardless of what the field says”. Forces are the troubleshooter’s best friend — and most dangerous tool:
- Force an input ON to test what the program does with that input. Useful for testing rungs without manually triggering the real sensor.
- Force an output OFF to disconnect a misbehaving actuator from the program’s control while you investigate.
- Forces persist until removed. A force left in place is a landmine — the program is no longer reflecting reality. Most platforms warn you with a flashing “FORCES ACTIVE” indicator. Take the warning seriously.
Cross-reference and search
When you want to understand “everywhere this output is driven from” or “everywhere this bit is used as a contact”, the cross-reference tool tabulates every rung that references a given address. Indispensable for debugging multi-rung interactions.
Trends and histograms
For analog values and rapidly-changing digital states, most platforms include a trend tool that plots a value over time. Watching a sensor’s value trace out a curve makes it obvious when the value is noisy, bouncing, drifting, or stuck. Use this to verify analog scaling, to tune PID loops, and to capture the exact moment a fault occurs.
Preventive Maintenance
A PLC system that runs without intervention for a decade is the result of disciplined maintenance, not luck. Without a schedule, every system slowly degrades — battery dies, dust accumulates, terminals work loose, firmware falls behind on bug fixes — until one or more of those issues turns into downtime at the worst possible moment.
The annual checklist
- Battery replacement. Many older PLCs use a lithium battery to retain the program through power loss. Battery life is typically 3–5 years; replace on a schedule rather than waiting for the “low battery” alarm. Keep the new battery on hand before you remove the old one — without battery backup, an SLC-500 loses its program the moment power drops.
- Cleaning. Vacuum panel air filters and processor cooling fins. Dust is an insulator — accumulated dust traps heat, raising component temperatures, shortening lifespan. A 10-minute vacuum visit twice a year prevents thermal failures.
- Visual inspection. Walk the inside of every cabinet looking for: discoloured terminals (overheating), green or white powder (corrosion), wires pulled slightly out of terminal blocks (vibration), evidence of moisture, melted or cracked insulation. Most failures broadcast warning signs visible to a careful eye.
- Terminal torque check. Loose connections cause heating, intermittent failures, and eventually arc damage. Once a year, verify each power and signal terminal is tight to manufacturer’s specification using a calibrated torque screwdriver.
- Firmware status. Check vendor bulletins for firmware updates that fix known bugs or close security vulnerabilities. Don’t blindly update firmware on a working system — read the changelog, test on a bench unit first, and update only when there’s a specific reason. “If it’s not broken, don’t fix it” applies double to PLC firmware.
- Spare parts inventory. Keep at least one spare of every module type used in the plant on a shelf, ready to swap. Not in a “we’ll order one when we need it” loop — that’s a 4-week downtime if a board fails on a Friday.
Troubleshooting — Processor, Inputs, Outputs, Ladder Logic
Sooner or later, every PLC system breaks. The systematic technician finds the fault in 30 minutes; the unsystematic one is still poking at the panel three hours later. The discipline is the same one used in every electrical-system diagnostic: follow the signal from where it’s supposed to start to where it’s supposed to end, and identify the first place where reality stops matching the design.
The five places a fault can hide
Any failure of an output to respond correctly traces to one of five places, in roughly increasing order of how often it’s the culprit:
- The processor itself. Hardware fault, fault code in the status file, processor in fault mode and not running. Look at the processor LEDs first. If the RUN LED is off and FAULT LED is on, the rest of the troubleshooting is moot — clear the fault, then start over.
- The input module. Sensor working, wiring correct, but the input bit doesn’t reflect the field state. Module’s status LED for that channel tells you whether the module sees what’s at its terminals.
- The ladder logic. Input is fine, but a rung condition isn’t being satisfied. Online monitor highlights the broken contact.
- The output module. Coil is energised in the program, but the module’s output terminal isn’t switching the field side.
- The field wiring or actuator. Output bit is on, the module’s LED is on, voltage is at the terminal — but the actuator doesn’t move. Cable broken, fuse blown, contactor coil burned out, actuator mechanically jammed.
Figure 13.2 — The follow-the-signal troubleshooting flowWalk from input to output: at each step, ask “is reality matching the design?” The first step where the answer is NO is the location of the fault. We’ll work through this exact flow on a real “pump won’t start” scenario in Case Study 4.
Status LEDs are your friends
Almost every modern PLC module has front-panel LEDs that summarise its status — power, run, communicating, fault — and per-channel LEDs that show the state of each input or output point. Read the LEDs first. They tell you, at a glance, whether the module sees power, whether each input is reading high or low, and whether each output is being commanded on. A 30-second LED scan often pinpoints the fault before you’ve opened the laptop.
PLC Programming Software (RSLinx, RSLogix)
The Allen-Bradley/Rockwell ecosystem dominates PLC installations across North America and a large fraction of the world’s industry; their programming software is the de facto standard most engineers and technicians work with daily. Even if your plant uses Siemens, Schneider, or Mitsubishi, the workflow concepts are nearly identical — only the menus differ.
RSLinx — the communications layer
RSLinx is the driver/gateway that handles all communication between the laptop and the PLC. Before any programming software can talk to a processor, RSLinx has to be configured with the right driver — Ethernet/IP, DH-485, DF1, or whatever the network uses. RSLinx runs as a background service; once configured, the programming software discovers PLCs through it. RSLinx Classic Lite is free; full RSLinx Classic adds OPC server functionality for SCADA integration.
RSLogix 500 — for SLC-500 and MicroLogix
RSLogix 500 is the editor for the SLC-500 family (the platform used as the reference throughout this textbook) and the MicroLogix family. Programs are organised as a single ladder file with subroutines; addresses are file-and-element based (I:1/0, O:2/3, N7:50, F8:0). The data table is fixed-size and addressed numerically. This is the platform you’ve been seeing in every worked example in Chapters 5 through 12.
RSLogix 5000 / Studio 5000 — for ControlLogix and CompactLogix
Studio 5000 Logix Designer (formerly RSLogix 5000) is the editor for the modern ControlLogix and CompactLogix processors. The major differences from RSLogix 500:
- Tag-based addressing instead of file-and-element. Addresses are named (e.g.
StartButton,Pump1_Run) instead of numbered (I:1/0). The tag database lives in the project; tags can be structures with members, like a programming language struct. - Multiple programs and tasks. Where RSLogix 500 has a single ladder file with subroutines, Studio 5000 has multiple programs grouped into tasks, each with its own scan rate and priority. Closer to a real-time operating system model.
- Add-on instructions (AOIs). Custom reusable instruction blocks — like writing your own SQO-equivalent and then dropping it into rungs as a single instruction.
- Function block diagram (FBD) and structured text (ST) alongside ladder. The same project can mix all three IEC 61131-3 languages.
Don’t get hung up on which platform
Once you understand any ladder programming environment thoroughly, every other platform takes about a week to become productive in. The instructions are 90 % identical (every PLC has TON, CTU, ADD, SUB, MOV); the menus differ; the addressing schemes differ in cosmetics. Engineers fluent in RSLogix 500 transition to Siemens TIA Portal or Schneider EcoStruxure with little drama. Master the concepts in one platform, then treat new platforms as different tools for the same craft.
Worked Case Studies
Four Case Studies — Real-World Installation and Troubleshooting Scenarios
Four scenarios that every working PLC technician meets within their first year on the job: a noisy VFD upsetting nearby inputs, a long cable run with leaky off-state voltage, an online edit on a running production line, and a systematic input-to-output troubleshooting walk on a pump that won’t start.
The scenario: a packaging line has a single PLC controlling eight stations along a 40 m conveyor. A maintenance team has just installed a new 10 HP VFD-driven pump on an adjacent line, sharing the same electrical room. Over the following days, operators start reporting strange behaviour: occasionally a station’s “part-present” input flickers ON for a fraction of a second when no part is there, causing false counts. The flicker is intermittent and only happens during certain pump operations.
Symptoms reported
WHAT OPERATORS SEE
Random false counts on station 5 and station 6 (closest to the new VFD)
Counter values 5–10 % higher than visual inventory at end of shift
Issue started exactly when the new pump was commissioned
Worse during pump speed-changes, almost absent at steady-state
WHAT THE PROGRAM SEES
Online monitor shows I:1/4 (station 5 part-present) flickers ON for ~1 scan when no part is there
The flicker correlates in time with pump VFD acceleration ramps
No fault codes, no comms errors, no other anomalies
Diagnostic walk-through
- Suspect the program first? The technician opens the rung driving station 5 and finds it’s a single XIC of
I:1/4— just reading the photo-eye. No timing, no debounce, nothing programmable that could go wrong. The fault is upstream of the program. - Suspect the sensor? Visual inspection of the station 5 photo-eye finds it clean, properly aligned, and mechanically secure. Disconnecting the photo-eye and shorting its terminals at the panel also shows occasional flickers — so the photo-eye itself is fine; the issue is in the wiring between the photo-eye and the panel.
- Confirm the noise hypothesis with timing. The technician sets up an oscilloscope on the station 5 input terminal and instructs the operator to ramp the new VFD-driven pump from 30 % to 80 % speed. The oscilloscope captures clean transient spikes on the input line, perfectly correlated with the moment the VFD’s acceleration profile causes its largest dV/dt switching. Confirmed: VFD-induced noise on the signal line.
- Investigate the cable routing. Tracing the cables, the technician finds the photo-eye signal wires for stations 5 and 6 share a conduit with the new VFD’s motor cable for the last 8 m of their run to the panel. The two cables run parallel for that entire stretch.
- Apply mitigations. Three changes are made: (a) the signal cables are pulled and rerouted into a separate conduit that crosses the VFD cable only at right angles; (b) the signal cables are replaced with shielded twisted-pair, with the shield grounded at the panel end only; (c) the VFD’s output cable is replaced with VFD-rated symmetric shielded cable to contain the radiated EMI at its source.
- Verify the fix. The technician re-runs the same VFD acceleration test with oscilloscope still on the station 5 terminal. The transient spikes are now invisible — well below the input’s noise floor. Counters track inventory correctly over the next shift.
Why this case matters more than it first looks
The original installer didn’t do anything obviously wrong — they ran cables in a conduit, which seems fine. The problem only appeared when a noisy load was added later. This is why “good cable routing” isn’t a one-time decision; it’s a permanent design constraint. Every signal cable should be routed assuming a noisy load might one day be installed nearby. The cost of separation at install time (running an extra conduit, using shielded cable) is a fraction of the cost of finding and fixing this kind of fault later — especially when the symptoms are intermittent and easy to misdiagnose.
The scenario: a level switch is added to a remote storage tank 200 m from the existing PLC panel. The switch is a 24 V DC NPN proximity sensor. After the installation, the technician reports a strange behaviour: the input I:2/3 reads ON in the program even when the tank is empty and the switch’s actuator is fully retracted. Operators verify the switch is mechanically working — but the PLC won’t see “off”. The system was working fine on the bench with a 1 m test cable.
Symptoms reported
WHAT THE TECHNICIAN OBSERVES
On the bench (1 m cable): switch open → input OFF; switch closed → input ON. Works perfectly.
In the field (200 m cable): switch open → input still ON; switch closed → input ON.
The input never drops to OFF in the field.
Sensor LED on the switch correctly indicates open vs closed in both cases.
VOLTMETER READINGS
At the input terminal, switch open: 18.5 V (should be near 0 V)
At the input terminal, switch closed: 23.8 V (correct, near supply)
Input module’s “ON” threshold (per datasheet): 10 V
The 18.5 V open-state voltage exceeds the threshold → input reads ON.
Diagnostic walk-through
- The voltmeter reading is the smoking gun. 18.5 V at the input with the switch open shouldn’t be possible if the switch were a perfect open. The input is seeing significant voltage from somewhere.
- Two leakage paths to consider. The 24 V DC NPN sensor has internal leakage current — typically 0.3–0.5 mA — flowing through its solid-state output transistor even in the “off” state. On a long cable run, the cable itself has substantial distributed capacitance — about 100 pF per metre, so 200 m of cable presents 20 nF of capacitance to ground. The combination of small leakage current and high-impedance input creates the leaky off-state.
- Calculate the equivalent leakage. The technician disconnects the input cable from the input module and measures the voltage at the cable’s far end (the input-module side) with the switch open: still about 18 V. This confirms the leakage is upstream of the input module — coming from the sensor and the cable itself, not from the input.
- Apply the bleeder resistor. The technician installs a 5.6 kΩ, 1 W resistor between the input terminal and 0 V (the input common). Re-measure: input voltage with switch open drops to 2.8 V — well below the 10 V threshold. Switch closed still shows 23.5 V — well above threshold. The input now correctly reads OFF when open and ON when closed.
- Verify the resistor’s power dissipation. When the switch is closed, the bleeder dissipates
(24 V)² / 5.6 kΩ = 0.10 W. The 1 W resistor handles this with margin to spare.
A back-of-envelope check
| Configuration | Open-state voltage | PLC sees |
|---|---|---|
| 1 m cable, no bleeder | ~0.5 V (negligible cable capacitance) | OFF ✓ |
| 200 m cable, no bleeder | ~18.5 V (substantial leakage path) | ON ✗ |
| 200 m cable, 5.6 kΩ bleeder | ~2.8 V (leakage now drops across bleeder) | OFF ✓ |
| 200 m cable, 1 kΩ bleeder (too small) | ~0.5 V (good for OFF state) | OFF, but bleeder dissipates 0.58 W when closed |
Bleeder sizing — the trade-off
Smaller bleeder → lower open-state voltage (good for clean OFF) but higher power dissipation and load on the source. Larger bleeder → higher open-state voltage (risk of being above threshold) but minimal power. The sweet spot for 24 V DC is typically 4.7 kΩ to 10 kΩ, which keeps open-state voltage at a few volts (well below typical 10 V threshold) while dissipating well under 1 W when the switch is closed. Always size the bleeder for at least 2× the calculated dissipation as a safety margin.
The scenario: a high-temperature alarm on a curing oven is set to 185 °C. Process engineering has reviewed the data and approved a 5 °C reduction to 180 °C to give earlier warning of overheat conditions. The line is in continuous production and shutting it down for an offline edit would cost the plant approximately $50 000 in lost throughput per hour. The change has to be made online while production continues.
What’s being changed
BEFORE
Rung 47 in the OvenAlarm subroutine
GEQ Source A: F8:12 (oven temp), Source B: 185.0 → OTE Alarm_Hi (B3:5/0)
Trips at 185.0 °C
AFTER
Same rung structure
GEQ Source A: F8:12 (oven temp), Source B: 180.0 → OTE Alarm_Hi (B3:5/0)
Trips at 180.0 °C
Step-by-step procedure (as performed by the controls engineer)
- Pre-change sign-off. The change has been approved by process engineering and the plant manager, with documented approval ID. The MOC (Management of Change) form references this approval and lists exactly which rung will be modified, by whom, when, and what the rollback plan is. No live edits without written approval — ever.
- Backup the current program. Before connecting in edit mode, the engineer downloads the running program from the PLC to a versioned backup file:
OvenLine_v3.4_2026-05-08_pre-alarm-change.RSS. The filename includes the date and the reason. This backup is the rollback target if anything goes wrong. - Notify operators. A radio call to the line operator: “Going online to make the alarm setpoint change we discussed. The line will keep running. If anything looks wrong, call me back immediately and I’ll revert.” The operator confirms understanding and agrees.
- Connect online. RSLogix 500 is launched, RSLinx is configured, and the laptop connects to the PLC. The status bar shows “ONLINE — RUN” — the program is running and editable.
- Navigate to the rung. Open the OvenAlarm subroutine, scroll to rung 47. Verify with the engineer’s process documentation that this is the correct rung.
- Make the edit (PENDING stage). Double-click the GEQ instruction, change Source B from
185.0to180.0. The rung now displays with a small “PENDING” indicator — the edit exists on the laptop but has not yet been sent to the PLC. - Promote to TEST stage. Click “Test Edits”. The PLC accepts the modified rung into its working memory; the rung now executes with the new value alongside the original program. The plant is now operating with both the old logic stored and the new logic running.
- Verify in monitor mode. Watch the rung live. The current oven temperature is 178 °C (below both old and new thresholds). The engineer waits for the next normal-operation temperature peak (typically reaches 182 °C during baking phase). When the temp passes 180 °C, the alarm correctly trips — confirming the new setpoint is active and working as intended. (The alarm is silenced at the operator’s HMI; it doesn’t stop the line.)
- Promote to ACCEPT stage. Confident the test edit behaves correctly, the engineer clicks “Assemble Edits”. The new rung overwrites the original in the PLC’s program memory. The “PENDING” and “TEST” indicators clear; the edit is now permanent.
- Re-download (or save online) and back up the new program. The engineer saves the new version to the laptop:
OvenLine_v3.5_2026-05-08_post-alarm-change.RSS. This becomes the new master copy; the v3.4 backup is retained as the rollback target for the next 30 days. - Update documentation and notify. The MOC form is updated with completion timestamp. The operator is told the change is in. The change is logged in the plant’s controls-change register.
What you would NEVER do online
The hard “no online edit” rules — every disciplined plant has them.
- Safety logic. E-stop chains, machine-guard interlocks, fire-suppression interlocks. Plant down for the change; no exceptions.
- Structural changes. Adding new I/O modules, renaming tags, restructuring subroutine flow. These require offline edit and full retest.
- Anything you’d be unable to undo within 5 seconds if it goes wrong. If the rollback plan is “stop the line”, the change is too big for online editing.
- When you’re tired, stressed, or alone. Online edits are concentration work. Two engineers — one driving, one watching — is the gold standard for any non-trivial online edit.
The scenario: the morning operator’s first call: “Pump 3 won’t start when I press the START button.” The pump’s RUN lamp doesn’t come on; the pump motor doesn’t turn. The maintenance technician arrives and starts at the panel with a multimeter, an HMI, and a copy of the ladder logic.
Walking the five-step trail (Figure 13.2 in action)
The technician opens the laptop and connects to the PLC. The START button is wired to I:1/2. With the operator pressing and holding the START button, the technician watches I:1/2 in the online monitor.
| Test | Result | Conclusion |
|---|---|---|
START button pressed, watch I:1/2 | I:1/2 goes from 0 to 1 in the program | Input is fine ✓ — move to step 2 |
If I:1/2 hadn’t changed when the button was pressed, the fault would have been at or before the input module. Test would have been: voltmeter at the input terminal, with the button pressed, expecting +24 V. If voltmeter shows 0 V, the fault is in the field wiring or the START button itself; if voltmeter shows +24 V but the input bit doesn’t change, the fault is the input module. But here, the input is fine; we move on.
The technician opens the rung that drives the pump’s run output. The rung condition is:
rung 23: XIC I:1/2 (start) — XIO I:1/3 (e-stop) — XIC I:1/4 (cabinet door closed) — XIO B3:0/5 (overtemp latch) — OTE O:2/0 (pump run)
The technician watches the rung in monitor mode while the operator presses START. I:1/2 highlights (true). I:1/3 (e-stop XIO) does not block (e-stop is fine, NC contact is closed). I:1/4 (cabinet door XIC) does not highlight — the rung breaks here.
| Test | Result | Conclusion |
|---|---|---|
| Watch rung 23 with START pressed | Rung breaks at I:1/4 (door interlock) | Logic believes the cabinet door is open |
Cross-reference I:1/4 | Wired to door limit switch DLS-3 | Check the door interlock |
The technician walks to the pump’s local cabinet. The door is closed and apparently latched. But the door limit switch DLS-3 has a small alignment tab that’s bent — the tab isn’t quite engaging the limit switch lever. Visually, the door looks closed; electrically, the limit switch sees the door as open. The technician straightens the alignment tab; the limit switch lever clicks into its operated position.
Back at the panel, the technician watches I:1/4 in the monitor. With the door now properly engaging the limit switch, I:1/4 reads ON. The operator presses START. I:1/2 goes ON. All XIO conditions on rung 23 are clear. O:2/0 energises. The pump starts. The RUN lamp lights. The motor turns.
A summary of the trail
| Step | Hypothesis | Test | Outcome |
|---|---|---|---|
| 1 | Input dead? | Press START, watch I:1/2 in monitor | Input fine — move on |
| 2 | Rung not satisfied? | Watch rung 23 in monitor mode | Door interlock I:1/4 not satisfied |
| 3 | Field issue at door? | Walk to cabinet, inspect DLS-3 mechanically | Alignment tab bent — switch never closes |
| 4 | Fixed? | Straighten tab, retest | Pump starts normally |
Why the systematic walk wins every time
An undisciplined troubleshooter would have looked at the symptom (“pump won’t start”), gone straight to the pump motor, swapped the contactor, swapped the motor relay, and only after an hour of replacing innocent parts would they have asked “is anything blocking the rung?” The systematic walk found the real fault — a $0 mechanical alignment — in under 10 minutes. The trail is always the same: input → ladder → output → wiring → field. Walk it in order. Don’t skip steps. The first step where reality stops matching the design is your fault.
Common Field Mistakes
Things to watch out for in the field:
- Running signal cables in the same conduit as VFD or motor cables. Even a short parallel run is enough to couple noise into the signal. Always separate by at least 30 cm or use a dedicated signal-only conduit.
- Grounding cable shields at both ends. Creates a ground loop. Ground the shield at the panel end only; leave the field end floating.
- Bonding the PLC’s signal common to chassis ground at multiple points. Same problem — multi-point grounding creates loops. Bond at one carefully-chosen point only, typically the panel ground bus.
- Skipping the bleeder resistor on long-cable inputs. The bench works fine, but the field installation has a leaky off-state. Always measure open-state voltage during commissioning of long-cable sensors.
- Forgetting RC snubbers across contactor coils. Each unsnubbed contactor coil is a noise generator firing every time it opens. Ten unsnubbed coils make a panel impossible to commission cleanly.
- Online-editing without backing up first. If the edit goes wrong and you can’t remember exactly what you changed, the rollback is impossible. Backup before every connect, every time.
- Online-editing safety logic. Hard rule of every disciplined plant: safety logic changes are offline only, with full retest. The cost of a wrong online safety edit can be catastrophic.
- Leaving forces in place after troubleshooting. A force keeps an input or output in a pretend state regardless of reality. Always remove forces and verify the FORCES indicator is off before leaving the system.
- Skipping the systematic troubleshooting walk and “guessing”. Replacing parts based on hunches wastes time, money, and creates new problems. Walk the input-to-output trail every time.
- Updating firmware on a working system without a documented reason. “If it’s not broken, don’t fix it” applies double to PLC firmware. Update only when there’s a known bug or vulnerability that affects you, and test on a bench unit first.
Quick Recap
The ten things to take away from Chapter 13:
- Specify the NEMA enclosure rating for the environment first; layout, ventilation, and door interlocks follow from that decision.
- Inside the enclosure, separate power from signal: opposite sides, dedicated wireways, right-angle crossings only.
- Electrical noise from VFDs, contactor coils, and welders is mitigated by physical separation, shielded cable, twisted pair, ferrite cores, and RC snubbers at the source.
- Leaky inputs on long cable runs are diagnosed with a voltmeter; the fix is a bleeder resistor across the input terminals.
- Single-point (“star”) grounding with each conductor returning to one ground bus prevents ground loops; never bond shields at both ends.
- Voltage protection has three layers: surge suppressor, isolation transformer, UPS — and per-channel surge suppressors on outdoor signal cables.
- Online editing follows the pending → test → accept lifecycle; back up before connecting, document before committing.
- Forces, online monitor, cross-reference, and trends are the day-to-day tools for understanding a running program.
- Preventive maintenance is the difference between a system that runs ten years and one that fails in three: battery, cleaning, terminal torque, visual inspection, spare parts.
- Troubleshoot every “device won’t respond” symptom by walking the signal: input → ladder → output → terminal → field. The first step where reality stops matching the design is your fault.
Review & Self-Assessment
Chapter 13 Review Questions
Try answering each question on your own first. Tap to reveal the answer when you’re done.
Q1Which NEMA enclosure rating would you choose for a PLC panel installed in a food-processing washdown area, and why?+
Q2A panel layout has the PLC processor at the bottom of the enclosure and the VFD at the top. Why is this a problem and what should be changed?+
Q3Why is grounding a cable shield at both ends a mistake, and what’s the correct practice?+
Q4A 24 V DC input on a 200 m cable run reads 14 V at its terminal when the field switch is open. Is this a problem? What would you do?+
Q5What’s the difference between a surge suppressor and a UPS, and when do you need each?+
Q6Describe the pending → test → accept lifecycle of an online edit and explain why each stage exists.+
Q7What’s a “force”, and why is leaving a force in place after troubleshooting dangerous?+
Q8An operator reports a pump won’t start. Walk through the systematic troubleshooting steps in order.+
Q9What annual preventive-maintenance items would you put on the schedule for a PLC system?+
Q10What’s the difference in addressing between RSLogix 500 and Studio 5000?+
letter:file/element — for example I:1/0 (input file 1, bit 0), O:2/3 (output file 2, bit 3), N7:50 (integer file 7, word 50), F8:0 (float file 8, word 0), T4:0.DN (timer file 4, instance 0, done bit). Addresses are numeric and the data table layout is fixed. Studio 5000 (for ControlLogix and CompactLogix) uses tag-based addressing — each piece of data has a name (e.g. StartButton, Pump1_Run, OvenTemp), and the underlying memory layout is invisible. Tags can be structured (a tag of type “Motor” might have members like .Run, .Speed, .Current). Tag-based addressing makes programs much more self-documenting and easier to extend, but file-based addressing is simpler to learn and matches the historical PLC tradition. Most engineers learn file-based first (RSLogix 500) and then transition to tag-based (Studio 5000) as their projects scale up.Q11A panel is being commissioned in a basement near three other VFDs. The integrator is worried about noise. Beyond cable separation, what are three additional noise-mitigation measures you’d specify?+
Q12You’re called to a plant where a process has been running fine for 8 years and just started randomly faulting. The first replacement engineer suggested updating the PLC firmware to the latest version. Is this good advice?+
Need help with panel design, commissioning, or troubleshooting on a real industrial system?
Get one-on-one consulting from Dr Ahsan Rahman — Head of Electrical Engineering, with two decades of experience in industrial PLC installation, noise mitigation, and field troubleshooting across power, manufacturing, and process plants.