PLC Wiring Mistakes: The 10 That Destroy Your Panel
A panel working on the test bench doesn't mean it's wired correctly.
We see it constantly in technical support: the integrator tests everything in the workshop, and it's perfect. They install it on the plant floor, and random reboots start, analog readings drift, and communications drop out "sometimes." 90% of those "sometimes" is wiring. The workshop is an electrically friendly environment — no drives nearby, no contactors switching, no hundreds of meters of cable acting as an antenna. The plant floor is exactly the opposite.
These are the 10 wiring mistakes that reach our support desk most often, with the physics behind each one and the fix. Number 10 deserves special attention: it's the direct cause of most of the burned-out PLCs we receive for repair, and it happens in the first minute of the installation.
Mistake #1 — Inductive loads without suppression
A contactor, a solenoid valve, or a motor brake are coils: they store energy in their magnetic field. When you disconnect them, that energy has to go somewhere, and it does so as a voltage spike of hundreds of volts (the induced voltage is proportional to how fast the current is interrupted). Without suppression, every switching event is a small lightning bolt inside your panel: arcing at the contacts that erodes them, plus radiated and conducted transients that corrupt communications and damage PLC outputs.
The fix, by coil type:
- DC coil: flyback diode in antiparallel, as close to the coil as possible. If actuator drop-out time matters (brakes, fast valves), a diode in series with a zener speeds up demagnetization.
- AC coil: RC network (snubber) or a varistor sized to the switching voltage. A diode doesn't work here — the current is alternating.
Golden rule: suppression gets mounted at the load, not at the contact. The goal is for the transient to never even reach the wiring.
Mistake #2 — Signal and power in the same wireway
Motor cables — especially a drive output, which switches at several kHz with very fast edges — running alongside 4-20 mA or 0-10 V analog signals is coupling noise by design: capacitive and inductive coupling between parallel conductors grows with the length of shared run.
The fix: separate wireways and trays for power and signal, with the maximum distance the panel allows. When a crossing is unavoidable, make it at 90°: a perpendicular crossing minimizes the coupling length. And inside the panel, the same discipline applies: routing signal wiring down one side and power down the other isn't aesthetics — it's EMC.
Mistake #3 — Shield mesh: floating or connected without a plan
Two ends of the same mistake. A shield mesh connected to no potential at all shields nothing: it's just another conductor floating next to your signal. And a mesh connected at both ends without a decent equipotential network creates a ground loop: the potential difference between the two grounds drives current through the mesh, and that current injects noise into the very signal you were trying to protect.
The practical rule for analog signals: ground the shield at one end only, the panel end (where the control system reference lives). Communication buses and drive motor cables follow different criteria — bonding at both ends with a verified equipotential and 360° connection — but if you're unsure about an analog signal, one end is the defensible choice.
Mistake #4 — Electronics 0V and field 0V mixed carelessly
Sharing the same 24V source between the PLC's CPU and solenoid valves that draw inrush current is an invitation to micro-resets: every load startup causes a voltage dip that the electronics reads as a micro-reset or corrupted data. The classic symptom: "the PLC reboots right when the compressor starts."
The fix, in order of preference: separate power supplies for electronics and for field/actuators; if there's no budget or space for that, at minimum size the supply for the inrush current of all simultaneous loads — not the average consumption on the label — and wire the CPU's power directly from the supply terminals, not at the end of a chain of loads.
Mistake #5 — Analog signals on untwisted cable
A 4-20 mA loop forgives a lot — that's the point of it: the information rides on the current, which is identical throughout the loop, and noise induced as voltage barely affects it. But a 0-10 V signal on 15 meters of parallel cable next to a drive is an antenna: the signal is the voltage, and every induced millivolt is a direct measurement error.
The fix: twisted, shielded pair, always, for every analog signal — the twist makes induced noise cancel between the two conductors. And if you can choose the signal type at the source: current loop before voltage, and digital (Modbus RTU) before either for long distances.
Mistake #6 — RS-485 wired as a star
The RS-485 bus is a transmission line: one single path, daisy-chained node to node, with its two 120 Ω termination resistors at the physical ends — and only at the ends. Every star branch creates an impedance discontinuity, and every discontinuity generates reflections: the frame and its echo overlapping. The result is intermittently corrupted frames — the worst kind of fault, because you can't reproduce it: it depends on which slave was talking, at what speed, and the length of each branch.
The fix: line topology with no exceptions (stubs as short as possible), termination at the two real ends of the bus, twisted shielded pair, and a third conductor as a common reference between nodes. If the plant layout forces a branched topology, RS-485 repeaters exist that isolate each branch as its own bus.
Mistake #7 — Trusting that "the terminals are already tight"
A loose terminal under vibration follows an inevitable sequence: intermittent contact → micro-arcing → oxidation and local heating → more resistance → more heating. Over time it produces a fault that will look like software ("we lose comms sometimes") or, in power circuits, a hot spot you'll see on a thermal camera before you see it as smoke.
The fix: torque per manufacturer spec (yes, terminals have a specified torque), crimped ferrules on every flexible cable — never tinned wire, which flows and loosens over time — and a scheduled re-torque after the first weeks of real vibration. Spring-cage (push-in) terminals eliminate the problem at the root and are the preferred choice wherever there's vibration.
Mistake #8 — Ground as an afterthought: no real equipotential bonding
The protective conductor daisy-chained terminal to terminal, the panel chassis without a proper connection, the door without its grounding braid… Grounding a control panel has two distinct jobs: safety (making sure a fault trips the protections) and EMC (giving high-frequency currents a low-impedance path). A ground that "already continues somewhere" can satisfy the first and fail badly at the second: at transient frequencies, what matters isn't resistance but the inductance of the path — length and geometry.
The fix: an equipotential bar or plate in the panel, with every ground connection arriving in a star with a short, generously sized conductor (flat braid beats round wire at high frequency), chassis and mounting plate with clean metal-to-metal contact (star washers on scraped paint), and cable shields reaching that reference by the shortest possible path.
Mistake #9 — Everything hanging off the supply with no per-circuit protection
The 24V supply feeding the PLC, sensors, valves, and HMI directly, with no individual fuses or circuit breakers per group. The day a field sensor gets crushed or a cable chafes against the chassis, the short circuit doesn't take down one circuit — it takes down power to the entire panel, PLC included. The whole machine goes down because of a €12 limit switch, and diagnosis is slow on top of that because nothing points to the guilty circuit.
The fix: 24V distribution protected by group — control electronics, inputs/sensors, outputs/actuators, communications — with fuses or electronic protectors per channel, sized selectively: the field fault should clear its local protection before the supply hits current limiting. Electronic 24V protection modules with remote signaling also turn every trip into an instant diagnosis.
Mistake #10 — The PLC killer: badly wired 24V power
Here's the one we promised: the direct cause of most of the burned-out PLCs that reach our repair service. And it happens in the first minute of the installation, with the first two wires connected: the power supply.
Three variants of the same mistake:
- Reversed polarity: +24V and 0V swapped. Two same-colored wires in the drawer, a bit of a rush, and there you go.
- Wrong voltage: powering a 24Vdc input with 48Vdc from another rail in the panel, or directly with 24 Vac from a control transformer — which, after the device's internal rectification, becomes a peak voltage far higher.
- Sustained overvoltage: a cheap power supply with its adjustment trimmer turned up "to compensate for the drop," or transients on a shared 24V rail from unsuppressed inductive loads (see mistake #1).
Why is it so lethal? Because the power input is the one point on the device connected directly to all of its electronics. A mistake on an optocoupled digital input burns, at worst, that one channel. A mistake on the power input attacks the entire regulation stage of the device. Well-designed industrial equipment builds in protection at this stage — the M-Duino 21+, per its official User Guide, accepts 12 to 24 Vdc input and includes fuse protection (2.5 A) and polarity protection at the input — but no protection is an excuse for skipping verification.
The fix (a ritual, not optional): verify polarity and voltage at the PLC's terminals with a multimeter before connecting it — at the terminals, not at the supply: something can cross in between. Standardized, respected wire colors for +24V and 0V throughout the installation. A quality industrial power supply with output voltage verified. And if the device allows it, a dedicated external fuse on its power line: it costs pennies and turns a wiring mistake into a blown fuse instead of a repair.
Quick checklist before powering up
- 24V polarity and voltage measured at the PLC's terminals ✔
- Every coil (contactors, valves) has its suppressor at the load ✔
- Power and signal in separate wireways; crossings at 90° ✔
- Analog shields grounded at one end only (panel side) ✔
- RS-485 wired in line, two 120 Ω terminations, common reference ✔
- Supplies (or selective protections) separating electronics from field ✔
- Star equipotential bonding, chassis with clean metal contact ✔
- 24V distribution fused by group ✔
- Torque values applied and re-torque scheduled ✔
- Analog signals on twisted shielded pair (current loop where possible) ✔
Ten minutes of checklist versus weeks chasing intermittent faults. The math always works out.
📥 Download the checklist: "From Arduino prototype to industrial installation"
The complete 12 steps — wiring included — for taking your project from the workbench to the electrical panel, in printable PDF.
Download the free checklist →
M-Duino PLCs are designed to survive environments where one of these mistakes has already been made: optocoupled inputs, isolated outputs, and a power stage with fuse and polarity protection. See how each I/O block is protected in the M-Duino 21+ technical documentation: M-Duino 21+ technical features (product: M-Duino 21+ Ethernet PLC).