Most bad-switch calls are not bad switches. I'm a controls engineer, not a belt salesman, so I'll tell you what I actually see: the switch is doing its job. Something upstream moved the belt, loaded it off-center, or let a return idler seize, and the protection device is just the messenger. This is written for whoever's standing at the drive with the line down at 2am, deciding whether to reset and run or lock out and fix it. When in doubt, lock it out.
A quick word before any of this, because conveyors hurt people. Full lockout/tagout before hands-on work, no exceptions for cleaning just one idler. On an inclined belt, the load and belt want to run backward the moment you release a brake or relax a takeup, so chock and block before you get near a gravity counterweight or a holdback you're servicing. Guards go back on before restart, and a pull-cord or e-stop is a safety function you repair, never defeat. Reclaim tunnels, transfer towers, and MRF pits are commonly permit-required confined spaces with real engulfment and atmosphere risk, so know which of your spaces are permit-required before anyone goes in. Everything past this point assumes the belt is either locked out for hands-on work, or you're observing it running from a safe position outside every guard and nip point.
Why belts drift, and the cardinal rule for fixing it
A belt runs off to one side because the forces across its width aren't balanced, and it keeps chasing that imbalance until something stops it, a frame, a drift switch, or its own edge shredding on structure. The cardinal rule: a belt steers off whatever it touches first as it approaches, which is why you correct tracking working in the direction of travel, and why yanking one idler over is almost always the wrong move. You've treated a symptom two frames downstream of the actual cause.
A belt steers off whatever it touches first. That's why you correct tracking working in the direction of travel, and why yanking one idler over is almost always the wrong move.
Off-center loading is the number one cause in aggregate and MRF work: if the chute dumps material to one side of centerline, that side is heavier, the belt cups toward it, and it walks. On MRF lines a surge of heavy mixed load, a wet cardboard slug, a chunk of C&D debris, does the same thing intermittently, so the belt tracks fine empty and drifts only under load. Material buildup on idlers and pulleys is close behind: a quarter inch of packed fines or film on one end of a pulley effectively makes that end a bigger diameter, and the belt climbs toward or away from the built-up side depending on where it packs. This is the single most common intermittent tracking problem, and cleaning the return run fixes more tracking calls than adjusting anything.
A seized idler stops turning entirely, so the belt drags across stationary steel, wears a flat, grabs unevenly, heats, and steers, and in a dry dusty tunnel a locked roll polishing against a moving belt is a real ignition source. Spin every return idler by hand, LOTO first, and mark the dead ones. A racked or settled frame throws idlers out of perpendicular to belt travel and the belt walks toward the low corner, common after ground settling or a forklift hit, and it masquerades as a chronic one-direction drift that no amount of idler tuning ever holds. And know which of your pulleys are supposed to be crowned: a crowned pulley self-centers the belt, and if one got swapped flat during a repair, or the crown wore off a rubber-lagged pulley, you've lost the self-centering.
The methodical order for fixing tracking
Do this with the belt running empty first, then loaded, one change at a time, and give it several full revolutions to respond before you judge. A long overland belt can take minutes to settle. Clean first, the return run, pulley faces, and idlers, and fix the belt cleaner if carryback is feeding the buildup; a shocking number of tracking problems end right here. Fix loading next, center the chute, repair skirting, check for a plugged deflector. Check the structure, square and level the frame before touching a single idler. Replace what's dead, seized idlers, worn-flat rolls, a flat-spotted crowned pulley.
Only then tune with idlers, working in the direction of travel, advancing the carrying idler on the side the belt is running toward. Self-aligning training idlers can help but they mask root causes; don't install one to avoid fixing loading or structure. If you find yourself cranking a single idler hard over to hold the belt, stop. You're fighting a cause you haven't found, and it'll tell on you the next time the load surges.
The protection devices that pull the interlock, and get blamed for it
Every one of these exists to stop the belt before something worse happens, and nine times out of ten a "nuisance trip" call is the device catching a real condition the plant would rather not deal with. Drift switches are roller-actuated devices at the belt edge; a trip means the belt moved that far, which is real, and the fix is the tracking work above, not bending the switch bracket outward so the belt has to travel farther before it trips, which just lets the belt shred its edge on structure before the switch ever sees it. If the switch trips with the belt visibly centered, suspect the switch itself, a bent actuator arm or a corroded pivot.
Pull-cord and e-stop switches are a safety function, full stop. You never defeat, widen, or jumper them. If one's nuisance-tripping, the honest causes are rope tension out of spec, a corroded switch, or a fouled rope, and every one of those is a repair. Zero-speed detection watches that the belt is actually moving at commanded speed, and it trips on real slip, a sheared coupling, or a stalled belt under a plugged chute; belt slip on a drive pulley polishes the lagging and can start a fire, and a slipping drive on an incline can let a loaded belt run backward against the holdback and catch with a shock load, so check lagging and takeup tension before you ever suspect the sensor.
Plugged-chute switches ask one honest question: is the chute actually plugging? On aggregate transfers wet fines and clay do exactly that. If the chute is genuinely clear, suspect buildup on a capacitance probe reading caked dust as material, and don't crank the sensitivity down until it stops calling, you'll miss the plug that buries a transfer tower. Belt-rip detectors are the highest-stakes call you'll make: assume a trip is real and inspect the belt before you call it false, because a rip caught at two feet is a patch and a rip missed becomes a full belt replacement and a tower full of spilled material. And a motor overload trip is often the end symptom of everything above, a seized idler dragging, a plugged chute loading the belt, or simply too much material, so before you reset it more than once, go find the load.
Telling a real trip from a nuisance trip
This is the discipline that separates a good troubleshooter from a switch-widener. Does the physical condition exist right now, belt off center, chute backing up, belt not at speed? Look before you judge the switch. Does it trip under a specific condition, a load surge, wet material, a cold morning, one particular product? That's the device catching a real intermittent, so chase the condition, not the device. Does it trip with the belt visibly fine, empty, and clean? Now you've earned the right to suspect the device, its actuator, wiring, or a corroded pivot.
And log every trip, time, load state, material, weather, and what you found. Three logged trips with a pattern tell you more than any single reset, and a trip log is the cheapest diagnostic tool in the plant and the one nobody keeps.
Why corrosive and dusty plants eat good switches
Aggregate plants are dust and grit, MRFs add humidity, rot, and mildly corrosive leachate, and switches don't fail because they're cheap, they fail because of where they live. Caked dust on a capacitance or proximity sensor reads as material, which is the classic false plugged-chute and false zero-speed trip, so clean the sensor face on a schedule rather than chasing the setpoint. Corroded pivots and actuator arms on drift and pull-cord switches stiffen or lose their return spring until the switch won't trip or won't reset cleanly.
Water ingress through a failed seal corrodes contacts and causes intermittent open circuits that look exactly like nuisance trips, so check that enclosure ratings actually match the washdown exposure. And vibration backs off actuator setscrews over months, so a switch set correctly last year trips early or late this year; re-verify switch positions during preventive maintenance, not just after failures. If a plant has a chronic nuisance switch, it's usually the right kind of switch in the wrong enclosure, on a corroded pivot, or reading buildup. Fix the environment problem and the nuisance stops.
If drift and nuisance trips are eating into your tonnage on top of everything else, it's worth reading alongside where the missing tons on a crushing plant actually hide. And if you want to see what these stops are costing you in tons per shift rather than just downtime minutes, the crushing plant throughput and utilization estimator does that math.
Collected in one place so they're easy to refuse: bending or relocating a drift switch, jumpering a pull-cord, widening the underspeed window to ignore slip, turning down chute-switch sensitivity to silence a real plug, disabling a rip detector, and resetting a motor overload on repeat without finding the drag. Every one trades a cheap, correct repair for an expensive, dangerous failure later.
The printable field version of this guide, with the full symptom-to-cause-to-check reference table, is free in the resources library. Conveyor faults are some of the worst to hand between vendors, the belt guy blames controls, the controls guy blames the belt, and the switch keeps tripping in the seam between them. If that's where you're stuck, get in touch and we'll find the actual condition it's catching.