A variable frequency drive that trips is doing its job. It saw a condition outside a limit and shut the motor down before that condition turned into a burned winding, a fried output stage, or a shaft failure with someone standing next to the machine. The mistake I see most at two in the morning is treating the trip itself as the problem. It isn't. The trip is the drive telling you something happened. The fault code is the drive telling you what it saw. Your job is to read what it saw, work out why, and fix the cause, not open the parameter list and widen the limit until the drive stops complaining.
Write the code, the timestamp, and the conditions down before you touch anything.
This guide is organized the way the drive thinks: by fault class. Overcurrent means one family of causes, overvoltage means a different one entirely, and the code narrows your search before you ever pick up a meter. It applies whether the drive is on a conveyor, a pump, a mixer, a fan, or a blower feeding a digester at an organics or composting site, and I'll flag where a corrosive or gas-classified plant changes the answer.
Read the log before you touch anything
Every modern drive keeps a fault log with a timestamp and, on most platforms, a snapshot of conditions at the moment it tripped. That snapshot is the single most useful thing on the machine, and most people walk straight past it. Before you change one parameter, pull the log and answer four questions: the exact fault code and what this drive family's own manual calls it, since codes aren't universal between manufacturers; what the drive was doing when it tripped, accelerating, decelerating, or holding steady speed, because overcurrent on accel and overcurrent at constant speed point at different causes; whether it was loaded, from the current and torque in the snapshot; and whether there's a pattern, a drive that trips every afternoon in July is telling you about ambient heat, one that trips when three big loads start together is telling you about line sag.
Write the code, the timestamp, and the conditions down before you touch anything. Change a parameter first and you've destroyed the evidence. Now you're guessing.
What has to happen before the covers come off
Most of this diagnosis happens at the keypad, watching the display and the load, and that's fine. It stops being fine the moment you open the enclosure or probe line or motor terminals. A VFD's DC bus holds capacitors that carry a lethal charge well after power is removed, so off is not safe until you've waited the stated discharge time and verified the bus voltage yourself. Before hands-on work on the motor or anything it drives, apply full lockout/tagout to a verified zero-energy state; a drive can be commanded to start remotely over a network you didn't know was armed. And if it runs a blower or mixer tied to a digester, that space is very likely permit-required confined space, with hazards a bump-tested four-gas monitor exists to catch.
Overcurrent trips are fast, and usually mechanical or electrical, not a ramp problem
An instantaneous overcurrent trip means the output current crossed a hard, fast limit, well above the motor's full-load amps, in milliseconds, the drive protecting its own output transistors. Work the causes in order. A short or ground fault in the motor cable or windings is the safest-first assumption, because it's the most dangerous to ignore; if the drive trips the instant it fires, before the motor even turns, suspect the cable or motor first. A mechanical bind or jam, a jammed conveyor, a seized bearing, a pump against a closed valve, spikes current as the motor tries to break the load loose, and usually trips right at the start of acceleration. Too-fast acceleration is real too, and it's also the cause people abuse, lengthening the ramp until the trip stops when the actual problem was a bind. Lengthen the ramp only after you've ruled out a mechanical cause.
With the drive isolated, LOTO applied, and the bus verified dead, disconnect the motor leads and check insulation with a megohmmeter, and turn the shaft by hand to feel for a bind before you touch a single parameter.
Overvoltage and undervoltage are usually about where the energy is going
The DC bus normally sits around 1.35 times incoming line voltage, so a 480V line gives a bus near 650V DC. Overvoltage means something pushed it higher, and the classic cause is regeneration on deceleration: slowing a spinning load turns the motor into a generator that pumps energy back into the bus, and high-inertia loads like big fans or flywheels do this hard. With nowhere to put that energy the bus rises and trips, almost always on decel, which the fault snapshot will show you; a missing or failed dynamic braking resistor is the usual reason. Undervoltage runs the opposite direction: a supply sag when a big load starts elsewhere, a loose input lug heating up under load, or a lost phase from a blown fuse or contactor pole. Measure all three incoming phases under load; a phase reading low or zero means you've lost it, and heat discoloration on a lug is a quiet but common cause of a "random" trip.
Widening a trip doesn't fix the cause. It just moves the failure downstream to something more expensive, and on a bad day that something is a person standing next to the machine.
Ground fault means isolate the drive before a megohmmeter touches anything
A ground fault trip means the drive detected output current leaking somewhere it shouldn't, the three phase currents no longer summing to zero. The usual cause is insulation breakdown in the motor or cable, or moisture bridging what's left of it, one of the most common real ground faults I find in wastewater and biogas plants.
Here's where people damage equipment doing the test wrong: isolate the drive and disconnect the motor leads at the output terminals before you megger, since a megohmmeter applies hundreds of volts DC and will destroy the drive's electronics if it's still connected. Megger each motor lead to ground at the voltage appropriate to the motor. A healthy motor and cable read high, typically many megohms; a reading in the low kilohms, or one that drops while you watch, is a ground fault. If it megs clean at the motor but bad with the cable attached, dry it and find the water path rather than resetting and running, since a ground fault in a classified gas area is an ignition source.
Overtemperature and motor overload are both thermal, but protect different things
Drive overtemperature means the heatsink or internal sensor crossed its limit, the power devices getting too hot, and the drive protecting silicon you can't replace at 2am. Blocked cooling, clogged filters, dust-caked heatsink fins, is the number one cause in dusty plants, and it's entirely maintenance. A failed cooling fan, high ambient heat in a sealed enclosure, or a carrier frequency bumped up to quiet motor whine round out the list.
Motor overload is a different, slower trip: the drive's thermal model decided the motor has drawn too much current for too long. Most of the time that's telling the truth, a dragging belt, a misaligned coupling, a worn bearing, product buildup, a pump running off its curve. Sometimes it's wrong motor parameters: verify the programmed full-load amps against the nameplate, a legitimate correction, not the same as raising the overload class to make the trips stop. Do that and you cook a winding.
Loss of comms or reference is a signal problem, and the shield is usually the cure
This trip means the drive lost whatever was telling it what to do, a network command over a fieldbus or an analog 4-20 mA reference. A dropped fieldbus, a failed switch, or a PLC that stopped scanning shows up in the drive's comms status LEDs and error counters if it exposes them; I go into that side of the fault in more depth in the field method for chasing comms dropouts. On the drive itself, a noisy or lost analog reference usually comes down to a broken wire or a shield grounded wrong, and the cure is almost always the shield: land it at one end only, usually the drive or panel end. Land it at both ends and you build a ground loop that injects the very noise you're trying to keep out; land it at neither and it does nothing at all.
A ten-minute triage that works from the code down
Work top to bottom and stop as soon as the code tells you where to go. First, read the fault and the log: the exact code, the timestamp, and whether there's a pattern. Then classify it, current, voltage, thermal, or signal fault, since each sends you a different direction, and spend a minute looking and listening at the keypad: hot enclosure, dead fan, comms LEDs, a burnt smell. Take one safe measurement that fits the class, and for a current or ground fault, don't meter live output; plan the isolate-and-megger instead. Then decide the safe next step rather than the quick reset. Never close out a VFD trip by widening the trip. The motor, the drive, or the machine pays for that later, and in a gas or dusty plant, later can mean a fire.
I see this pattern constantly on crusher and conveyor drives across quarrying and aggregates plants, and nuisance trips add up fast in stopped-line minutes; our OEE and downtime cost estimator puts a number on what that costs across a year. And if the drive is running a lift-station pump rather than a conveyor, the trip pattern is often tangled up with duty-cycling logic worth its own look, covered in the money hiding in lift-station control logic.
On a VFD, the ugliest faults are the ones that get handed between the electrician who owns the panel, the motor shop that owns the winding, and the integrator who owns the network, and each one tests their piece, finds it clean, and hands it back. Reading the same fault log across all three domains, vendor-neutral, usually finds the seam they were passing over. That's the fault I get the most calls about.
The printable field version of this guide, including the full symptom-to-check reference table, is free in the resources library. If your drive keeps tripping and the fault log is telling a story nobody's connected yet, get in touch and we'll read it together.