Servo Drive & Motor Troubleshooting: From Fault Code to Replacement Decision

Servo faults fall into four layers: power, feedback, control parameters, and mechanical load. Work them in the order safety, simple checks, then isolation, and you will locate most failures without guessing. This guide covers the four layers, how to read servo drive fault codes correctly, how to decide whether to repair or replace, and how to source the right unit when the model is discontinued.

Find your symptom first.

If you smell burning or see smoke, go straight to Symptom 6. Do not power up again to "confirm" it. Before you test anything, read the next section.

1. Before You Touch Anything: Safe Isolation

A servo drive stays dangerous after the main switch is open. DC bus capacitors hold charge, and discharge time depends on capacitance and the bleeder resistor. Here is the part people get wrong: when the panel LED goes dark the bus may still be live, because that LED runs off the control supply. Measure the bus.

Five step isolation

1. Open the main disconnect and remove control power.
2. Apply lock out tag out, one lock per person working in the cabinet.
3. Wait for bus discharge, using the time printed in the drive manual, not a number you remember from another machine.
4. Verify with a meter on the DC bus terminals, below the threshold the manufacturer gives. Indicators are not verification.
5. Support any vertical axis before disconnecting the motor. The holding brake releases when it loses power or when you unplug its cable, and a loaded Z axis will drop. Lower it to its bottom stop or block it mechanically. Dropped axes wreck tooling more often than they damage drives.

Tools, and what each one proves

One habit saves hours: read the fault history, not just the code on screen. The first stored event is usually the real one.

2. The Four Layer Isolation Method

Every check below answers two questions: what reading is abnormal, and what does it rule out?

Layer 1, power and wiring. Check three phase voltage and imbalance, phase loss, motor cable insulation to ground, shield grounding at one end only, and earth continuity. Compare deviations against the drive manual. A clean result rules out supply and cable damage. Watch shared cabinets: a variable frequency drive starting a big load on the same supply can pull the servo bus low enough to trip undervoltage, and its output cable in the same tray couples noise into feedback lines.

Layer 2, feedback. Confirm the encoder type and that the drive is configured for it. Inspect the connector for oil, moisture, and oxidation, check the cable where it flexes in the drag chain, and check the backup battery on absolute encoders. Fast proof: move the feedback cable to an identical spare axis and see whether the fault follows it.

Layer 3, control and parameters. Many "servo faults" are not in the servo. Verify the enable chain first: safety relay, e-stop string, STO terminals, and the enable bit from the controller. Then check control mode, torque and speed limits, and ramp times. Ask who worked on the machine last, because parameter changes from an earlier repair are a common root cause with no alarm attached. If the PLC or motion module never issued a move command, the servo is correct to stand still, and an operator change made on the HMI may appear in no log.

Layer 4, mechanical and load. Look for a loose coupling, misalignment, worn bearings, a jammed rail or ball screw, chips, dry lubrication, an unreleased brake, and inertia ratio out of range. The cheapest test in this guide: uncouple the motor and run it free. If the fault disappears, motor and drive are fine and the problem is downstream.

Swap testing: proving the fault instead of betting on it

Swapping is fast, and it is also how people destroy a second drive. Each test has a precondition.

• Swap the motor. Proves motor side against drive side. Precondition: the replacement matches the loaded parameter set, including encoder type.
• Swap the drive. Proves the same from the other end. Precondition: confirm there is no short in windings or cable first. Skip that and you burn the second drive in seconds.
• Uncouple the load. Separates motor from machine at almost no risk. Do this before either swap.
• Cross swap two identical axes. Fastest answer, highest risk, because a damaging fault now has two victims.

Do not swap test at all if you smell burning, see discoloured terminals, or measure failed insulation. The answer there is already replacement, and the question is what caused it.

In short: identify the failed layer before spending money. That alone prevents most wrong purchases.

3. Reading Servo Fault Codes

Codes differ by brand because each maker chose its own numbering. The physical faults do not. So map your code to a fault family, then troubleshoot the family. Searching the code string alone sends you to threads about a different firmware version.

Where to look up your code. Take the model and firmware version from the nameplate, then use the manual matching that version. Siemens SINAMICS separates faults and alarms by letter prefix, Allen-Bradley Kinetix uses grouped numeric codes with sub codes, and Schneider Lexium, Mitsubishi MELSERVO, Omron, and ABB each use their own scheme. The same number can mean different things across firmware releases, so check the exact model specifications before trusting a code table.

Codes that lie

The displayed code is a symptom, not a diagnosis. Four cases where it names the wrong part:

• An intermittent encoder signal makes the drive read a wrong position, so it commands full current. Screen says overcurrent. Fault is feedback.
• A jammed axis draws rated current continuously. Screen says overload or overtemperature. Fault is mechanical.
• A brake that never fully releases makes the axis lag behind command. Screen says following error. Fault is the brake circuit.
• A chain of alarms fires in milliseconds and the display keeps the last one. The first fault history entry is the one that matters.

Rule: if you replaced the part the code named and the fault returned, go one layer back and start again.

In short: a fault code gives you a direction, not a part number. Replacing parts by code is the most common way to waste money on a servo.

4. Symptom by Symptom

Symptom 1, motor will not turn or has no torque. Check three non-faults first: the brake never released, the enable signal never arrived, the controller never sent a command. Measure brake release current, read the enable state in the drive software, read the commanded position in the controller. Replace only when insulation fails, phase resistances are unbalanced, or the output stage is damaged.

Symptom 2, motor overheating. Measure at the flange and the case centre, and compare against the motor datasheet, not a general number. Heat with no load points to a dragging brake, wrong commutation or encoder configuration, or blocked cooling. Heat only under load means the duty cycle exceeds the rating, a sizing problem.

Symptom 3, noise and vibration. Low growl means load or bearing drag, dry rubbing means bearings, high whine means gains too high. Lower the gain: if the noise fades, it is tuning. If it persists, uncouple the load. Still shaking means motor or mounting, quiet means coupling, alignment, or machine.

Symptom 4, following error. Actual position fell outside the allowed window around commanded position. Error growing with speed points to ramps and gains. Error growing with load points to capacity, drag, or an unreleased brake. If the torque trace saturates at the limit, the motor is undersized for the machine's current condition.

Symptom 5, position lost after power cycle. Almost always the absolute encoder battery. Replace it with control power on where the manual allows, because changing it with everything dead can wipe multi turn data and force re-homing. Put the battery on the annual list.

Symptom 6, smoke, burning smell, or ozone. Stop. Do not re-energize to confirm. Isolate and treat the unit as scrap. Then find out why it burned, because a shorted cable, failed brake circuit, or supply problem will destroy the replacement too.

Symptom 7, intermittent faults. These are the hardest, so use timing instead of the meter. Motion-linked faults point to cable flex fatigue in the drag chain. Contactor-linked faults point to EMI, often from a VFD sharing the tray or a shield grounded at both ends. Faults appearing only after hours of running point to thermal drift. Correlate fault history timestamps with the machine cycle, and trend cabinet temperature on the operator panel.

In short: symptoms point to a component, but only measured values convict it.

Diagnosis ends here. What follows is a spending decision.

5. Repair or Replace

Four numbers decide it. First, cost per hour of downtime: lost output plus labour plus penalties or expedited freight. Second, the repair quote and, more importantly, the repair turnaround in days. Third, price and lead time of a new unit. Fourth, remaining machine life and whether the model is still available. Multiply repair turnaround by hourly downtime cost before comparing prices. That step reverses many decisions, and it differs sharply between a continuous line and a standalone machine. See our industry solutions.

When repair is genuinely better. If the damage is peripheral, if a qualified repair channel turns it around inside your tolerance, if the machine is near retirement, or if the axis must stay identical to others for parts commonality, repair is the better call.

Why the same part keeps failing. Four causes cover most repeats. Cooling: cabinet temperature, blocked filters, dead fans. Sizing: inertia ratio pushed out of range by a tooling change. Cabling: drag chain bend radius tighter than the cable specification. Power quality: sags and harmonics from large loads on the same supply. Repeat damage is a system condition, not a parts quality problem.

In short: the price gap between repair and replacement is usually smaller than one day of downtime. Compare lead times first.

6. Sourcing the Right Replacement, Including Obsolete Models

When the model is discontinued. Three paths exist. The manufacturer's migration model, which usually needs new cables, a new parameter file, and sometimes a new mounting plate. A compatible unit from the same family, with the specs above verified line by line. Or the original model from remaining stock, which keeps machine and program untouched. Cross generation units rarely accept the old parameter file, so build commissioning time into the schedule.

Verifying you are not buying refurbished or counterfeit stock. Check nameplate print quality and label consistency, whether serial and batch codes can be verified, whether packaging and accessories are complete, and whether the warranty is written. A supplier who cannot say where a unit came from is telling you something. Our supplier credentials are on About Us.

Lead time is part of the spec. A cheaper unit arriving in six weeks costs more than a higher priced unit in stock, once you multiply the gap by hourly downtime cost. Ask for stock status and realistic clearance time before comparing quotes.

Not sure which replacement matches your unit?

Send a photo of the nameplate. We will confirm the model, compatible alternatives, and lead time within 24 hours.  | WhatsApp +86 189 2643 1543 | jerry@szct-automation.com Or browse the Model Library: Siemens, Allen-Bradley, Mitsubishi, Omron, Schneider, ABB.

Request a Quote

In short: checking the nameplate twice costs ten minutes. Ordering the wrong model costs a week.

7. After Replacement: Commissioning Checklist

1. Back up parameters before removing anything. Export from the commissioning software, memory card, or panel. With no backup, copy from an identical axis, or load defaults and rebuild each value against the documentation.
2. Re-zero the encoder after a motor change. Rotor magnet position relative to the feedback zero has changed. Skipping this produces start vibration, noise, following errors, or uncontrolled motion. After a drive change it is usually unnecessary, but confirm the motor and encoder selection in the parameters.
3. Auto-tune carefully. Tune uncoupled first, then add load in steps. Back the gain off one step when whine or oscillation appears. If the inertia ratio is out of range, tuning will not fix it.
4. Return to production in stages. Jog with no load, then single axis at low speed, then light load, then full load. Watch current stability, temperature rise, position deviation, and noise at each stage.
5. Log it. Date, machine, fault code, root cause, part and serial number, hours lost.

In short: a replacement is only half finished until parameters and the zero position are restored.

8. Preventing the Next Failure

Rank spares by lead time, not price. Any long lead item on a critical axis belongs on the shelf, and cutting the number of different models across your machines reduces inventory more than negotiation will.

FAQ

 

 

Is the problem the servo motor or the servo drive?

Uncouple the load and run the axis free. If it still faults, move the motor to an identical spare axis. If the fault follows the motor, the motor is bad. If it stays with the original drive, the drive is. Check insulation before any drive swap.

Can I test a servo motor without a drive?

Partly. You can measure phase to phase winding resistance and balance, insulation to ground, and brake release, and turn the shaft by hand for roughness. You cannot verify closed loop performance, encoder accuracy, or torque output.

What does a following error fault actually mean?

Actual position fell outside the allowed window around commanded position. Ranked causes: load beyond capability, ramp too steep, gains too low, brake not fully released, encoder problems, mechanical drag.

Why does my servo motor overheat with no load?

Three usual causes: a brake dragging because it never fully releases, wrong commutation or encoder configuration causing continuous current, or blocked cooling with high ambient temperature. Measure phase current at standstill to separate the first two from the third.

Is it cheaper to repair or replace a servo drive?

It depends on four numbers: hourly downtime cost, repair turnaround, new unit price and lead time, and remaining machine life. Repair wins for peripheral damage on machines near retirement. Replacement wins when repair turnaround exceeds the downtime you can absorb.

My servo drive model is discontinued. What are my options?

Three: the manufacturer's migration model, a compatible unit from the same family, or the original model from remaining stock. The first two need new cables, parameter rework, and commissioning time. The third keeps the machine unchanged. Send us the nameplate and we will tell you which paths exist.

Where You Are Now

Still diagnosing? Run the uncoupled test in section 2 first. Know the part? Check the nameplate against the eight specs in section 6. Model discontinued? You have three paths, each with different commissioning work.

Need Help Sourcing a Replacement?

Shenzhen Chentuo Technology has supplied automation parts since 2016 to distributors and end users in more than 30 countries, covering Siemens, Allen-Bradley, Mitsubishi, Omron, Schneider, and ABB. We hold stock on common HMI, PLC, PLC module, and drive items, and confirm model, availability, and lead time within 24 hours of a nameplate photo.

Request a Quote | WhatsApp +86 189 2643 1543 | jerry@szct-automation.com

Free consultation