Module 9 of 16
Inverters & Variable Frequency Drives
DC bus, IGBTs, and diagnosing drives without destroying a meter
Learning objectives
- Explain rectifier, DC bus, and inverter stages and what each contributes
- Measure DC bus voltage safely and interpret it
- Separate a drive fault from a motor or load fault using drive fault codes and current balance
- Recognise the supply and installation problems that cause repeat drive failures
- Explain how an inverter controls motor speed and torque, and why that changes what a meter can tell you
- Interpret DC bus behaviour as evidence about the rectifier, the bus capacitors, and the supply
- Diagnose communication faults between a main control and an inverter or motor-control board
- Apply the right method for residential inverter compressors versus commercial VFD installations
Three stages, three failure families
An inverter rectifies incoming AC into DC, smooths it across a bank of bus capacitors, then switches that DC with IGBTs to synthesise a variable-frequency, variable-voltage output. Motor speed follows frequency; torque follows the voltage-to-frequency relationship the drive maintains.
Each stage fails differently. Rectifier faults show as low or unstable DC bus and often as a blown input fuse. Bus capacitor ageing shows as ripple, nuisance undervoltage faults, and bulged cans. Output stage faults show as one weak or missing phase, unbalanced motor current, and overcurrent trips on acceleration.
- Expected DC bus on a 230 V single-phase input: roughly 320 VDC
- Expected DC bus on 400–480 V three-phase input: roughly 560–680 VDC
- Low bus with correct input = rectifier or bus capacitor problem
Bus capacitors hold lethal charge for minutes after disconnect. Wait the manufacturer's stated time and measure the bus before touching anything.
Measuring a drive output correctly
Do not judge inverter output with an ordinary averaging meter. PWM output is a train of high-frequency pulses; averaging meters report meaningless values and some meters are damaged by the switching transients. Use a true-RMS meter rated for the application, or better, judge output by current balance with a clamp.
Current balance is the practical field test. Clamp each output leg with the motor running at a steady speed: the three legs should be within a few percent of each other. A leg that is significantly low or zero points to the output stage or a motor winding, and comparing against motor winding resistance tells you which.
- Output legs should balance within about 5 percent
- Unbalanced with balanced motor windings = drive output stage
- Balanced current but no torque = load or coupling, not the drive
Reading drive fault codes as evidence
Drives log why they stopped, and those codes are more precise than most appliance codes. Overcurrent on acceleration usually means the ramp is too fast for the load or the load is dragging. Overvoltage on deceleration means regenerated energy has nowhere to go — a braking resistor or a longer ramp is the fix. Undervoltage means supply, not drive. Overtemperature means cooling: blocked heat sink fins, a failed fan, or an enclosure that has been closed up.
Clear the code only after recording it. The code plus the moment in the cycle when it appeared is often the entire diagnosis.
Installation faults that kill drives
Repeat drive failures usually trace to the installation rather than the drive. Long unshielded motor cables produce reflected-wave voltage spikes that erode winding insulation. Missing or shared grounds create common-mode currents that damage bearings and confuse control signals. Poor ventilation cooks the output stage.
Check supply quality too: voltage imbalance between phases above about 2 percent produces disproportionate current imbalance and heating, and is a common cause of drives that fail seasonally when the building load changes.
How an inverter actually controls a motor
An inverter does not vary voltage the way an old speed control did. It synthesises a waveform. The output stage switches the DC bus on and off thousands of times per second, and by varying the width of those pulses it produces an average that traces a sine wave at whatever frequency the control asks for. Motor speed follows that frequency almost exactly; slip accounts for the small difference on induction motors, and there is effectively none on the permanent-magnet motors used in modern compressors and direct-drive washers.
Torque comes from the relationship between voltage and frequency. The drive raises output voltage in proportion to frequency so magnetic flux in the motor stays constant; that is the classic volts-per-hertz behaviour. Below a few hertz the drive adds a voltage boost, because winding resistance eats a disproportionate share of a small voltage. Above base frequency the drive runs out of voltage and torque falls off — which is why an overspeed request on a loaded machine produces a stall, not more work.
Permanent-magnet motors add a further requirement: the drive must know where the rotor is before it can energise the right winding. It learns this from hall sensors, from a resolver, or by sensing back-EMF in a sensorless design. Lose that feedback and the drive will not commutate, which the machine reports as a start failure or a motor fault even though the motor is fine.
- Speed follows frequency; torque follows the volts-per-hertz relationship
- Low-speed torque depends on the drive's boost setting, not on the motor alone
- Permanent-magnet drives cannot start without valid rotor position feedback
A drive that refuses to start a healthy motor is usually protecting the machine because feedback or a permissive is missing. Prove feedback before condemning the drive.
The DC bus as evidence
The DC bus is the most informative single measurement on a drive, and each way it misbehaves points somewhere specific. A bus at the expected value with the drive faulted tells you the front end is healthy and the problem is in the output stage, the control, or the load. A low bus with correct incoming voltage points to the rectifier or to bus capacitors that no longer hold. A bus that rises during deceleration is regeneration — the motor is being driven by its load and pushing energy back — and needs a longer ramp or a braking path, not a new drive.
Ripple matters as much as level. Ageing capacitors let the bus sag between line cycles; the drive sees momentary undervoltage and trips at unpredictable moments, usually under load, usually described by the customer as random. On a three-phase input, a lost phase produces heavy ripple and a bus that reads plausibly on average while the drive trips constantly.
None of this is worth guessing at. The value, the stability, and the behaviour of the bus during acceleration and deceleration are three separate observations, and together they usually name the failing stage before you open anything.
- Correct bus, drive faulted: look at output stage, control, or load
- Low bus, correct input: rectifier or bus capacitors
- Bus rises on deceleration: regeneration, needs ramp or braking path
- Heavy ripple on three-phase input: suspect a lost phase before suspecting the drive
Bus capacitors store lethal energy. Observe the manufacturer's discharge time and verify the bus is safe with a rated meter before contact — this is not a step to compress.
Communication between control and drive
On residential inverter platforms the main control and the inverter are separate boards that exchange commands and status over a dedicated link. Most 'inverter' fault codes are really communication codes: the main control asked for a compressor start and heard nothing back within its timeout. The machine then reports a compressor or inverter fault, and technicians replace one or both boards while the actual fault sits in the harness between them.
The order that works: confirm supply and ground at the inverter board, confirm the ground reference is shared and solid, confirm link continuity with the harness flexed, and read whatever the inverter board itself indicates — most have an LED pattern or a service display that reports its own state independently of the main control. An inverter board that is powered, grounded, and reporting a healthy state while the main control claims it is absent is a link fault, not a board fault.
Commercial VFDs communicate too, over Modbus, BACnet or a proprietary building bus. The same logic applies with higher stakes: verify address, baud rate and termination before assuming a device has failed. A drive that runs correctly from its keypad and ignores the building control is a network problem.
- Most inverter fault codes are communication timeouts, not power-stage failures
- Read the inverter board's own indication before trusting the main control's account
- Runs from the keypad but not from the building control = network, address, or termination
Residential inverter compressors versus commercial VFDs
The physics is shared; the service reality is not. A residential inverter compressor is a sealed permanent-magnet motor driven by a matched inverter board, with no keypad, no parameter set, and no user-accessible fault log beyond the appliance's own codes. Your diagnostic instruments are winding resistance and insulation to ground at the compressor terminals, the inverter's own indication, the DC bus, and the machine's service mode. Winding balance and a clean insulation reading with a healthy bus and an inverter that will not drive is a strong case for the inverter; low insulation to ground condemns the compressor and predicts that a new inverter would fail too.
A commercial VFD is a configurable industrial device. It has a keypad, a parameter set, a fault history with timestamps and often with bus voltage and current captured at the moment of the trip, and it can be run locally to separate a drive problem from a control-signal problem. Its faults are frequently installation faults: cable length producing reflected-wave stress, missing or shared grounding, inadequate ventilation, or supply imbalance between phases.
The judgement difference matters. On residential equipment you are deciding between two sealed assemblies with limited evidence, so you gather every reading available before committing. On commercial equipment you have abundant evidence and the discipline is to read it — fault history first, parameters second, hardware third.
- Residential: winding balance, insulation to ground, bus, inverter indication, service mode
- Commercial: fault history with captured values, local keypad run, parameters, then hardware
- Repeat commercial drive failures are usually installation, supply, or cooling — not drive quality
Failure modes and what confirms them
| Symptom | Mechanism | The tell |
|---|---|---|
| Drive trips on overcurrent during acceleration | Ramp too aggressive or mechanical drag | Current spikes at start; load turns roughly by hand |
| Drive trips on overvoltage during deceleration | Regenerated energy with no braking path | Bus voltage rises sharply as speed falls |
| Undervoltage faults at random | Supply sag or ageing bus capacitors | Bus voltage low or with visible ripple; input voltage dips under building load |
| Motor runs rough, one leg low | Output stage or motor winding fault | Unbalanced output current with balanced winding resistance = drive |
| Repeat drive failures at the same site | Cable length, grounding, or ventilation problem | Unshielded long motor run, missing ground bond, or blocked heat sink |
| Machine reports an inverter or compressor fault, compressor never attempts to start | Communication timeout between main control and inverter board | Inverter board powered and indicating a healthy state while the main control reports it absent |
| Compressor draws heavily and the inverter shuts down immediately | Locked rotor, or winding insulation breakdown to ground | Low insulation resistance at the compressor terminals, or unbalanced winding resistance |
| Drive runs from its keypad but ignores the building control | Fieldbus address, baud rate, termination, or wiring fault | Local operation is normal while the network shows no response |
| Drive nuisance-trips under load with no pattern | Bus capacitor ageing or supply sag under building load | Bus ripple visible, or supply voltage dips at the drive terminals when loaded |
| Repeat drive or motor failures at the same installation | Long unshielded motor cable, missing ground bond, poor ventilation, or phase imbalance above about 2 percent | Installation survey finds the condition; the replaced parts show the same damage pattern each time |
Test procedures
Manufacturer differences
Samsung (digital inverter compressors)
Sealed permanent-magnet compressor driven by a matched inverter board, with the main control commanding capacity over a communication link.
What it changes: Check compressor winding balance and insulation to ground before touching the inverter, then read the inverter board's own indication rather than accepting the main control's fault code at face value.
LG (linear and inverter compressors, direct-drive motors)
Inverter-driven sealed compressors and stator-and-rotor direct-drive laundry motors with hall-effect rotor position feedback.
What it changes: Rotor position feedback is part of the drive circuit. Motor faults on direct-drive platforms should always include a hall output check before any board is condemned.
Whirlpool
Inverter-driven laundry and refrigeration platforms with the motor control integrated into or adjacent to the main control, and diagnostic modes that step outputs.
What it changes: Use the platform's service test to command motor operation directly; it separates a control command problem from a drive or motor problem in one step.
GE
Variable-capacity compressor platforms and inverter laundry drives with documented service tests and fault histories.
What it changes: Pull the fault history in order. On variable-capacity refrigeration the sequence of logged faults usually distinguishes a sealed-system limitation from a drive fault.
Bosch
Integrated control and drive modules with limited component-level access and condition-based error codes.
What it changes: Verify the condition the code describes physically. Modules are typically non-returnable, so the standard of evidence before ordering has to be higher.
Commercial VFDs (ABB, Danfoss, Yaskawa and similar)
Standalone programmable drives with keypads, full parameter sets, timestamped fault logs, and fieldbus connectivity.
What it changes: Read the fault log and captured values first, run locally from the keypad to isolate the control signal, and check parameters before hardware. Repeat failures point to cable length, grounding, ventilation, or supply imbalance.
DC bus behaviour and what it means
Value, stability, and behaviour during ramps are three separate observations. Record all three.
| Observation | Likely stage | Next test |
|---|---|---|
| Bus at expected value, drive faulted | Output stage, control, feedback, or load | Check output current balance and rotor position feedback |
| Bus low with correct input voltage | Rectifier or bus capacitors | Inspect for bulged capacitors; check input fusing and rectifier |
| Bus unstable with heavy ripple | Ageing capacitors, or a lost phase on three-phase input | Measure each input phase and phase-to-phase balance |
| Bus rises sharply during deceleration | Regeneration with no braking path | Extend the deceleration ramp or verify the braking resistor circuit |
| Bus collapses under load only | Supply capacity or a failing connection upstream | Measure supply voltage at the drive terminals while loaded |
Common drive fault families and where they actually live
| Fault family | Usual cause | Not the cause | Confirming evidence |
|---|---|---|---|
| Overcurrent on acceleration | Ramp too fast, or mechanical drag in the load | A failed drive | Load turns roughly by hand; current spikes only at start |
| Overvoltage on deceleration | Regenerated energy with nowhere to go | Supply overvoltage | Bus voltage rises as speed falls |
| Undervoltage | Supply sag or degraded bus capacitors | Motor fault | Input dips under building load, or bus shows heavy ripple |
| Overtemperature | Blocked fins, failed cooling fan, or a closed-up enclosure | Drive age | Heat sink hot with fan stopped or airflow blocked |
| Communication / no response from drive | Harness, ground reference, address, baud rate, or termination | Either board on its own | Drive powered and reporting healthy while the controller says it is absent |
| Ground fault | Motor winding insulation breakdown or a damaged motor cable | Drive output stage, usually | Insulation resistance to ground measured low at the motor |
Residential inverter compressor versus commercial VFD
| Aspect | Residential inverter | Commercial VFD |
|---|---|---|
| Fault information | Appliance codes and inverter board indication | Timestamped fault log with captured bus voltage and current |
| Parameters | Fixed, matched to the compressor | Fully configurable and frequently the actual fault |
| Local operation | Not available | Run from the keypad to isolate the control signal |
| Motor access | Sealed compressor terminals only | Full access to motor, cable and terminations |
| Primary evidence | Winding balance and insulation to ground | Fault history, then current balance, then installation |
| Repeat failure causes | Sealed-system load, low insulation, supply quality | Cable length, grounding, ventilation, phase imbalance |
Safety and professional boundaries
- Bus capacitors remain lethal after disconnect. Observe the stated discharge time and verify with a rated meter — no exceptions, no shortcuts under schedule pressure.
- Three-phase drive panels require arc-flash rated protection, lockout/tagout, and the qualification your jurisdiction demands. Decline work outside that qualification.
- Opening a sealed system requires refrigerant certification and recovery equipment. If you cannot recover legally, the job stops and gets handed to someone who can.
- Do not raise protective trip thresholds to keep a machine running. Protections exist because the alternative is hardware damage or a fire.
Measure the DC bus safely
- 1.Open and lock the supply disconnect; wait the manufacturer's discharge time.
- 2.Set a CAT III/IV rated meter to DC volts and measure across the documented bus terminals.
- 3.Confirm the bus is below the manufacturer's safe threshold before further work.
- 4.To evaluate under power, use only the drive's own display or documented test points and follow the manufacturer's live-work instructions.
Bus voltage is lethal and persists after disconnect. Never assume it has bled down.
Check output current balance
- 1.Run the motor at a steady commanded speed under normal load.
- 2.Clamp each output leg in turn and record the current.
- 3.Calculate the deviation from the average; over about 5 percent is significant.
- 4.Isolate and measure motor winding resistance; balanced windings with unbalanced current implicates the drive.
Triage a residential inverter compressor
- 1.Record the appliance fault codes and the order they were logged before clearing anything.
- 2.Confirm supply and ground at the inverter board and read the board's own indication.
- 3.Verify link continuity and ground reference between main control and inverter with the harness flexed.
- 4.With power removed and the bus verified safe, measure compressor winding resistance across all three combinations and confirm balance.
- 5.Measure insulation resistance from each winding to the compressor shell using an appropriate insulation tester.
- 6.Condemn the compressor only for low insulation or unbalanced windings; condemn the inverter only when the compressor is proven good and the link is proven sound.
Bus capacitors hold lethal charge. Observe the discharge time and verify the bus with a rated meter before touching compressor or inverter terminals.
Survey a commercial VFD installation after a repeat failure
- 1.Download or photograph the fault log with timestamps and captured values.
- 2.Measure each incoming phase and calculate voltage imbalance between phases.
- 3.Measure motor cable length and confirm whether shielded cable and correct glanding were used.
- 4.Verify the ground bond from drive to motor frame and to the panel, and confirm it is not shared with signal returns.
- 5.Check heat sink cleanliness, cooling fan operation, and enclosure ventilation against the drive's requirements.
- 6.Record the parameter set and compare it against the documented configuration for the application.
Three-phase panels demand appropriate arc-flash protection and lockout/tagout. Work within your qualification and your employer's electrical safety programme.
Expected readings and what they mean
| Measurement | Expected | Meaning |
|---|---|---|
| DC bus, 230 V single-phase input | ≈320 VDC | Low means rectifier or capacitor problem; unstable means ripple |
| DC bus, 400–480 V three-phase input | ≈560–680 VDC | Same interpretation at higher voltage |
| Output leg balance | Within 5% | Imbalance points at output stage or motor winding |
| Supply phase imbalance | Under 2% | Higher causes heating and premature drive and motor failure |
| Heat sink temperature | Below the drive's derating threshold | Overtemperature faults are cooling faults |
| Compressor winding resistance, all three combinations | Balanced within a fraction of an ohm of each other | Imbalance condemns the compressor and predicts that a new inverter would fail with it |
| Compressor winding insulation to shell | High, per the manufacturer's insulation test guidance | Low insulation is a compressor failure; fitting an inverter alone repeats the failure |
| Output current balance across three legs at steady speed | Within about 5 percent leg to leg | Imbalance with balanced windings points to terminations or the drive output stage |
| Incoming phase-to-phase voltage imbalance | Below about 2 percent | Higher imbalance produces disproportionate current imbalance and seasonal drive failures |
| Bus voltage captured at the moment of a trip | Within the drive's documented operating band | High at a stop event is regeneration; low under load is supply or capacitors |
Field scenarios
Mini-scenario: refrigerator reporting an inverter fault
- Inverter-compressor refrigerator, not cooling, compressor never starts
- Main control reports an inverter or compressor communication fault
- Condenser fan runs; condenser is clean
- Inverter board has supply voltage present at its connector
The code names the inverter and the inverter has power. What is the correct next step?
Mini-scenario: commercial VFD tripping on stop
- Belt-driven exhaust fan on a VFD trips only when commanded to stop
- Fault log shows overvoltage, timestamped at each stop event
- Captured bus voltage at trip is well above the running value
- Motor current is balanced while running and the drive is cool
Overvoltage on every stop, healthy while running. What is the fix path?
Mini-scenario: rough-running motor on a VFD
- Three-phase motor on a VFD runs rough and vibrates at all speeds
- Clamp readings show one output leg significantly lower than the other two
- Motor winding resistances measure balanced within a fraction of an ohm
- Insulation resistance to ground is high on all three windings
Unbalanced output current with balanced windings and good insulation. What does that prove?
The drive that fails every August
- A walk-in condensing unit's inverter drive has failed twice, both times in late summer. The replacement works fine for months.
- The drive is mounted in a closed enclosure against a south-facing wall, and the heat sink fins are packed with dust.
What is the real fault?
Thermal. The drive derates and then fails when ambient plus dust-blocked fins push the output stage beyond its rating, which happens in the hottest weeks. Clean the fins, restore or add enclosure ventilation, verify the cooling fan runs, and confirm ambient at the drive against its rating. Replacing the drive again without fixing cooling buys one more summer.
Takeaway: Seasonal electronics failures are almost always thermal or supply-related, not random.
Knowledge check
Why should you not judge inverter output with an averaging multimeter?
Output currents are unbalanced but motor winding resistances are balanced. Where is the fault?
A drive trips on overvoltage every time it decelerates a large fan. What is the fix?
A sealed inverter compressor shows balanced windings and high insulation to ground, the bus is at the expected value, and the inverter will not drive it. What is the defensible conclusion?
A commercial drive has failed three times in two years at one site, always the output stage. What deserves your attention?
You want to check inverter output voltage on a residential drive with your standard averaging meter. What should you do?
Key takeaways
- Rectifier, bus, and output stage each fail with their own signature.
- Bus capacitors stay lethal after disconnect — measure before touching.
- Current balance is the practical field test for drive output health.
- Repeat drive failures are usually cooling, cabling, or supply quality.
- Speed follows frequency, torque follows volts-per-hertz, and permanent-magnet drives cannot start without valid rotor feedback.
- The DC bus value, its stability, and its behaviour during ramps are three separate pieces of evidence — record all three.
- Most residential inverter fault codes are communication timeouts; prove the link and the ground before buying either board.
- Commercial drives hand you a timestamped fault log. Read it, run locally from the keypad, check parameters, then look at hardware.
- Repeat drive failures are installation problems until the installation has been surveyed and cleared.
Finished this module?
Mark it complete to track your progress toward the Technical Systems Certificate.