A used drive that powers up with a lit display, an empty fault log and a clean heatsink has proven exactly three things: the control board boots, the precharge circuit closes, and the switch-mode supply regulates. It has proven nothing about the DC bus capacitors, nothing about the output stage, and nothing about the fans. Treat any listing that says "tested" without a motor on the output as a power-up check and price it at core money. The only claim worth a premium is a drive that ran a real load long enough to reach thermal equilibrium, sold by someone who will take it back.
Decode What the Seller Actually Means
The word carries four completely different claims. Sort the listing into one of them before you look at the price.
- "The display came on." Control power and precharge work. Says nothing about the power section.
- "It ran when we pulled it off the machine." Best case it was healthy on removal day. It also tells you the machine was probably being scrapped or retrofitted, and that the drive has been sitting since - which is its own problem.
- "It worked five years ago." Worthless. Electrolytics degrade on the shelf faster than they degrade in service. This drive is a reforming job at minimum and a vent-and-spray risk at worst.
- "We ran a motor on it." The only version worth paying for - and only if you can see how long, at what load, and with what current on all three output legs.
None of these are health indicators on their own. Paper does not help either: fault-log printouts, service tags and test reports are trivially fabricated. What you are really buying is the seller's return policy. A vendor with a track record and a warranty they will honor is worth more than any document in the listing.
Match the Symptom to the Failure You Cannot See at No Load
| Symptom on the bench | Most likely cause | Shows up at power-up only? |
|---|---|---|
| Motor hums, will not accelerate, one output leg reads near zero current | Dead output phase - failed IGBT or gate driver in one leg | No - needs a motor connected |
| Overcurrent trip on acceleration only, clean at steady state | Output stage weakness, current-sensor drift, or a shorted motor lead | No |
| Loud bang or vented capacitor seconds after first power-up | Long shelf storage, oxide layer degraded, no reforming performed | Yes - and it is the dangerous one |
| Runs perfectly on the bench, faults intermittently once installed | Cracked solder joints or marginal gate drive that only opens up with thermal cycling and vibration | No |
Every row except the vented cap requires a motor, a load and time. That is the whole argument against no-load "testing" in one table.
Understand Why the Bus and the Output Stage Hide
At no load the DC bus is charged almost to the peak of the line - roughly 1.35 to 1.41 times the RMS input voltage on a six-pulse diode front end - and it stays there because nothing is drawing from it. Ripple current through the bus capacitors is nearly zero. A capacitor bank that has lost most of its capacitance and gained ESR will still hold that voltage and still let the drive boot and idle. Put load on it and the inverter starts pulling pulsed current out of the bus at twice line frequency plus switching ripple. The degraded bank cannot supply it, bus voltage sags on load steps, the ripple amplitude climbs, and internal heating from I²R in the ESR accelerates the whole thing. That delay is what makes weak caps the silent killer of used-drive purchases.
The output stage hides for the same structural reason. With no motor attached there is no current path through the IGBTs. One shorted, open or ungated leg is electrically invisible. The drive reports no fault because there is nothing to detect. Hook up a motor and the single-phased output shows immediately as a hum, a stall, or an unbalanced current reading.
The nameplate tells you the rating, not the duty. A drive rated for 10 hp may have run a decade of 500,000-plus hours near its overload limit in a hot cabinet. Elevated ambient is the single biggest life multiplier on electrolytics - a drive that lived in a 50 °C panel has burned through its design life far faster than the same unit in a conditioned room. Plan around a nominal 10-year design life for the drive as a whole and assume a used unit has already spent most of it.
Inspect and Reform Before You Apply Full Line Voltage
Start here, before anything is energized.
- Read the nameplate and photograph it. Full catalog number, voltage class, output amps, and the manufacture date code. If the nameplate is unreadable in the listing, walk.
- Pull the covers. Look for bulged or domed capacitor tops, split or lifted rubber vent plugs, dried electrolyte crust, browned or lifted PCB laminate around the bus caps and snubbers, and green corrosion on the power terminals.
- Check the fans by hand. Any roughness, drag or wobble means order the fan now, not after the load test.
- Blow out the heatsink and the control-board conductive dust. Metallic or carbon dust between traces is a fault waiting for humidity.
- Meter the input diode bridge and the output IGBTs terminal to terminal with a diode-test function, DC bus to each of U, V, W and L1, L2, L3. Any leg that reads short or dead open condemns the power section before you spend an hour on it.
- Never megger the drive terminals. Test motor insulation with the drive leads disconnected.
If the drive has been on a shelf for a year or more, reform the bus capacitors. Bring the DC bus up gradually with a variac on the input or a current-limited DC supply across the bus, holding at steps and letting leakage current settle before you increase, over several hours. If you cannot do it properly, the fallback is to leave the drive powered up and idle overnight before you put any load on it. Standing behind a drive when a capacitor vents is not a lesson you want to learn - treat the first energization of a stored drive as a safety event, doors closed, nobody in front of it.
Run the Load Test That Actually Proves the Drive
Assume it dies 30 seconds after you turn it on. Test in that order - cheapest fault first.
- Energize with no motor. Verify the display, the keypad, and the DC bus reading against 1.35-1.41 x the applied line RMS. Confirm the precharge relay or SCR bypass actually closes; a chattering precharge or a bus voltage stuck low means a failed precharge resistor.
- Read the fault history and, if the drive keeps them, the powered-hours and running-hours counters. Logs get cleared, so treat an empty log as no information rather than good news.
- Connect a small motor, uncoupled. Run it up to base speed. Clamp all three output legs. Currents should track within a few percent of each other. One low leg is a dead output phase - stop.
- Couple a real load. A dyno, a brake, a pump loop, a loaded gearbox - anything that will hold the drive at 50% of rated current or better.
- Cycle it. Accelerate and decelerate under load repeatedly, then let it idle and reload. Intermittent gate-drive and solder-joint faults surface on thermal cycling, not on steady state.
Verify the Numbers, Not the Display
What you want at the end of the soak:
- Output current balanced leg to leg within a few percent across the speed range.
- DC bus voltage steady under load steps, with no visible sag on acceleration. Scope it AC-coupled through a correctly rated differential probe - if the ripple amplitude climbs sharply as load increases, the bank is finished regardless of whether the drive tripped.
- Heatsink temperature plateauing well below the trip threshold, with airflow you can feel at the exhaust.
- No fault entries added to the log during the run.
- Motor current and drive-reported load percentage agreeing with your clamp meter - a large discrepancy points at current-sensor drift.
If a seller is claiming a load test, the equivalent evidence from them is a continuous video of the drive spinning a motor with the keypad readout visible, not a still photo of a lit display. Below that standard, pay parts money.
Price the Risk, and Know When Used Is the Wrong Answer
A workable ceiling: no more than 50% of the new price for the same model. Go to 60% only when there is a written warranty covering at least half the duration of the new-product warranty, from a vendor whose returns you have actually collected on before. Everything else is core value.
Used only makes sense in two situations. First, the part is out of production and the alternative is re-engineering the machine around a modern drive - at 50 to 250 kW that re-engineering cost dwarfs the gamble, so buying used and expecting to lose the money is a rational trade. Second, you are in a jam and need to restore production today. Outside those cases, migrate. Repaired and refurbished units fail too; a reputable firm will replace one, but you still lose a day of production while the replacement ships.
Stop bench-diagnosing and go to the manufacturer once the drive fails your load test with a shorted or open leg, a vented capacitor, or a bus fault that survives capacitor replacement - the power stack, gate drive and control board are matched, and parts-level repair on an obsolete unit rarely comes out ahead of a factory-remanufactured exchange. Contact the manufacturer's technical support or an authorized service center for a remanufactured unit with warranty coverage and for the documented migration path to the current family. If the drive is still in production, buy new and spend the bench time on something that pays.
Frequently Asked Questions
How do I test a used VFD without a dynamometer?
Couple the drive to any motor driving real work - a pump against a throttled valve, a loaded gearbox, a fan, or a second motor driven into a resistor bank.
How do I know if the DC bus capacitors need reforming?
Reform any drive that has been out of service for a year or more, regardless of how it looks. Ramp the bus up gradually with a variac or a current-limited DC supply over several hours, letting leakage current settle at each step; if you cannot do that, leave the drive energized and idle overnight before applying any load.
How do I spot a dead output phase before I install the drive?
Run a small uncoupled motor and clamp all three output legs - U, V and W currents should match within a few percent. One leg reading near zero, or a motor that hums and will not accelerate with no fault code displayed, is a failed IGBT or gate driver in that leg. A power-up test with no motor connected cannot detect it.