Understanding Generator and Transformer Stall Testing Procedures
The terminology around electrical machine testing gets messy fast. People throw around terms like stall cup, locked rotor, and short circuit interchangeably when they actually mean different things. I spent years dealing with field failures where the wrong test method caused more damage than it prevented. This guide walks through the actual procedure, the common mistakes, and the edge cases that blow up on site. A stall cup test, more formally called a locked rotor test, is performed to determine the starting characteristics of an electric motor or generator winding. You mechanically prevent the shaft from turning, apply reduced voltage to the stator, and measure current, power, and temperature rise. The results tell you whether the winding can handle the thermal stress of an actual stall event. Transformers get a similar but distinct test where the secondary is shorted and voltage is raised on the primary until rated current flows. The difference matters. A motor stall test reveals rotor bar integrity and starting torque curves. A transformer short-circuit test reveals copper loss at load and the impedance voltage percentage. Mixing them up on a work order has caused me to pull equipment that was perfectly fine right back out of the bay.
Step-by-Step Procedure for a Generator Stall Cup Test
Before you touch anything, verify the generator is disconnected from all loads and the prime mover is secured. This sounds obvious but I once saw a technician start the test sequence while the coupling was still connected to a compressor train. The backdrive spun the shaft and ruined the instrumentation in about three seconds. Step one: Install the torque reaction arm and lock the rotor. Use a proper mechanical brake or the manufacturer's provided locking pin. Do not improvise with chain hoists or wedges. I learned that the hard way on a 500 kW unit in 2019. A 3/8-inch hex key I used as asheared off inside the rotor slot and we spent two days disassembling the machine to remove it. Step two: Connect your test instrumentation. You need a true-RMS ammeter on each phase, a wattmeter or power analyzer for input power, and thermocouples placed on the stator windings at the hottest spots. I use type K thermocouples with adhesive-backed pads, clamped with high-temperature wire mesh. Cheap snap-on sensors fall off within minutes and you lose half your data points.
Step three: Apply reduced voltage gradually. Start at about 10 percent of rated voltage and increase in 5 percent increments. At each step, record current, power, and winding temperature. The goal is to reach rated locked rotor current without exceeding the winding temperature limit specified in the nameplate or NEMA MG 1 standards. For most induction machines this limit is 125 degrees Celsius for Class F insulation during a short-duration test. Step four: Hold at rated locked rotor current for the time specified by the manufacturer. This is typically 5 to 15 seconds for small motors and up to 60 seconds for large generators. Monitor temperature continuously. If the rate of temperature rise exceeds the expected curve, shut down immediately. A rapid spike usually indicates a grounded turn or a rotor bar defect that will become catastrophic under full load. Step five: After the test, perform an insulation resistance measurement before touching the windings. Hot insulation reads artificially high. Let the machine cool to near ambient, then measure. A drop of more than 50 percent from the pre-test value suggests insulation damage occurred during the stall.
Get the Full Details
Transformer Short-Circuit and Impedance Testing
Transformers don't have a stall cup per se, but the equivalent procedure is the short-circuit or impedance test. You short the low-voltage terminals with heavy copper straps rated for the expected current, then apply variable voltage to the high-voltage side using a variac or test transformer. Raise the voltage until the HV current equals the rated full-load current. At that point the applied voltage is the impedance voltage, usually expressed as a percentage. For a typical distribution transformer this might be 4 to 6 percent. For a large power transformer it could be 8 to 12 percent. The wattmeter reading at this point gives you the full-load copper loss. This is critical data for transformer loading calculations and thermal modeling. A counter-intuitive thing about this test: the copper loss you measure at rated current during the short-circuit test will be higher than the nameplate value if the windings are warm. Copper resistance increases about 0.4 percent per degree Celsius above 20 degrees. If your test room is 35 degrees and the windings have absorbed some heat from previous tests, your measured loss could be 6 to 8 percent higher than the guaranteed value. I've seen this cause unnecessary rejection of perfectly serviceable transformers. Always correct the measured resistance to the standard reference temperature, usually 75 degrees Celsius for oil-immersed units.
Motor and Compressor Coupling Considerations
When testing motors that drive compressors, you need to decide whether to test the motor standalone or coupled to the compressor. Standalone is simpler and gives you baseline motor data. Coupled testing is more realistic but introduces variables from the compressor side. Friction, internal clearances, and refrigerant charge all affect the load profile. I recommend standalone testing first. Document the results. Then couple and run a performance test under normal operating conditions. Comparing the two lets you isolate motor issues from compressor issues. The biggest pitfall here is assuming a motor that passes standalone is fine when coupled. A motor might draw acceptable current at no load but exceed its thermal limit when the compressor begins cycling. Check the locked rotor amps against the compressor's breakdown torque requirement. If the compressor needs more torque to overcome refrigerant pressure at start-up than the motor can provide at rated voltage, you will have a problem.
Common Pitfalls and When This Approach Fails
Stall cup testing has real limitations. It does not detect inter-turn faults that only manifest under thermal cycling. A winding can pass a cold stall test and fail two weeks later when the hot expansion closes a turn-to-turn gap. If you suspect this, combine the stall test with a surge comparison test. Surge testing catches problems that thermal and insulation tests miss entirely. Another failure mode: stall testing is destructive by nature. Every test degrades insulation slightly. For older equipment with worn insulation, the test itself can be the final straw. I always check insulation resistance and polarization index before committing to a full-stall test. If the PI is below 1.5, I do a reduced-voltage test instead and monitor carefully. Better to catch degradation early than to ruin a machine on the test bench. Synchronous generators add complexity. The excitation system must be disabled or bypassed during the test, and any residual magnetism needs to be checked beforehand. A generator with zero residual magnetism won't pick up voltage even with field current applied, which can confuse the test setup. I keep a small permanent magnet and a compass on my test tray for this. Ten seconds with the magnet and the field is usually enough to restore usable residual flux.
For compressor motors specifically, the locked rotor torque requirement is often 150 to 200 percent of full-load torque. Standard design NEMA B motors provide about 175 percent. If you are starting against a high head pressure condition, you may need a NEMA D motor or a soft starter. Stall testing alone won't tell you this. You need to simulate the actual starting condition or calculate the breakaway torque from the compressor specifications. Skipping this step is how I ended up with a 480-volt, 200-horsepower unit sitting on a pad because it couldn't start against 400 psig discharge pressure on a cold morning.
Documentation and Records
Keep a test record for every unit. Include ambient temperature, winding temperature, applied voltage, current per phase, power per phase, calculated losses, and insulation resistance before and after. Future troubleshooting depends on having baseline data. A current unbalance of 3 percent might look fine in isolation but becomes significant if your baseline was 0.5 percent. I found this on a 750 kW generator last year. The phase current unbalance had slowly climbed from 0.4 to 3.2 percent over eight years of annual testing. Catching that trend saved us from an unexpected failure during peak season. Standard reference documents for these procedures include IEEE 112 for motor testing, IEEE C57.12.90 for transformer testing, and NEMA MG 1 for general requirements. None of these are especially readable, but they are the authority your inspector will cite when something goes wrong.
