Air Compressor Motor Humming But Not Starting: Capacitor and Winding Issues

That hum is the sound of a single-phase motor stalled. The main winding is energized, the magnetic field is established, and the rotor is locked in place unable to move. Below, the two core fault lines of capacitors and windings get taken apart.

Single-phase AC can only produce a magnetic field that swings back and forth inside the stator, completely different from a three-phase motor. A three-phase motor's field starts rotating the moment power is applied. A single-phase field pushes and pulls along a single axis, the force on the rotor is symmetrical in both directions, net torque is zero. The rotor can't move, vibrates at line frequency, and that vibration is the hum.

Every single-phase compressor motor has an additional start winding installed, offset from the main winding by roughly 90 electrical degrees, with a start capacitor in series. The capacitor advances the phase of the current in the start winding, pulling apart the phase angle from the lagging current in the main winding, synthesizing an approximately rotating field. The motor runs in a quasi-two-phase mode for the first one to two seconds, dragging the rotor up to speed. Once the rotor reaches somewhere past seventy percent of synchronous speed, a centrifugal switch or relay cuts out the start winding, and the motor continues running on the main winding alone.

The start capacitor is the most common cause in this type of fault, and also the component that gets diagnosed most crudely. Most people's approach is to set the multimeter to capacitance mode, get a reading, and declare it not broken. What this approach misses is substantial.

The number 270μF is not a rough pick. Each motor's start winding has a specific impedance value under locked-rotor conditions, and the capacitor value is calculated against that impedance at line frequency, with the goal of getting the phase angle between the two winding currents as close to 90 degrees as possible. Swap in a 180μF capacitor, the phase angle shrinks, starting torque shrinks. If the tank is empty maybe it still turns over. If there's back pressure in the cylinder, that gap of a few dozen μF is the line between starting and not starting.

When capacitance has decayed to about seventy percent of the rated value, a continuity check reads perfectly normal, charge and discharge behavior looks normal too. Only a capacitance meter will show the reading is off. This type of capacitor on a machine behaves as sometimes starting, sometimes not, depending on how much pressure is left in the cylinder at that moment. It's not an intermittent motor fault. The capacitor's margin is already gone, and variation in the load is exposing it.

This takes a bit to explain.

Electrolytic start capacitors have a printed tolerance of ±20%. A capacitor labeled 270μF is considered in spec anywhere from 216 to 324. Manufacturers use less dielectric material to keep costs down, so cheap ones ship right at the lower limit. Not after a year or two of degradation. Brand new, still sealed, put a capacitance meter on it and it reads two hundred twenty, two hundred forty. Within the tolerance band, no basis for a quality complaint.

What this means in the field: some compressors come from the factory assembled with the cheapest capacitors the supply chain offers, and the starting torque margin is thinner than the design engineer calculated from day one. A little bit of aging on top of that and it's no longer enough. If the replacement capacitor is again the cheapest one off the internet, it amounts to swapping one bottom-of-tolerance capacitor for another bottom-of-tolerance capacitor. The symptom might improve briefly or might not improve at all.

There's an even worse situation online. The label says a brand name, the dielectric stack inside doesn't match at all, marked 270 but measures under 170, marked 330V but punctures at 280. Put one of these in, loaded start fails on the first attempt. Unloaded it might barely turn over, which tricks you into thinking it's fine. A few weeks later the capacitor body breaks down, possibly with an audible pop.

Electrolytic capacitor capacitance follows temperature. The ionic conductivity of the electrolyte drops sharply in cold conditions. A capacitor that reads 270μF at 25°C might only deliver 170 to 190μF near 0°C. Down to minus ten, it may not even hold sixty percent.

Every autumn a batch of compressors develops this problem. Worked fine all summer, first cold morning it hums and won't start. Nothing broke, nothing changed. By noon when the garage warms up, the same machine fires right up. Pull the capacitor off and test it indoors, reading is perfectly normal, because at room temperature it is normal. There are plenty of people who make multiple trips chasing this one without finding the cause.

The fix is to install a start capacitor with a higher rated capacitance, building in the low-temperature decay ahead of time. Or add a hard start kit, where the PTCR thermistor relay inside keeps the start capacitor connected for an extra fraction of a second, extending the effective start window in cold conditions.

While on this topic, the scenarios where a hard start kit is useful are basically two: cold weather starting difficulty and incomplete unloader valve pressure relief. Outside these two scenarios, installing a kit is most likely masking a different fault. More on this later.

This matters when doing replacements.

At the moment of starting, the start winding is an inductive load. When current through it is interrupted, the inductive kickback adds to the voltage across the capacitor. On a 230V motor, the transient voltage peak across the start capacitor can reach 350V or higher. That's why the standard capacitor voltage rating for a 230V compressor motor is 330V, not 220V.

No 330V on hand, use a 220V to get by for now, will it work? It'll work. How long is a matter of luck. Every start subjects the 220V dielectric layer to voltage stress beyond its rating, micro-puncturing a little, then a little more next time. Capacitance silently drops, ESR silently climbs. A few months later the motor starts having occasional trouble starting. New capacitor fixes it, cause gets filed as "capacitor aging." It did age. It was accelerated by the wrong voltage rating.

So replace capacitors at the original rated capacitance, don't substitute across ratings. If more starting torque is needed, use a hard start kit to add capacitance in parallel. Don't change the original capacitor's value.

This section is shorter. The information density isn't as high as the start capacitor section. Two failure modes to note.

The run capacitor in a CSR (capacitor-start capacitor-run) configuration stays in the circuit during normal operation, improving power factor.

Short circuit failure: the run capacitor is in parallel with the series combination of start capacitor and start winding. When the run capacitor shorts internally, it becomes a low-impedance bypass. During starting, most of the current that should flow through the start winding gets diverted through it instead. Pull the start capacitor and test it, it's good. Measure the start winding, it's good. Motor still hums. Someone with experience will think to check the run capacitor. Someone without can get stuck at this stage for a long time.

Open circuit failure: the motor can still start and run, but efficiency drops, current rises, and the main winding runs hotter than it should. Thermal protection trips more and more frequently, cooling intervals between trips aren't long enough, winding temperature ratchets up with each cycle. A few weeks later one day it won't start, just hums. Looks like a start circuit problem. Tracing back reveals the run capacitor went open at some unknown point and the winding has accumulated thermal damage.

Winding faults take more effort to write about than capacitors, because by the time diagnosis reaches the winding level, simple swap-and-test logic is no longer sufficient.

C to R is the main winding, thick wire, low resistance. C to S is the start winding, thin wire, high resistance. S to R spans both windings in series, the reading should roughly equal the sum of the other two.

An absolute resistance reading by itself tells you nothing. 3.2 ohms is normal on one motor model and means a short on another. Comparison to factory data is needed, or better yet, baseline readings recorded when the motor was healthy (almost nobody does this, but it should be done).

The hardest winding fault to catch. Two adjacent turns of wire lose their enamel insulation and touch each other, forming a shorted loop. This loop circulates large current within itself, generates intense localized heat, while its effect on the overall winding resistance can be very small.

When enough turns are involved, resistance measurement picks it up, the total winding resistance drops noticeably. When only two or three turns are shorted, the resistance change drowns in the measurement uncertainty of an ordinary multimeter. Data comes back looking normal. Motor hums.

The proper diagnostic tool is a surge comparison tester, which injects a steep pulse into the winding and examines the oscillation waveform. Shorted turns change the winding inductance, shifting the oscillation frequency and decay envelope compared to a healthy winding. This equipment isn't found in most general repair shops.

A field alternative that can be done on the spot: infrared temperature scanning. After a brief controlled start attempt (power on for two or three seconds, then immediately cut), sweep an IR thermometer around the stator housing right away. If there's a turn-to-turn short, the circulating current in those two or three seconds generates heat concentrated at the short location that hasn't had time to spread. The IR gun can pick up that spot reading several degrees hotter than its surroundings. Speed matters. Wait ten or twenty seconds and conduction evens out the temperature, eliminating the detectable difference.

This method works reasonably well for locating where the short is. For determining whether a short exists at all, it's less sensitive, since only a couple of shorted turns may not generate enough heat to create a distinguishable temperature difference on the housing surface. That's where the limitation sits.

This is directly related to the increase in winding failure rates.

Pick up a 5HP compressor motor from twenty years ago and one made today, same rating. Weigh them in your hands and the difference is obvious. The new one is noticeably lighter. Wire gauge is thinner, turn count is lower, slot depth is shallower. The nameplate parameters are the same. Performance specs measured under lab standard conditions are the same. The difference is in the margin.

The old motor's winding temperature during normal operation might have sat twenty or thirty degrees below the insulation class limit. That temperature gap is the tolerance for error. Voltage a little low, ambient a little hot, run capacitor slightly degraded causing current to run a bit high, temperature rises, still within the safe zone.

On current production motors that gap has been compressed to ten or so degrees. The same external deviations push the winding straight into the accelerated insulation aging zone. After a few years the insulation starts developing problems. Capacitor and supply conditions need to be maintained to a higher standard than before to get the same winding lifespan.

Insulation life follows an exponential relationship with temperature. Every 10°C above the rated thermal class cuts life in half. Class B insulation rated at 130°C can run twenty thousand hours at 130°C, ten thousand at 140°C, five thousand at 150°C. This rule is well established.

What matters more to field repair is the effect of moisture, which textbooks usually cover in a single sentence, while in the field it causes far more trouble than that sentence would suggest.

A compressor sitting in an unheated garage goes through dew point cycling all autumn and winter. Nighttime temperatures drop below the dew point, moisture condenses on internal motor surfaces including winding insulation. Daytime temperatures rise and evaporate some of it, not all. Over a winter the insulation's moisture content gradually climbs and its dielectric strength gradually drops. This process is silent. Nothing looks abnormal from the outside.

Motor rewind shops can observe a very clear seasonal peak: every April through May they receive a concentrated wave of burned compressor motors. Come summer the volume drops, because once temperatures are up, the cold and moisture factors no longer stack.

Once insulation diagnosis comes up, it's megohmmeter territory. A regular multimeter's resistance setting has close to zero ability to assess insulation condition. Insulation resistance is in the megohm range. A regular meter reads "OL" (over limit), and it reads "OL" for both healthy insulation and insulation that's about to fail. No differentiation.

A megohmmeter applies 500V DC, forces a tiny leakage current through the insulation, and uses the reading to quantify insulation quality. Two measurements: winding to winding, and winding to motor frame.

Above 100 megohms, no issues. Between 5 and 100 megohms, insulation is aging, still usable, start paying attention. Below 2 megohms, actively failing. Below 1 megohm, winding is condemned, changing the capacitor is pointless.

When megohmmeter readings fall in the borderline zone, a polarization index test can be added. The method: keep applying 500V DC, record the insulation resistance at 1 minute and at 10 minutes, calculate the ratio. If the 10-minute reading is significantly higher than the 1-minute reading (ratio above 2), the insulation structure is intact. The initial low reading was surface moisture being driven off by the DC voltage; resistance is improving over time. If the ratio is close to 1 or below 1.5, the leakage paths are permanent. Not a moisture problem. Drying and baking won't help. The insulation has structural damage.

The polarization index test, one megohmmeter and ten minutes, gives a level of accuracy in assessing remaining winding life that's a step above resistance measurement alone. Unfortunately very few people use it in the context of compressor repair.

This section needs emphasis, because the secondary damage from this operational mistake can be more severe than the original fault.

Locked-rotor current is several times normal running current. Copper loss is proportional to the square of current. The heating intensity during a stall is tens of times what it is during normal running. The thermal protector will trip within a few seconds. In those few seconds the winding temperature can spike by several tens of degrees.

If the thermal protector is reset immediately after tripping and another start attempt is made, the winding begins heating again from the elevated temperature baseline left by the last stall. Three or four rapid successive starts can push the start winding (thin wire, thin insulation, low thermal mass) past its insulation temperature rating, even if the motor was completely healthy before the first stall.

What's commonly seen in the field goes like this: the capacitor has failed, the operator doesn't know the cause, tries five or six times. Each time it hums for a few seconds, thermal protection trips, waits ten or fifteen seconds for the reset, tries again. Eventually gives up and calls someone. The repair person arrives, tests the capacitor, finds low capacitance, replaces it, motor starts normally. Problem appears solved.

The issue is that the start winding insulation has been repeatedly overtemped during those five or six stalls. Can't be seen with the eye. Can't be detected by resistance measurement (unless it's already severe enough to have caused a short). A megohmmeter might show some decline in insulation resistance. The remaining lifespan of the winding has been drastically shortened. A few months later the motor hums again. This time the winding is burned. No longer connects to the earlier capacitor fault. If those five or six stall attempts had each been spaced fifteen minutes apart to let the winding cool fully, the insulation most likely would not have sustained such severe damage.

Capacitor and windings both test good, motor still hums. Before concluding it's an internal winding fault, check the switch in between.

Open-frame motors use a centrifugal switch, mounted on the shaft, which flings its contacts open once the rotor reaches speed to disconnect the start circuit. Hermetic and semi-hermetic compressor motors use a potential relay or current relay, mounted externally.

The centrifugal switch stuck in the open position is fairly common. Spring broke, contacts oxidized and stuck on the open side, the flyweight mechanism got gummed up with oil contamination. The start circuit is open before the motor is even energized. Start capacitor and start winding don't participate at all. Main winding powers up, motor hums. Pull the capacitor and test it, good. Measure the windings, good. The problem is a switch that costs a handful of dollars.

The centrifugal switch also has a gradual degradation process worth discussing. Each time the motor reaches speed and the switch opens, the magnetic field in the start winding collapses suddenly, inducing a voltage spike that arcs across the just-separated contacts. After a few thousand start-stop cycles the contact surfaces are chewed up with pitting, covered in a layer of carbon buildup, and contact resistance is high. Eventually the mechanical action is still normal, the contacts are touching, but electrical conduction is no longer reliable. The motor sometimes starts, sometimes hums, with no discernible pattern at all, entirely dependent on the microscopic contact conditions at the moment the surfaces meet. A light pass with a fine file or fine sandpaper across the contact surfaces restores conduction.

Going the other direction, a centrifugal switch stuck in the closed position or a relay with welded contacts keeps the start capacitor in the circuit during running. The start capacitor is designed for intermittent duty of a second or two per start. Continuous connection for several minutes causes it to overheat. The start winding likewise isn't designed for continuous energization. Both components fail in short order. The presenting symptom is a motor that was working fine last week and now suddenly hums, because the capacitor has been destroyed by the stuck switch. Replace the capacitor, it works for a few days, then the capacitor burns again. At that point someone finally thinks to check the switch. So if a replacement capacitor burns out again within a short period, check the switch.

The parts cost about ten dollars or so. Packaged in a plastic box with a label, sold for forty to eighty. High margin product.

Inside is an additional start capacitor and a PTCR thermistor relay. At room temperature the PTCR has very low resistance, essentially conducting, allowing the additional capacitor into the start circuit. Once energized, the PTCR heats itself, its resistance spikes, and within a few seconds it becomes near-open-circuit, disconnecting the additional capacitor. The effect is increased total capacitance during the starting phase, increased starting torque, and a slightly extended effective start duration.

Two applicable scenarios. First, all electrical components are healthy, but the unloader valve doesn't fully relieve pressure, the cylinder has back pressure, and the motor's factory starting torque can't overcome it. Second, cold-weather environment where capacitor capacitance loss reduces starting torque below the threshold. In these two scenarios, installing the kit makes sense.

After the motor stops, the unloader valve should release residual compressed air from the cylinder head and discharge pipe, so that the next start faces minimal resistance. When the unloader valve is stuck closed, the piston has to start from zero speed against residual pressure. A single-phase motor's starting torque can't handle that.

Diagnosis: open the tank drain valve, let pressure bleed to zero completely, start the motor. If it starts up normally, the unloader valve is the problem.

Induction motor starting torque is proportional to the square of voltage. This squared relationship means a small percentage drop in voltage costs a disproportionately large percentage of torque. Voltage dropping from 230V to 207V, a ten percent drop, costs close to twenty percent of torque.

Long extension cords, undersized branch circuits, sharing a circuit with other high-current loads, all capable of pulling voltage down to where it's not enough.

In cold winter conditions, voltage insufficiency and capacitor capacitance loss compound each other. Compressor oil thickens, requiring more torque. Capacitor capacitance drops with cold, providing less torque. If the supply line voltage is also sagging, every factor stacks in the same direction. A configuration that still had margin in July can be severely short of torque in January.

Start from the cheap and simple end.