How a Capacitor Behaves in a Motor Circuit
Why this matters
A capacitance reading inside the printed tolerance gets a capacitor exonerated and a tech sent home, and that is the single most common way a motor callback happens twice. The capacitor is not a pass/fail part sitting beside the motor. It is a series element in a second electrical path, and the number on your meter describes only that element, never what the path is doing. Judge the path and you find the corroded terminal, the drifted winding, or the genuinely dead can. Judge the number alone and you replace a good part.
Before the leads touch anything
A motor-run capacitor holds charge after the disconnect opens, and a can that reads zero on your meter can still be holding a surface charge that reappears. De-energize at the disconnect, lock and tag it per 29 CFR 1910.333(b)(2) (29 CFR 1926.417 is the construction counterpart for electrical lockout and tagging), then discharge each capacitor terminal pair through a resistor rather than shorting it with a screwdriver, which throws molten metal and can rupture the can. Prove dead the way NFPA 70E-2021, 120.5 requires: verify your meter on a known live source, test the terminals, verify the meter again.
Where this article tells you to take a reading with the motor running, you are working inside the troubleshooting exception at 29 CFR 1910.333(a)(1), which permits energized work only where de-energizing is infeasible because the test cannot be performed any other way, and only with equipment rated for the circuit under 29 CFR 1910.334(c)(2) - meter, leads, probes and clamp all rated at or above the system, with an arc-flash risk assessment done first under NFPA 70E-2021, 130.5. An under-rated meter across a line-voltage fault is the failure that injures people, and it does not announce itself before it happens.
What the capacitor is actually doing
A single-phase motor has no rotating field of its own. Two windings sit at different physical angles inside the stator, and if current in both peaks at the same instant, the field just pulses back and forth and the rotor sits there humming. The capacitor sits in series with the auxiliary winding and delays that branch's current relative to the main winding's. Two currents, offset in time, in coils offset in space, add up to a field that sweeps around the stator. That sweep is the torque.
So the capacitor's job is a phase shift, and its value sets the size of the shift. Capacitive reactance is the opposition it presents, and it falls as capacitance rises: reactance in ohms equals 1 divided by (2 x pi x frequency x capacitance in farads). At 60 Hz, a 43 microfarad capacitor presents about 61 ohms. Drop the capacitance and the reactance climbs, the auxiliary current falls, the shift degrades, and the field goes lopsided. The motor does not stop. It runs with less torque, draws more current in the main winding to make up the shortfall, and gets hot.
Start and run capacitors are not interchangeable, and the difference is duty, not just value. A start capacitor is intermittent-duty, sized for a few seconds per start and switched out of circuit by a relay or a switch once the rotor comes up to speed; a run capacitor stays energized continuously and is built for it. Read the duty and the voltage rating off the can rather than inferring either from the capacitance value, and never substitute across that line even where the microfarads match.
The gate that decides serviceable
State the rule with its unit of analysis before running any case against it. Per capacitor, at operating temperature, under real load:
- Bench capacitance falls inside the tolerance printed on the can, and
- in-circuit capacitance computed from the voltage across the can and the current in its lead agrees with the bench figure within your meter's own uncertainty, and
- line current at the motor is at or under nameplate full-load amps.
All three, joined by AND, not any one of them. Fail 1 or 2 and the capacitor is the part. Pass 1 and 2 and fail 3 and the capacitor is exonerated and the fault is elsewhere in the branch. The in-circuit figure comes from the same physics as the bench figure: capacitance in microfarads is about 2652 times the current in amps divided by the volts across the capacitor, at 60 Hz.
Two motors, same reading, opposite verdicts
Both are permanent-split-capacitor units on a nominal 240 V single-phase supply, nameplate full-load current 6.9 A, can printed 45 microfarads with a tolerance of plus or minus 6 percent. Read the tolerance off the can rather than assuming it - 6 percent is common on run capacitors but it is printed for a reason. That band runs 42.3 to 47.7 microfarads.
Motor A. Bench reading after discharge: 43.1 microfarads. Inside the band. Re-energized, with the clamp on the capacitor lead and the meter across the can: 246 V and 4.0 A. Computed: 2652 x 4.0 / 246 = 43.1 microfarads. The two figures agree, which tells you the can is doing in the circuit exactly what it did on the bench. Main-winding branch reads 5.2 A, auxiliary branch 4.0 A, line current 6.6 A against a 6.9 A nameplate. Under. All three gates pass. The capacitor is not the fault, and neither is the motor - go look at what the customer actually called about.
Note what happened to the arithmetic there. The two branch currents sum to 9.2 A, but the line reads 6.6 A. That is not a bad clamp. The branches are roughly a quarter cycle apart, so they add as a right triangle rather than end to end: the square root of (5.2 squared plus 4.0 squared) is 6.56 A, which is what the line shows. The arithmetic sum overstates the line current by about 2.6 A, roughly 40 percent above the 6.6 A actually flowing. A tech who adds branch currents and concludes the motor is 33 percent over its nameplate is reading a phase relationship as an overload.
Motor B. Same equipment type, adjacent unit, same nameplate. Bench reading: 44.1 microfarads. Inside the band, and closer to nominal than Motor A. Under the older habit that is a pass and the capacitor goes back in. Re-energized: 175 V across the can, 2.9 A in the lead. Computed: 2652 x 2.9 / 175 = 44.0 microfarads. Gate 2 passes too - the can is genuinely healthy, twice confirmed, in place.
Then gate 3. Main branch 7.8 A, auxiliary branch 2.9 A, line current 8.1 A against 6.9 A nameplate. That is 8.1 divided by 6.9, or about 17 percent over rated current, running continuously. Gate 3 fails while gates 1 and 2 pass, and that combination has exactly one reading: the capacitor is fine and the auxiliary branch is not carrying what it should. Motor A's auxiliary branch ran 4.0 A; this one runs 2.9 A with a good capacitor in it, which means something else in that series path is adding opposition. Low current with low voltage across the can is the signature of resistance in series with the branch, not of a weak capacitor - a weak capacitor would show up as low current with HIGH voltage across the can, because a smaller capacitance means more reactance to drop across.
On this unit the added resistance was a heat-darkened spade terminal on the capacitor's auxiliary lead. The fix was a new terminal and a cleaned tab. Line current came back to 6.5 A and the auxiliary branch to 3.9 A. Swapping the capacitor, which was the reflex, would have changed nothing measurable and put the callback on the schedule for the following week with the same fault and a new part in it.
What a capacitance reading cannot see
The bench number is a property of one component measured in isolation, and it is blind to four things that will end the same motor:
- Series resistance anywhere in the branch. Terminals, the internal fuse link on some cans, the winding's own leads. Motor B's whole fault lived here.
- Winding condition. Shorted turns in the auxiliary winding lower its impedance and raise its current; an open or high-resistance section lowers the current. Both leave the capacitor reading normal.
- Applied voltage. Reactance is fixed by capacitance and frequency, so branch current tracks the voltage across the branch. A supply sagging under load moves the whole picture, which is why the line-voltage reading belongs on the same page as the current readings.
- Temperature and time. A can that measures in band cold on a bench can be outside it at cabinet temperature after an hour of running. Where the complaint is heat-related and the bench figure is near the edge of the band, take the in-circuit computation hot rather than trusting the cold bench number.
What flips the verdict
Two conditions genuinely invert the method rather than adjusting it. First, a capacitor with a bulged case, vented top, or leaked dielectric is condemned on sight and no reading argues it back into service, because the case is the pressure boundary and it has already told you it failed. Second, on a capacitor-start motor with a switched start winding, the auxiliary branch is only energized for the first seconds of a start, so gates 2 and 3 as written do not apply to it - you are looking at the start relay or centrifugal switch and the start capacitor's condition, and a continuously energized start capacitor is itself the fault, not a symptom.
How to verify you got this right
Take the same three readings after the repair and write all three down, not just the one that changed. Line current should sit at or under nameplate, both branch currents should have moved in the direction your diagnosis predicted, and the in-circuit and bench capacitance figures should still agree. If line current dropped but the auxiliary branch did not move, you fixed something real but not the thing you named, and your write-up should say so rather than claiming the diagnosis.
Then let the unit run to operating temperature and re-clamp. A branch fault that involves a resistive connection grows as the joint heats, so a reading taken two minutes after a repair is the most flattering one you will ever get from it.
References
- 29 CFR 1910.333(a)(1) and (b)(2) - de-energizing and safe work practices for electrical work; 29 CFR 1926.417 for the construction counterpart on lockout and tagging of circuits
- 29 CFR 1910.334(c)(2) - test instruments and equipment rated for the circuits to which they are connected
- NFPA 70E-2021, 120.5 (establishing an electrically safe work condition) and 130.5 (arc flash risk assessment)
- Manufacturer documentation for capacitance tolerance, voltage rating and duty class printed on the component
- See related: How Capacitors Fail (Generic); Why a Loose Connection Gets Hot; How to Read Current as a Diagnostic