Test a Run Capacitor While the System Still Runs
A bench meter reads a capacitor cold, at low voltage, out of circuit — the three conditions where a dying part looks healthiest. The under-load method measures what the capacitor actually delivers at operating voltage and temperature: clamp the amps, read the volts, and this calculator turns them into microfarads and a keep-or-replace verdict.
HVAC Capacitor Under-Load Calculator
Clamp the amps, read the volts — µF and a keep-or-replace verdict, live.
µF = 2652 × A ÷ V
Constant at 60 Hz mains: 2652. The same formula uses 2652 at 60 Hz and 3183 at 50 Hz.
Awaiting input
Enter your clamp-meter amps, capacitor volts and the rated µF from the label.
Live-voltage work. Measuring under load means the capacitor terminals sit at 300–450 V while the system runs, and a stored charge lingers after power-off. Discharge through a 20 kΩ 5 W resistor before touching anything, keep one hand out of the panel, and read the full safety protocol below before attempting this test. Educational aid only — defer to a licensed technician.
Why a Bench Test Can Lie to You
Pull a run capacitor out of a condenser, carry it to the bench, clip on a capacitance meter, and you can get a reading of 35.1 µF on a part that is three service calls away from killing a compressor. The meter is not wrong. The test conditions are. A bench instrument evaluates the part cold, at a volt or two of test signal, and completely out of circuit — the three conditions a working capacitor never experiences in real life.
In the condenser, the same part bakes on top of the cabinet where summer ambient reaches 55–60 °C, passes continuous ripple current from the motor circuit, and sits under a DC bias with an AC component riding on top. Film dielectric behaves differently in that environment: effective plate area shrinks a little every time the metallized film vaporizes around a fault during one of its self-healing events, and each event is triggered by heat and electrical stress — not by a quiet shelf.
The under-load method flips the question. Instead of asking what the part would do in isolation, it measures what the part is delivering right now, while the motor runs at operating voltage and temperature. That is a performance measurement, not a specification check — and it is the number the compressor actually cares about.
Field tools for the under-load methodPhoto: UnsplashFrom Reactance to the 2652 Constant
A run capacitor sits in series with the motor's auxiliary (start) winding. Its job is twofold: drop a big chunk of the supply voltage, and shift the current through that winding so it arrives out of phase with the main winding's current. The phase offset between the two windings is what creates the rotating field that gives a compressor the torque to start and the efficiency to keep running. The amount of shift depends on capacitive reactance, and reactance depends on frequency and capacitance:
Xc = 1 / (2πfC) → I = 2πf · C · V → C = I ÷ (2πf · V)
60 Hz: 1 ÷ (2π × 60) × 10⁶ = 2652.6 → µF = 2652 × A ÷ V
50 Hz: 1 ÷ (2π × 50) × 10⁶ = 3183.1 → µF = 3183 × A ÷ V
Rearranging the current equation isolates capacitance on one side. Everything else on the right is measurable with two ordinary tools: a clamp meter reads the current flowing into the capacitor, a voltmeter reads the voltage across its terminals, and mains frequency is the most stable number in the entire building. At 60 Hz, the arithmetic produces the famous constant: 1 divided by (2π × 60) is 1 ÷ 376.99, or 0.0026526 farads for every amp-per-volt, and multiplying by 10⁶ to convert farads to microfarads gives 2652.6 — rounded on every service truck in North America to 2652. On 50 Hz mains the same steps run through 1 ÷ 314.16 to give 3183.1, so the constant becomes 3183.
This is why the method beats a bench meter for field work. No LCR bridge, no out-of-circuit handling, no cooling the part down first: the reading you get already includes every real-world effect — can temperature, ripple current, voltage stress — because those effects are inside the measurement, not absent from it.
±5%, ±10% and the Point of No Return
Industry practice grades a run capacitor against its nameplate value in three bands. These are the bands the verdict card above applies automatically, and they are worth understanding rather than just obeying:
Good
The delivered capacitance matches what the motor's start winding was designed around. Phase shift and auxiliary-winding current are where the manufacturer intended them, and no action is needed beyond the next routine check.
Marginal
The system still runs, but the film has drifted past the healthy window and values almost never drift back. Heat accelerates the loss, so a marginal reading in spring is usually a replace reading by August. Plan the swap on your schedule, not the part's.
Replace
Torque is down, run current is up and the compressor is running hot — and a capacitor that has drifted this far is usually an open failure in waiting. Replace it with the same µF and an equal-or-higher voltage rating.
Why is a few percent of drift dangerous at all? Because the start-winding circuit is tuned around the rated capacitance. The auxiliary winding and the capacitor together decide how much phase shift the motor gets and how much current flows through that winding. Let the capacitance sag and the phase shift narrows, the rotating field weakens, and the compressor loses starting torque exactly when it needs it most — on a 40 °C afternoon with head pressure through the roof. The motor compensates by drawing more current for the same load, the windings heat, and the thermal protector starts tripping the system down on the hottest days of the year.
The same logic cuts the other way. More microfarads is not more power: an over-capacitance section pushes excessive current through the start winding too, which is why a replacement should always match the nameplate value rather than "rounding up". The ±5% window protects the motor from both directions of drift, and the bands get wider only because the consequences get more expensive.
Reading the 35/5 Label
Most North American condensers carry a single oval or round can marked something like 35/5 or 45+7.5. That is not one capacitor — it is two electrically independent sections sharing an oil-filled case and a common terminal. The label reads run µF / fan µF: the big section serves the compressor, the small one serves the condenser fan motor, and the three terminals — HERM, C and FAN — are where the outside world connects. Inside, the two sections never touch except at the common lug.
The big section is a true run capacitor: it stays in the compressor's start winding circuit the whole time the compressor runs, providing permanent phase shift. The small section does the same job for the fan motor, which needs far less torque and therefore far less capacitance. Because the two sections age independently, they fail independently — and each must be judged against its own rating, never against the other.
HERM section — compressor
Clamp the HERM lead and measure volts from HERM to C. Compare against the first (large) number on the label. Expect terminal voltage in the 300–450 V range on a healthy 440 V-rated can while the compressor runs.
FAN section — condenser fan
Clamp the FAN lead and measure volts from FAN to C. Compare against the second (small) number. This is the section that quietly dies and leaves a compressor running against a stalled, overheating fan.
The common failure modes follow from the construction. Heat dries or leaks the oil and bulges the can; terminal corrosion adds resistance that mimics capacitance loss; and section bleed — dielectric breakdown that partially bridges a section — shows up as a reading that climbs above rating rather than below it. A dual cap can also fail at the internal common, which makes both sections read open at once. And some failures are mechanical rather than electrical: a bulged top or an oil stain around the vent is a condemned part regardless of what the microfarad number says.
So test both sections, every time. "Compressor side fine, fan side dead" is one of the most common dual-cap findings, and its symptom — a compressor that runs fine while head pressure climbs and the unit short-cycles — is easy to misdiagnose as a failing fan motor. The capacitor is the cheaper suspect; check it first.
How to Measure µF Under Load, Step by Step
The whole measurement takes about ten minutes once the panel is off. The sequence matters: discharge before you touch, measure only at steady state, and discharge again before your hands go back in. Read the safety protocol in the next section before the first step.
- 1
Identify the capacitor and read the label
Find the µF rating, the voltage rating (370 or 440 VAC is typical) and whether the part is single or dual. The rating you compare against is printed on the can — never taken from a parts list. Note the terminal markings too: HERM, C and FAN on a dual, two or three marked lugs on a single.
- 2
Kill power and verify it is dead
Open the disconnect, lock and tag it, then prove the absence of voltage across L1 and L2 with your meter before the panel comes off. Verification takes ten seconds and is exactly the step people skip on the day they get hurt.
- 3
Photograph the wiring
Take a phone photo of the capacitor top with the wires still attached. Three terminals, several same-colored conductors and an interrupted service call are the classic recipe for a miswire — the photo is your reassembly map.
- 4
Discharge through a 20 kΩ 5 W resistor
Hold the resistor leads across each terminal pair (HERM–C and FAN–C on a dual) for about thirty seconds, then verify zero volts with your meter. Never short the terminals with a screwdriver blade: the arc pits the lugs, can weld the blade to the can, and dumps the full stored charge through your hand.
- 5
Restore power and let the system reach steady state
Reconnect, restore power and let the compressor and condenser fan run for a few minutes. Startup-surge readings are garbage — the numbers you want come from a system running at operating temperature and pressure, which is the entire point of testing under load.
- 6
Clamp the start-winding lead
Clamp the ammeter around the single lead feeding the section under test — the HERM lead for the compressor section, the FAN lead for the fan section. On a single capacitor either lead works, because the current that enters must leave. Read steady running amps, not inrush.
- 7
Measure volts across the capacitor terminals
Probe directly across the section's two terminals (HERM to C, for example) while the system runs. Expect roughly 300–450 V on the compressor section of a healthy 440 V-rated can, so set the meter range before the probes touch anything. Record both numbers at the same moment.
- 8
Compute, compare, then discharge again
Enter the amps and volts in the calculator above — µF = 2652 × A ÷ V at 60 Hz — and judge the result against the label using the ±5% / ±10% bands. Whatever the verdict, discharge through the resistor again before your hands go back into the panel.
Clamp the lead, read the volts, do the mathPhoto: UnsplashLive Terminals Are Not a Bench
Under-load testing is the most informative run-capacitor measurement available — and the only one performed on an energized circuit. The rules below are not suggestions; they are the difference between a routine reading and an arc-flash injury. Read them before the first clamp goes on.
- Live terminals: while the system runs, the capacitor sits at 300–450 VAC. Treat every lug, spade and wire in that box as energized at all times.
- Stored charge persists: after power-off a run capacitor can hold a dangerous voltage for minutes. Always discharge through a 20 kΩ 5 W resistor and verify zero volts before touching.
- One-hand rule: keep one hand in a pocket whenever you probe a live circuit, so fault current cannot cross your chest and through the heart.
- PPE: safety glasses at minimum — oil-filled cans can vent violently. Insulated probes, dry hands, no rings or metal watch straps.
- Never short the terminals to discharge. A screwdriver arc pits the terminals, can weld the blade to the can and risks a burn or an eye injury.
- Work dry and grounded: a sweaty grip on a live probe in a humid mechanical room is how routine readings become emergency calls.
- Know the limit: if any part of live-voltage measurement is unfamiliar — or the panel layout does not match this procedure — stop and call a licensed HVAC technician.
Educational aid, not a service manual. This page exists to explain the method and do the arithmetic. It cannot see your meter, your wiring, your equipment model or your training, so it deliberately stops short of instructing anyone to open a live panel. If you are not qualified for live electrical work on 240 V HVAC circuits, use the information here to understand the diagnosis — and let a licensed HVAC technician perform the measurement. No microfarad reading is worth an injury.
Reading the Numbers: What Each Result Means
A µF value is a sentence, not a verdict on its own. High, low, zero and "impossible" each point at a different failure mode — and a couple of them usually point at the measurement itself. Work down this table before you order the part.
| What you see | Most likely cause | What it means for the motor |
|---|---|---|
| µF a little above rating (up to ~+10%) | Section bleed from film breakdown, or a measurement error — film capacitors rarely gain capacitance honestly. | Re-measure with solid probe contact and the voltmeter truly across the capacitor terminals. If the high reading persists, treat it as marginal: the film is failing and the drift only goes one way. |
| µF far above rating | Almost always a misread — supply voltage measured instead of the voltage across the section, or a loose probe on the C terminal. | Fix the measurement before condemning the part. A genuine doubling of capacitance in the field is essentially unheard of; never replace on a reading you cannot repeat. |
| µF 5–10% below rating | Cumulative self-healing losses and heat aging eating away effective plate area. | The compressor still starts, but torque and efficiency are slipping and the trend is one-directional. Schedule replacement before peak cooling season. |
| µF more than 10% below rating | Advanced dielectric loss — the section is near end of life. | Replace now with the same µF and an equal-or-higher voltage rating. The motor is overheating and the thermal protector is the next component in line to quit. |
| Zero amps — no current flowing | Open section: the film has self-healed straight through, or the internal pressure-induced cutout has fired. | The compressor hums but will not start, and the overload trips within seconds. Replace the capacitor; if a good part does not cure the symptom, investigate the start winding and its wiring. |
| Amps high, volts across the section near zero | Shorted section — the metallized plates are bridged internally. | The start winding sees no phase shift and pulls destructive current; breakers trip or windings cook quickly. Kill power immediately, replace the capacitor and check the start winding for damage before restarting. |
| Readings jump around between measurements | Loose or corroded spade terminal, or poor clamp-to-lead contact. | Reseat every terminal and re-test before drawing any conclusion. A flaky connection at the capacitor can mimic every other fault on this list. |
One habit ties the table together: never act on a single anomalous reading. Re-measure, reseat the terminals, and confirm the voltmeter was genuinely across the capacitor and not across the contactor. Half of the "impossible" capacitor readings in the field are really loose spade lugs wearing a capacitor costume.
Related Calculators
The under-load test finishes a diagnosis, but capacitor questions rarely stop there. If you are still decoding or sizing parts, the main calculator on the home page decodes any 3-digit marking into pF, nF and µF with the tolerance range applied; the calculators hub lists every dedicated tool in one place; and the guides library explains the physics behind each result — from EIA tolerance letters to why series banks need balancing resistors. Every tool runs live in the browser, with no account and nothing sent to a server.
Capacitance Converter
Convert capacitance between pF, nF, µF, mF and F live, and translate any value to its closest EIA 3-digit code with the tolerance range shown.
Open calculatorSeries & Parallel Capacitors
Combine up to 12 capacitors in series or parallel with per-cap unit selectors. Live total capacitance, the rule behind it and voltage-sharing notes.
Open calculatorParallel Plate Capacitor
Compute capacitance from plate area, separation and dielectric: C = K·ε₀·A/d with 25 real materials, mm²/µm to sq in/mil inputs and fringing warnings.
Open calculatorCapacitor Energy
Compute energy stored in any capacitor from E = ½CV² with live J/mJ/µJ/nJ/Wh outputs, plus the Q²/(2C) and ½QV forms.
Open calculatorRC Time Constant
Get τ in seconds, ms, µs, ns from any R and C pair, plus the 5τ fully-settled time, charge/discharge curve values and cutoff frequency.
Open calculatorCapacitor Charge
Calculate stored charge Q = CV in C, mC, µC, nC, pC from capacitance and voltage, plus peak displacement current over any R.
Open calculatorCapacitive Reactance
Compute Xc in Ω/kΩ/MΩ from frequency and C, with parallel-R || C and series-R + C impedance, phase angle and admittance.
Open calculatorCapacitor Color Code
Decode 3-band, 4-band and 5-band capacitor color codes to pF/nF/µF with tolerance and voltage, plus a complete reference table.
Open calculatorMotor Capacitor Sizing
Size start and run capacitors for split-phase, cap-start, cap-run and dual-value single-phase motors with HP/voltage inputs and standard ranges.
Open calculatorCrystal Load Cap
Calculate C1 and C2 for a crystal oscillator: C = 2CL − Cstray, with PCB stray cap, transistor input cap and an AT-cut package comparison table.
Open calculatorCoupling & Bypass
Size AC coupling, input/output, emitter and cathode bypass capacitors from impedance and cutoff: C = 1/(2πfR), with an audio stage value table.
Open calculatorRC Filter Cap
Solve C = 1/(2πfR) for low-pass, high-pass, band-pass and notch RC filters, with a live Bode plot, gain readout and an R×C pairs table.
Open calculatorLC Resonance
Get f₀ = 1/(2π√LC) for parallel or series LC tanks, plus Q factor, bandwidth and a capacitor self-resonance (SRF) warning, from any L and C.
Open calculatorPower Supply Cap
Size rectifier bulk, buck-converter output and hold-up capacitors: C = I/(f·ΔV) and C = 2Pt/(V₁²−V₂²), with a ripple and voltage-rating table.
Open calculatorCap Voltage Divider
Split voltage across 2–4 series capacitors instantly: Vn = Q/Cn with equal charge Q = Ceq·Vtotal, plus per-cap bars, charge and rating warnings.
Open calculatorUnder-Load Testing FAQ
What µF deviation from the rating is acceptable?▼
Field practice grades a run capacitor in three bands. Within ±5% of the nameplate value the part is doing its rated work and needs nothing. Between 5% and 10% off, the capacitor is marginal — it will keep the system running, but the drift almost never recovers, so plan a replacement before the next heavy cooling season. Beyond ±10%, replace the part now: the motor is losing torque and efficiency and running hot. Note that the tolerance printed on a new capacitor (often ±6%) describes the factory band for a brand-new part, not the serviceability window.
Can I test a capacitor while the system is running?▼
That is precisely what the under-load method is: the capacitor stays wired in the circuit, the compressor runs, and you measure real amps and real volts while the part is hot and working. The physics only asks for a clamp meter around the lead and a voltmeter across the terminals. What it demands in return is live-voltage competence — the terminals sit at 300–450 V while the system runs. If you are not trained for live electrical measurement, use an out-of-circuit capacitance meter instead or call a licensed technician.
Where does the 2652 constant come from?▼
It falls straight out of the reactance formula. A capacitor's reactance is Xc = 1/(2πfC), and the current it passes is I = 2πfCV. Solve for C and you get C = I/(2πfV). At 60 Hz mains, 1/(2π×60) = 1/376.99 ≈ 0.0026526, and multiplying by 10⁶ to express the answer in microfarads gives 2652.6 — the familiar 2652. At 50 Hz the same arithmetic gives 1/(2π×50) × 10⁶ ≈ 3183.1, which is why technicians on European, Asian and Australian mains divide by 3183 instead.
Why does my brand-new capacitor read about 3 µF high?▼
Manufacturers deliberately ship film capacitors near the top of their tolerance band, because capacitance only drifts downward with age — starting high keeps the part inside spec for longer. A 35 µF ±6% capacitor can legitimately measure 37 µF on day one. Against this calculator's strict ±5% service window such a part may even show as marginal out of the box. That is not a defect: check the tolerance printed on the label, and judge the part by the direction of its drift across seasons rather than by a single reading.
Can a capacitor test good and still be bad?▼
Yes. A µF number is one snapshot. A section can open intermittently only at full operating temperature, a dual capacitor can carry a healthy compressor section and a dead fan section at the same time, and a can with a bulged top or an oil stain should be replaced on sight no matter what the meter says. Heat-driven and mechanical failures are exactly the ones a cold bench test misses — which is the whole argument for measuring the part hot, in circuit, under load.
Is it safe to test a capacitor under load?▼
It is safe only for people trained in live electrical work. The measurement itself is non-invasive — clamp around the wire, probe the terminals — but the terminals live at 300–450 VAC and the capacitor stores a dangerous charge after power-off. Follow the protocol: discharge through a 20 kΩ 5 W resistor and verify zero volts, keep one hand out of the panel, wear eye protection, and never short the terminals. If any of that is unfamiliar, stop and call a licensed HVAC technician. This page is an educational aid, not a substitute for qualified service.
Do 50 Hz and 60 Hz systems give different readings?▼
The physics is identical; only the constant changes. The same amps and volts across the same capacitor correspond to a smaller capacitance at 50 Hz, where the constant is 3183 rather than 2652. Compute a 50 Hz measurement with the 60 Hz constant and you will overstate the capacitance by about 20% — enough to condemn a perfectly good part or clear a dying one. Always match the frequency toggle to the local mains before you trust the verdict.
What actually kills run capacitors?▼
Heat first: a can mounted on top of a condenser lives at 55–60 °C in summer, and every extra 10 °C roughly halves film life. Ripple current, voltage spikes from the utility or lightning, terminal corrosion and oil loss through the pressure vent all shorten life further. Because most of those killers act while the part is hot and loaded, a dying capacitor often still measures fine on a cold bench — and fails the under-load test in the field. That is the single best argument for testing under load.
Put a Number on That Capacitor
Two meter readings and the label value are all it takes: the calculator above applies the field thresholds and hands you a keep-or-replace verdict before the compressor finishes its next cycle.