Testing a Run Capacitor: Bench, Under Load, and What Each Misses

The short version

A capacitor that reads good on the bench has not been cleared. The bench capacitance measurement is taken with the part out of the circuit, and it shows that the measured capacitance is inside its range. That measurement does not establish that the motor and the rest of the circuit are working, which is the question you were called out for.

These are two different tests, answering two different questions. A meter with capacitance mode measures the part disconnected. Amperage and voltage readings on a running system give you an estimated capacitance while the capacitor is working, and that is energized work with rules of its own, covered below. These are two measurements rather than two verdicts, and neither one is the complete answer.

What the capacitor is doing

A capacitor motor uses a second winding carrying current out of phase with the main winding, and the capacitor is what produces that phase shift. Not every single-phase motor is built this way.

Fluke describes a capacitor motor as a split-phase motor with one or two capacitors added.
Fasco’s literature separates permanent split capacitor and capacitor-start designs from other single-phase types including shaded pole.
Oriental Motor’s fundamentals describe the main and auxiliary winding arrangement for the types that use one.

Run capacitors stay in the circuit while the motor runs. Start capacitors drop out once the motor is up to speed.

Fluke’s guidance on the instrument side states the failure modes plainly: a capacitor can short, it can go open, or it can deteriorate physically to the point of failure. Fluke notes that deterioration can change capacitance, which is what a capacitance measurement is looking for.

A dual run capacitor serves two motors from one can. The label shows two values, something like 45/5, and three terminal groups: C for common, HERM for the compressor, FAN for the condenser fan. Copeland’s wiring documentation identifies HERM as the compressor connection and FAN as the outdoor fan. Go by the terminal marking and the equipment wiring diagram, not by which number on the label is larger.

Getting to it safely

The part stores energy and it can hold that energy after the power is off.

Isolate the equipment and keep it isolated. What matters is that the circuit cannot be re-energized while your hands are in it, not a fixed number of switches. Lock-out requirements come from your employer’s program and the rules where you work, so follow those rather than a count of disconnects. Fluke’s motor procedure is to open the safety switch or combination starter, lock and tag it out, and then verify voltage at the terminals. HVAC School’s capacitor procedure is to pull the disconnect or otherwise ensure the unit is without potential.

Verify dead on everything you will be exposed to, and prove the meter twice. The Occupational Safety and Health Administration (OSHA) rule is written around exposure rather than a count of readings: test the circuit elements and equipment parts the worker will be exposed to, including for voltage that arrived by backfeed or induction. On the ordinary two-line single-phase circuit, that comes to the three checks in HVAC School’s procedure: L1 to L2, L1 to ground, and L2 to ground.

Three-phase equipment, circuits with a neutral, and equipment with more than one source need more checks than three. The absence-of-voltage test in NFPA 70E, the National Fire Protection Association’s electrical safety standard, covers every phase-to-phase and phase-to-ground combination, which on a three-wire three-phase circuit ordinarily comes to six readings. Neutrals, control power, and any other source that can reach the parts you are exposed to get tested too.

Use a meter you have pre-tested against a known live source, and prove it against that live source again once the dead check is done. Fluke describes this live-dead-live sequence as what NFPA 70E requires. The first check is for a meter that has already failed, which would read zero on a live circuit and tell you the circuit is dead. The second is for one that failed while you were using it.

Discharge the capacitor with a tool made for the job, not a screwdriver. HVAC School describes bridging the terminals with an insulated screwdriver as a common practice among technicians, and a controversial one. Even the manufacturers’ pages are not uniform about the practice, and some Carrier consumer material describes the screwdriver method.

The strongest published statement is Trane’s, in its service literature: screwdrivers and other hand tools are not designed to safely discharge capacitors, and using them can result in death, serious injury or equipment damage. The short a screwdriver makes is uncontrolled, and HVAC School’s own caution about discharge concerns the arcing. Use the tool made for the job.

On equipment carrying a flammable refrigerant, there is a second reason. HVAC School’s A2L guidance is to always use a proper capacitor discharger on A2, A2L or A3 systems, because that work starts with keeping ignition sources away from the refrigerant. An uncontrolled arc across the terminals is exactly the risk that instruction is written against.

The tool made for the job is a resistor or a purpose-built discharge tool, and the published values are procedure-specific rather than universal.

Fluke’s capacitance guidance gives 20,000 ohms at 5 watts for five seconds, followed by verification with the meter.
Fluke’s own motor troubleshooting procedure gives 20,000 ohms at 2 watts.
Fasco’s literature gives a different value again for a particular bench test.

Use what the procedure you are following specifies, and check the equipment instructions where they exist.

Discharge both sections of a dual capacitor, not one. A dual capacitor is two capacitors sharing the common terminal, and discharging one section establishes nothing about the other. HVAC School’s procedure addresses HERM to C and FAN to C separately for that reason. Treat the section you have not touched as charged until the meter says otherwise, because the terminal you are about to put a hand near is the one you did not address.

Verify with a voltmeter, because a spark is not a test. No arc does not prove there was no charge, and one arc does not prove the charge is gone. The verification is a voltage reading, not capacitance or resistance. Put a voltmeter rated for the capacitor’s voltage on the voltage function across each section, C to HERM and then C to FAN, and read zero volts before your hands go near the capacitor.

Trane’s service literature has the same instruction. That literature also gives a reason not to assume the part discharged itself while you were opening the panel: many capacitors have internal bleed circuits, but those circuits can fail, and some take up to 30 minutes to finish. Copeland’s component literature and Fasco’s motor literature both describe discharge arrangements that differ by capacitor type and by how the motor is wired.

Photograph the wiring before you pull anything. HVAC School’s procedure calls for taking a picture before removing wires, because a dual capacitor has three terminal groups with multiple wires landing on them.

The bench test

This is the familiar version of testing a capacitor. HVAC School defines the bench test as removing both leads after the safety sequence above is complete, power verified off and the capacitor discharged, then putting a meter that measures capacitance across the terminals.

The capacitor has to be out of the circuit. Fluke’s instruction is to detach it before measuring, to avoid measurement errors. In Fasco’s literature, discharge comes before removal, and a microfarad-reading tester is among its field checks. What form the error takes depends on the circuit and the meter, so the useful rule is the instruction itself rather than an explanation of why.

Set the meter to capacitance mode. Fluke notes the symbol usually shares a spot on the dial with another function, so you often have to press a mode button as well. Leave the leads on for a few seconds and let the meter settle. The meter is charging the capacitor with a known current, measuring the resulting voltage, and calculating capacitance from that. Between autoranging and that charge cycle the displayed capacitance can take a few seconds to settle, and how the display behaves in the meantime varies by meter.

On a dual capacitor, out of the circuit means all the way out: three terminal groups can carry more than two wires, so disconnect every wire. Then you are taking two readings, not one. C to HERM covers the compressor side, and C to FAN covers the fan side. When both readings are done, put the wires back from the photograph, checked against the equipment wiring diagram.

HVAC School’s guidance is to bench test in these situations:

When the system is not running.
When the motor is not turning at all and the capacitor is a suspect.
On blower capacitors, because you do not want to be in the wiring with the system running.

HVAC School presents bench testing as one of two options rather than as the test that settles the question.

The under-load test

The other method reads the capacitor while the system runs, and it needs a different capability rather than a different set of tools: reliable amperage and voltage readings instead of capacitance mode.

HVAC School’s procedure requires an accurate multimeter that can measure both, and one clamp multimeter covers both, provided the meter and leads are rated for the circuit and the environment and get inspected before use. That condition is OSHA’s rule for test instruments, not a preference.

The clamp goes on one specific wire: the wire between the capacitor terminal and the motor’s start winding. On a dual capacitor that means the wire leaving HERM for the compressor or FAN for the fan, not just any wire on the can. HVAC School’s under-load diagram places the amp clamp between the capacitor and the start terminal.

Read start winding amps there, measure the voltage across that same section of the capacitor, and calculate. HVAC School gives the formula as start winding amps multiplied by 2,652, divided by capacitor voltage, which returns microfarads.

The answer is only as good as what goes into it: the two readings, the right wire and terminal pair under the clamp, and the right frequency constant. If you doubt a capacitance reading, HVAC School’s guidance is to retake it, reposition the clamp or check it against another meter, and to confirm a weak under-load result with a bench test.

That constant is frequency-dependent, not universal. It comes from capacitive reactance, which WEG’s motor documentation states as one over two pi times frequency times capacitance. Rearranged for microfarads, the constant is one million divided by two pi times frequency. At 60 Hz that computes to about 2,652.6, which the field references round to 2,652. At 50 Hz it is about 3,183, which is the constant HVAC School gives for 50 Hz equipment. Use the constant that matches the supply you are standing in front of.

What comes out is a capacitance estimate, and nothing more. The arithmetic turns the measured current, the measured voltage and the line frequency carried in the constant into an estimated effective capacitance. You then compare that estimate against the microfarad rating and tolerance on the label.

The arithmetic does not, by itself, identify every way a capacitor can fail. It does not show why a current or voltage reading is abnormal, only what capacitance that pair of readings implies.

HVAC School advises against running this test on a blower capacitor. The reason is that meter leads and a spinning blower wheel in the same cabinet is how leads get wrapped. Bench test blower capacitors instead.

The reason to bother is what the capacitance represents. HVAC School describes the under-load reading as a more accurate picture of the operating capacitance, and in other material calls the bench measurement slightly more accurate as a measurement. Both are right in their own sense: the bench measurement is the more accurate number, while the under-load reading shows the capacitance under real conditions.

The reason not to is that you are measuring inside an energized cabinet with operating equipment, which can mean a running compressor and condenser fan. That is energized work, and OSHA’s electrical rules reserve it for qualified persons. Being a qualified person there means being trained for live circuits and familiar with the precautionary techniques, protective equipment, insulating and shielding materials and insulated tools the conditions call for.

OSHA’s baseline is that exposed live parts get de-energized unless the employer can demonstrate that de-energizing adds hazard or is infeasible. HVAC School’s own guidance is to run the test only with proper protective equipment and only when it is safe. Whether you run it at all is governed by those rules as carried out through your employer’s electrical safety program, never by how careful you feel.

	Bench test	Under-load test
What you need	Meter with capacitance mode	Multimeter that reads amps and volts reliably
System state	Off, discharged, leads removed	Running
What it tells you	The measured capacitance of the part	An estimated operating capacitance while running
Use it when	System is down, or it is a blower capacitor	System is running and the energized-work conditions above are met
Main limitation	Taken out of circuit, so it does not represent operating conditions	Live work; the result is a capacitance estimate only, and HVAC School advises against it on blower capacitors

Reading the result

Where a tolerance is printed on the capacitor, that is the specification.

The percentages you will hear quoted are not one tolerance, and they are not all the same kind of thing.

Packard’s run capacitor literature gives 5 percent.
HVAC School describes about 6 percent as a common tolerance, and separately uses 10 percent as a field rule for when to replace.
Fluke’s motor troubleshooting procedure accepts a reading within 20 percent of the printed rating.

Some of those are part specifications and some are procedure thresholds, and the two are not interchangeable: a generic procedure number or a field rule does not widen the tolerance printed on the part.

Daikin’s service manual treats the under-load result the same way: compare the result against the tolerance stated on the capacitor, and replacement may be recommended once the result falls outside that tolerance. The range you work to comes off the part in front of you, or off the equipment or motor manufacturer’s specification where nothing is printed, not off a remembered figure.

Where the tolerance is printed, the range comes straight off the label: rated value times one minus the tolerance for the floor, times one plus the tolerance for the ceiling.

OL does not automatically mean the capacitor is open. Fluke’s guidance is that OL indicates either a capacitance above the meter’s measuring range or a faulty capacitor. OL is the meter’s over-limit display, meaning the value is beyond the range it is set to. So before treating OL as evidence of a fault, check the meter’s capacitance range against the microfarad rating printed on the can. Whatever the capacitance, you have to interpret it against the specified range for that part and that procedure, and Fluke lists short, open and physical deterioration as separate failure conditions rather than one.

A reading inside tolerance is a pass on the measurement, not on the call. Fluke describes three ways a capacitor fails: short, open, and physical deterioration. A capacitance measurement shows the failures that change the measured value.

An in-tolerance reading establishes the capacitance under the meter’s test conditions, with the part out of the circuit rather than doing its job. It does not establish that the motor, the wiring, the controls or the supply are right, and it does not override what you found on the can.

An in-tolerance reading also does not exhaust the ways the capacitor itself can fail. Carrier’s service literature distinguishes a volt-ohmmeter resistance check from a capacitor analyzer, which reads microfarads and catches insulation breaking down under load. So a part can pass a simple low-voltage check while failing in a way that check never sees.

If the capacitance looks fine and the symptom is still in front of you, keep diagnosing. The rest of the circuit, and the capacitor behaviors your instrument did not assess, are both still open questions.

Check the part physically while it is in your hand, which by this point in the sequence means power verified off and both sections discharged. HVAC School’s guidance is that a bubbling top or oil leaking is a replacement regardless of what the meter says, and that ordinary rust on its own is not a reason to condemn a capacitor.

The replacement starts with what the manufacturer specifies; where substitution is allowed, voltage goes equal or higher, never lower, and the microfarads move only where the manufacturer’s own literature allows. Nidec’s motor literature gives that requirement first: always use the correct capacitor as specified by the motor manufacturer.

Where a substitution is permitted, a 440 volt capacitor can replace a 370 volt capacitor carrying the same rated capacitance. Fasco’s literature allows a higher voltage rating where necessary, while saying the rated capacitor size should not be changed. Fasco describes the opposite direction, a lower voltage rating, as impairing capacitor life rather than causing an immediate failure. That is a reason not to do it rather than a promise about when the capacitor will quit.

On the microfarad side, Carrier’s guidance is to match the exact microfarad and voltage requirement, and Copeland publishes run capacitor values per compressor model. Copeland’s application bulletin for one reciprocating compressor family also shows what an authorized exception looks like. That bulletin allows only the specified value, and if that capacitor is not available, a capacitor rated 5 microfarads higher, at equal or higher voltage, may be substituted.

The required value, and any permitted substitution, belongs to the equipment rather than to a general rule.

Common mistakes

Isolate the equipment so it cannot be re-energized, then verify dead on every conductor you will be exposed to, meaning L1 to L2, L1 to ground and L2 to ground on a two-line single-phase circuit. Prove the meter against a known live source both before the dead check and again after it.
Discharge with a purpose-built discharge tool or the resistor your procedure specifies, then confirm the discharge with a voltmeter on the voltage function rather than assuming the discharge worked.
Photograph the terminals before removing a single wire.
Take two readings on a dual capacitor, C to HERM and C to FAN, identifying the sides by the terminal markings rather than by which label number is larger.
Read the tolerance printed on the part, where there is one, and calculate the range from that rather than from a remembered percentage.
Look at the top of the can and the base for bulging or oil while the part is in your hand.

Don’t

Do not close the diagnosis on a bench reading alone when the symptom is still there, because the measurement is taken out of circuit and clears the capacitance, not the motor, the wiring, the supply, or the failure modes a low-voltage capacitance test does not see.
Do not read OL as proof the capacitor is open before checking what OL means on your meter. On the meters Fluke describes, OL also indicates a capacitance above the measuring range, and some testers signal open and shorted explicitly instead; the instrument manual defines the display.
Do not discharge one section of a dual capacitor and move on, because discharging one section does not establish that the other is discharged.
Do not short a capacitor with a screwdriver or any other hand tool. Trane’s service literature is blunt about the practice: hand tools are not designed to discharge capacitors, and using them can end in death, serious injury or damaged equipment.
Do not judge discharge by an arc or the lack of one. You verify with the voltmeter, on the voltage function.
Do not measure capacitance with the part in circuit, because Fluke’s procedure calls for detaching it to avoid measurement errors. The under-load method is its own separate measurement, not an in-circuit capacitance reading.
Do not run an under-load test on a blower capacitor. That rule is HVAC School’s, and the reason is that meter leads and a spinning blower wheel are a bad combination.
Do not replace a capacitor against a remembered percentage when the tolerance is printed on the part in front of you.
Do not use 2,652 on 50 Hz equipment, because the constant is set by line frequency and the 50 Hz figure is 3,183.
Do not fit a lower voltage rating than the original.
Do not replace a capacitor that keeps failing without checking why: whether the installed value and the operating limits match the equipment, and whether the part simply reached the end of its life.

Frequently asked questions

How do I test a run capacitor with a multimeter?

Isolate the equipment so it cannot be re-energized, then verify dead on every conductor you will be exposed to, meaning L1 to L2, L1 to ground and L2 to ground on a two-line single-phase circuit. Use a meter proven against a known live source before the check and proven again after it, and discharge the capacitor.

Fluke’s capacitance procedure uses a 20,000 ohm resistor across the terminals for five seconds followed by verification with the voltmeter. The wattage Fluke specifies differs between that procedure and its motor troubleshooting procedure, so follow the procedure you are working from.

Disconnect every wire from the capacitor so it is fully out of the circuit, set the meter to capacitance mode, and read across the terminals. On a dual capacitor that is two separate readings, C to HERM and C to FAN.

What tolerance is acceptable on a run capacitor?

Where a tolerance is printed on the capacitor, that is the specification, and no procedure number or field rule widens it. The two figures in circulation describe different things: HVAC School gives about 6 percent either way as a common capacitor tolerance and 10 percent as a field rule of thumb for replacement. Calculate the range from the printed tolerance rather than from either shortcut, and where nothing is printed, work from the equipment or motor manufacturer’s specification rather than an assumed percentage.

Why does my capacitor test good but the unit still will not start?

A capacitance reading tells you the measured capacitance is inside its range under the meter’s test conditions, and that is the whole of what it tells you. It does not test the motor, the wiring, the controls or the supply. If the capacitor is within specification and the motor still will not run, keep diagnosing: the motor itself, the supply, the controls, the wiring and the mechanical load are all still open questions. An in-tolerance reading is a reason to keep looking at the rest of the circuit rather than a reason to stop.

Can I test the capacitor without removing it from the circuit?

Fluke’s capacitance procedure calls for detaching the capacitor from the circuit before measuring, to avoid measurement errors. Fasco’s literature has the capacitor discharged before removal, and describes the field tests separately. There is a separate method that reads a running system through amperage and voltage rather than through capacitance mode.

Is it safe to discharge a capacitor by shorting the terminals with a screwdriver?

No, shorting the terminals with a screwdriver is not a safe discharge method. The practice is common, and HVAC School describes it as widespread and controversial and cautions about the arcing. Trane’s service literature is direct: screwdrivers and other hand tools are not designed to safely discharge capacitors, and using them can result in death, serious injury or equipment damage.

Use a resistor or a purpose-built discharge tool; Fluke and Fasco both publish resistor procedures, with different values for different procedures, so use what the procedure you are following specifies. On A2, A2L or A3 refrigerant systems HVAC School’s guidance is to always use a proper capacitor discharger, because the work starts with keeping ignition sources away from the refrigerant. And no discharge is finished until a voltmeter reads zero volts across both sections of a dual capacitor.

Can I use a capacitor with a different rating?

Start with what the equipment or motor manufacturer specifies: Nidec’s motor literature requires you to always use the correct capacitor as specified by the motor manufacturer. Where a substitution is permitted, voltage rating can go higher than the original but never lower, so a 440 volt part can replace a 370 volt part with the same rated capacitance. HVAC School’s guidance is explicit on that direction.

The microfarad rating is not flexible the same way. Per Fasco’s literature, the rated capacitor size should not be changed, and Carrier’s guidance is to match the exact microfarad and voltage requirement. Copeland publishes run capacitor values per compressor model, with one of its bulletins allowing a capacitor rated 5 microfarads higher when the specified capacitor is not available. The value, and any permitted substitution, belongs to the equipment manufacturer rather than to a rule of thumb.

What if the same capacitor keeps failing?

Find out why before fitting another capacitor, and leave room for the answer to be the part itself. The causes HVAC School names include a capacitor that is simply the wrong part, installed after a motor or compressor change or fitted wrong in the first place. HVAC School also names poor manufacturing quality, overtemperature and overvoltage. WEG’s capacitor manual makes the manufacturer’s side of the same point: temperature, voltage and current have to stay within the product’s limits. That manual also notes that capacitors simply reach the end of their service life, which nothing external caused.

Check the installed value and the operating conditions before fitting the next capacitor. Copeland’s application guides publish model-specific data, run capacitor values included. Confirm the installed value is what the equipment calls for before replacing the capacitor again.

Glossary

A2L: the ASHRAE class for mildly flammable refrigerants, which changes how you discharge a capacitor.
Bench test: a capacitance measurement taken with the capacitor out of the circuit.
Capacitance: the charge a capacitor stores per volt, measured in microfarads.
Fasco: a motor manufacturer whose field checks are cited here.
Fluke: a test instrument manufacturer whose troubleshooting guidance is cited here.
Hz: hertz, cycles per second, which differs between 60 Hz and 50 Hz supplies.
HVAC School: a practitioner training and reference organization cited here.
Lockout and tagout: locking and tagging a disconnect so the circuit cannot be re-energized while you work.
Microfarad: the unit capacitance is measured in, printed on the capacitor can.
NFPA 70E: the National Fire Protection Association’s standard for electrical safety in the workplace, whose absence-of-voltage test is cited here.
OL: the over-limit display on a meter, meaning the value exceeds the selected range.
OSHA: Occupational Safety and Health Administration, whose verification rule is cited here.
Run capacitor: the capacitor that stays in circuit while the motor runs, shifting phase to keep it turning.
Under-load test: a capacitance figure calculated from motor amps and volts while the motor runs.

Drafted with AI assistance and reviewed by the author.

← All posts