What a Continuity Check Proves and What It Misses

Why this matters

The beeper is the most-used function on a field multimeter and the least-examined. It is fast, it is a dead test so it feels safe, and it produces a yes. Techs clear whole circuits with it, write "rang out good" on a ticket, and go looking somewhere else.

A continuity check asks one question: is there a path from here to there that a milliamp or so of test current can travel? That is a question about topology. Whether that path can carry the circuit's actual working current without dropping most of the voltage across itself is a completely different question, and the beeper does not ask it, does not answer it, and gives no indication that it has not.

This card is about the gap between those two questions, and about a joint that beeps, reads a plausible-looking handful of ohms, and still starves the load it feeds.

Before you ring anything out

A continuity or resistance test is a dead-circuit test. Applying it to an energized circuit gives a meaningless reading, can damage the instrument, and puts you in contact with a circuit you have declared dead in your own head. Isolate under 29 CFR 1910.333(b)(2), apply your lock and tag, and verify absence of voltage with a tester proved live, then dead, then live again on a known source (NFPA 70E-2021, 120.5, binding through your employer's electrical safety program or your contract in whatever edition that program has adopted).

The second half of this card requires a voltage-drop reading taken while the circuit is working, which is live work. That reading exists only under load, which is one of the recognized cases under 29 CFR 1910.333(a)(1) where de-energizing is infeasible because of operational limitations, but it is a determination you make and defend, not a default. If you take it: work to the shock and arc-flash boundaries your employer's electrical safety program sets, and provide and use the electrical protective equipment required by 29 CFR 1910.335(a). A 24 V control circuit inside an enclosure that also contains line-voltage terminals is a line-voltage job, and the low control voltage is irrelevant to the boundary you are standing inside.

What the test actually does

The instrument puts a small, fixed voltage across your leads and measures the resulting current. The test current is typically on the order of a milliamp, though it varies by instrument and by function, and the specification sheet for your own meter is where that number lives.

That small test current is the whole story. Ohm's law does not care that it is small, so the voltage developed across a resistance during the test is proportionately small too. A joint that will drop a significant fraction of the supply under working current drops almost nothing under a milliamp, which is exactly why it does not announce itself.

The beeper is a threshold detector

The beep is not a measurement. It is a comparator: below some resistance the tone sounds, above it the tone stops. That threshold is set by the instrument's designer, it varies widely between instruments, and on many meters it is not printed anywhere on the case. A meter that beeps below several tens of ohms and a meter that beeps below a few ohms will give opposite answers on the same corroded splice, and neither will tell you which one you are holding.

So the first discipline is to stop using the tone as the reading. Switch to the resistance function and read the number. Then subtract your lead resistance, which you get by touching the probes together first, because on a low-resistance measurement the leads themselves are a meaningful share of what you are reading.

That change alone converts a yes-or-no into a number, and a number can be compared against a budget. The rest of this card is about building that budget.

What "a path exists" does not mean: capacity

Whether a path is good enough is not a property of the path alone. It is the relationship between the path's resistance, the current the circuit draws, and how much voltage the load can lose and still work. Those three together set an allowable resistance, and nothing about a continuity test touches any of them.

The arithmetic is one line. Voltage lost in the path equals the working current multiplied by the path's resistance. Turn it around and the allowable resistance equals the voltage the load can afford to lose divided by the working current. A high-current load with a tight voltage tolerance has almost no resistance budget; a low-current load with a wide tolerance has a lot.

The load's tolerance is not something to invent. Pull-in voltage for a coil, minimum operating voltage for a board, and acceptable drop for a branch circuit all live in the equipment manufacturer's literature or in the governing installation standard, and that is where the number comes from.

The case: a contactor that would not pull in

A rooftop unit whose compressor contactor sometimes closed and sometimes did not, more often late in the afternoon. The control circuit is nominally 24 V from a transformer, running out to a remote thermostat and back through a long conduit run with a splice in a pull box halfway.

The first tech isolated, locked out, verified dead, and rang out the control run end to end. It beeped. Ticket read: control wiring continuity confirmed, suspect contactor. Contactor replaced. Fault returned in under two weeks.

The second tech went back to the same splice with the resistance function instead of the tone.

The number the resistance test gave. 18 ohms across the run, after subtracting lead resistance. On its own that number means nothing, because 18 ohms is fine in some circuits and fatal in others. It has to be run against a budget.

Building the budget. The coil's own literature states its pull-in voltage; say for this coil it is 19 V, and its inrush current at pull-in is 0.6 A. Both figures are illustrative and stand in for the ones in your own equipment's literature.

• Voltage the circuit can afford to lose: 24 V supply minus 19 V pull-in is 5 V.
• Allowable total circuit resistance at pull-in current: 5 V divided by 0.6 A is about 8.3 ohms.
• Measured: 18 ohms in the wiring alone, before anything else in the circuit is counted.
• 18 divided by 8.3 is about 2.2, so a single splice was carrying more than twice the entire circuit's resistance budget.

What that predicts, and what confirming it looked like. At 0.6 A through 18 ohms, the drop across the splice is 0.6 multiplied by 18, which is 10.8 V. The coil would see 24 minus 10.8, or 13.2 V, which is 55 percent of nominal and well below the 19 V it needs. The prediction is that the coil cannot pull in at all through this run.

That prediction is what turns a suspicion into a test. The second tech re-energized under the live-work conditions above and read voltage directly across the splice with the thermostat calling. The meter read close to 10.8 V across two inches of wire, which is the whole finding in one number: 10.8 V that should be at the coil is being spent in the splice instead.

Why it worked in the morning. Contact resistance in a corroded joint rises with temperature, and the coil's own pull-in voltage is not perfectly sharp. Early in the day, with the joint cooler and its resistance somewhat lower, the coil was seeing just enough. By afternoon it was not. The mechanism runs the same way at both ends of the range: cooler joint, lower resistance, less drop, coil pulls in; hotter joint, higher resistance, more drop, coil drops out. The intermittency was not random, it tracked the one variable that moves the resistance.

The failure mode this cost. A contactor was replaced for a wiring fault. The replacement was correct hardware fitted for the wrong reason, it did not change the drop, and the fault came back with a part on the invoice. That is the standard shape of a continuity-driven misdiagnosis: the test cleared the thing that was broken, so the investigation moved to the thing that was not.

What would change the answer. Move the same 18 ohms into a circuit that draws a tenth as much current with the same 5 V allowance and the allowable resistance becomes 5 divided by 0.06, or about 83 ohms, and 18 ohms is comfortably inside budget. Nothing about the joint changed; the budget did. This is why a resistance reading without a stated current and a stated tolerance is not yet an answer.

Four other ways a continuity result misleads

It cannot tell one path from two. Test current will happily travel a path you did not intend: through a parallel device, a backfed common, a shared neutral, or another leg of the same control circuit. A beep proves a path exists between the probe tips, not that it is the path on your drawing. Lift one end before you believe a continuity result on anything with more than one route between the two points.

It cannot see a fault that is not present when you test. A joint that opens with vibration, with thermal expansion, or only when the equipment is running is closed and quiet on the bench. A clean continuity result on an intermittent complaint is close to no information at all, and treating it as an exoneration is how the same fault survives three visits. Load-testing the path, or reading drop across it while the equipment runs, is what reaches those.

It cannot confirm identity. Two conductors in a bundle, two terminals on a strip, a numbered wire that was re-pulled and not re-labelled. The instrument reports on whatever is between the probe tips and has no opinion about whether those are the points you meant. Confirm both ends physically before the reading, not after it disagrees with you.

It can read open on a joint that would carry load. This is the mirror of the whole card and it catches people who have learned the first lesson. A thin oxide or tarnish film on a contact face can block a milliamp of test current and still be punctured by working current, so a meter can call a joint open when the circuit works. If a resistance test says open and the equipment demonstrably runs, do not disbelieve the equipment. Re-seat the probes, apply firm contact pressure, and confirm with a test that uses more current before you cut anything out.

What a continuity result is genuinely worth

It is excellent at proving a hard open: a broken conductor, a blown element, an unseated terminal, a switch that never closes. Those are its home ground, and it finds them faster than anything else you carry. It is also a fast, cheap way to confirm a wiring path before you energize something for the first time, and to prove a fuse or a thermal cutout has opened.

State the result at that level. "No open in the control run" is defensible and useful. "Control wiring good" is a claim about capacity that a continuity test never made.

How to verify you got this right

• Did you read a number or listen to a tone? If it was the tone, you do not know the threshold and you do not have a reading.
• Did you null the leads? On anything under a few tens of ohms, unsubtracted lead resistance is a meaningful share of the result.
• Do you have a working current and a voltage tolerance to divide? Without both, a resistance reading has no budget to be judged against and cannot support a pass.
• Did you lift one end? If more than one path exists between the two points, you have not tested the one you named.
• Does the complaint involve intermittency, heat, or vibration? If so, a clean dead-circuit result did not clear the path, and a drop reading under working conditions is the test that will.

References

• 29 CFR 1910.333(b)(2) - lockout and tagging of circuits for dead-circuit work on electrical installations
• 29 CFR 1910.333(a)(1) - the requirement to de-energize, with the infeasibility and increased-hazard exceptions that cover a measurement only available under load
• 29 CFR 1910.335(a) - use of electrical protective equipment for work on or near energized parts
• NFPA 70E-2021, 120.5 - verification of an electrically safe work condition, binding through your employer's electrical safety program or your contract in the edition adopted
• See related: Continuity vs Voltage vs Resistance: Which Test to Run; What a Resistance Reading Tells You and What It Hides; How to Load Test a Component Instead of Trusting a Static Reading