How to Work Out What a Piece of Equipment Is Actually Fed From

Why this matters

Half the electrical work a service shop does happens in a building whose panel schedule was last accurate during a different tenancy. The cost of guessing is not abstract: a tech locks out the wrong breaker and opens a still-energized disconnect, or shuts down a walk-in cooler to prove a circuit and finds out what was in it the next morning. Establishing a supply path is a procedure with a defined end point, and the end point is not a label on a door.

What counts as evidence, and what does not

Five kinds of clue turn up on these jobs, and they are not equally trustworthy. Ranked by how often each one survives contact with a real building, weakest first:

1. A panel schedule or a legend. The weakest. It records an intention from an installation date, not a state.
2. Conductor colour. Nearly as weak, and worse because it feels authoritative. Colour convention is an installation practice, not a physical property.
3. Conductor size and raceway route. Better, because it is physical, but a raceway can carry more than one circuit and can pass through a junction you cannot see.
4. Overcurrent device rating and pole count matched against the equipment nameplate. Good for narrowing a candidate list to two or three, useless as a single answer, because identical equipment gets identical breakers.
5. A deliberate, observed change at both ends. The only proof. Everything above it produces candidates.

The whole procedure below is a way of climbing that list cheaply, so that the expensive step, the one that interrupts something, is taken once and on the right candidate.

Before you start: the gate and the authorization

Reading a nameplate is not live work. Everything after it is. 29 CFR 1910.333(a)(1) permits work on or near energized parts only where the employer can demonstrate that de-energizing introduces additional or increased hazards or is infeasible due to equipment design or operational limitations, and identifying an unknown supply is one of the recognised cases, because the presence and origin of voltage is exactly what is being determined. Establish the shock and arc-flash boundaries and select PPE on the basis in NFPA 70E-2021, 130.5 and 130.7, a consensus standard that reaches you through your employer's electrical safety program or your contract rather than on its own authority. Use a meter and leads rated CAT III or better at or above the voltage present, keep one hand clear of the enclosure, and stand off the hinge line when a cover comes off.

The second gate is the one techs skip. Interrupting a circuit to prove it is a deliberate outage, and the outage is a hazard you create. Never use interruption as the test on a circuit that serves life safety or fire alarm equipment, a medical or laboratory load, refrigeration holding temperature-critical product, sump or ejector pumping, or any process that cannot be restarted safely from a cold stop. For everything else, get the customer's authorization for a named window in writing before the first breaker moves, and ask specifically what is on that panel that they would notice losing, because the answer is regularly something nobody wrote down.

The third gate is that a piece of equipment may have more than one source. A transfer switch, an uninterruptible supply, an inverter or a control transformer fed from elsewhere can all keep part of an assembly energized after the obvious breaker is open. Proving one source is dead is not proving the equipment is dead.

Step 1: read what the equipment says it needs

Take the nameplate first: voltage, phase, frequency, and either a full-load current or the pair of numbers packaged equipment carries, minimum circuit ampacity and maximum overcurrent protective device rating. This costs nothing and it sets the target. Skipping it means you match candidates against a guess, and every later step inherits that guess.

Understand what those two numbers are before you use them. Minimum circuit ampacity is a conductor sizing number, built from 125 percent of the largest motor's rated load current plus the rated currents of the others; it is not the current you should expect to clamp. The maximum overcurrent device rating is a ceiling, so a device larger than it is a finding in its own right under Article 440 of the NEC as adopted in the edition in force in your jurisdiction, not a sign you found the wrong circuit.

Step 2: measure what the equipment actually has

At the equipment's own terminals or line-side disconnect lugs, take all three line-to-line readings and all three line-to-ground readings. This tells you the voltage class and the system configuration, which immediately eliminates every panel in the building that is not that configuration. A sibling article covers what those line-to-ground readings mean; the short version is that a leg reading roughly 208 V to ground on a 240 V system is a high leg and normal, and that a system with nothing bonded gives line-to-ground readings that are not voltages at all.

Step 3: follow the raceway as far as your eyes go

Trace the conduit or cable physically from the equipment toward the building, noting size, raceway type, every junction box and every point where it disappears into structure. Photograph each junction. The value here is not that you will follow it all the way, it is that you will usually eliminate two of the four panels in the building because the raceway leaves in the wrong direction.

Step 4: build a candidate list from the devices

At each surviving panel, list every device whose pole count matches the equipment's phase count and whose rating is at or below the nameplate maximum overcurrent device rating. Do not stop at the first plausible one. A candidate list of three is a normal, useful result at this step; a candidate list of one usually means you stopped early.

Step 5: correlate with a clamp before you interrupt anything

With the equipment running, clamp one conductor of each candidate device in turn and record the current. This is still not proof, because two similar loads can read similarly, but it is nearly free and it usually leaves one candidate standing. Do not expect the clamp reading to equal the minimum circuit ampacity, for the reason given in Step 1.

Where the equipment cycles on and off, the stronger version of this test is to have a second person switch the equipment at its own controls while you watch the clamp. A candidate whose current moves with the equipment's cycling, twice, in both directions, is a very strong candidate.

Step 6: prove it with a coordinated interruption

Inside the authorized window, open the surviving candidate device and confirm at the equipment's own terminals that all legs have gone dead, using the live-dead-live sequence in NFPA 70E-2021, 120.5: prove the meter on a known live source, prove the conductors dead, prove the meter again on the known live source. Then close the device and confirm voltage returns. Both halves matter. Voltage disappearing tells you the candidate feeds this equipment; voltage returning tells you nothing was damaged and no other source is quietly holding it up.

If the work that follows is on the equipment or its conductors, this is where lockout applies: lock and tag the device under 29 CFR 1910.333(b)(2) for the electrical exposure, and separately under 29 CFR 1910.147 for any mechanical or stored-energy hazard such as an unexpected fan or compressor start. Releasing stored ELECTRIC energy is part of the electrical isolation itself under 29 CFR 1910.333(b)(2) rather than an afterthought: run and start capacitors, drive DC buses and UPS batteries hold a charge after the disconnect opens, so wait the manufacturer's stated discharge time, prove each terminal dead, and on any bus of size apply a shorting and grounding device rated for it before your hands go in, noting that 1910.147 excludes the electrical exposure itself at (a)(1)(ii)(C). Where the job falls under construction, the electrical counterpart is 29 CFR 1926.417.

Worked example: three candidates and one clamp

A three-phase rooftop condensing unit, 208 V, nameplate minimum circuit ampacity 34.6 A and maximum overcurrent device 50 A. The panel schedule says "ROOF" against four different circuits.

Step 2 gives 207, 208 and 207 V line to line and 120, 120 and 120 V line to ground, so it is a grounded wye system and the two 240 V panels in the building are out. Step 3 puts the raceway into the north electrical room. Step 4 finds three three-pole candidates there: two rated 50 A and one rated 60 A. The 60 A device exceeds the nameplate maximum of 50 A, so if it turns out to be the feed, the finding is a device that needs correcting rather than a mismatch that clears it as a candidate; it stays on the list.

Step 5, with the unit running on a mild afternoon:

Candidate B is carrying something small and constant, almost certainly not a running condenser. Candidate C is carrying more than this unit's whole minimum circuit ampacity, which is possible but points at larger equipment. Candidate A at 25.6 A is the shape you expect: a compressor in the low twenties plus a pair of condenser fans, comfortably under the 34.6 A minimum circuit ampacity because that number was built at 125 percent of the compressor's rated load current and is a sizing figure rather than a running one.

A tech who took the 34.6 A on the plate as the expected reading would have rejected Candidate A for reading too low and gone after Candidate C, which is the single most common wrong turn in this procedure.

Step 6, in a window the customer authorized in writing after confirming nothing temperature-critical shares the panel: opening Candidate A drops all three legs at the rooftop disconnect to zero, proven live-dead-live on a known source before and after. Closing it restores 207 V on all three. The path is now established rather than believed.

When you cannot open the breaker

Some circuits cannot be interrupted at any hour, and the honest answer is that you will not get proof, you will get a better candidate. Two methods narrow it further:

A signal tracer injects a tone onto a conductor and follows it with a receiver. It works well on a de-energized branch circuit and it degrades on energized runs in a crowded raceway, because the signal couples into neighbouring conductors and the receiver responds to the strongest field rather than the correct conductor. Treat a tracer result as a strong candidate, and where two adjacent circuits both respond, say so rather than picking one.

A momentary load signature works where the equipment itself can be cycled without an outage. Have the equipment started and stopped on its own controls while a logging clamp sits on the candidate, and match the timestamps. Two matched transitions in each direction is a defensible result, and it is often the best available on a live process.

Whichever you use, write down that the path is unconfirmed. A candidate recorded as proven is worse than no record at all, because the next tech will lock out on your note.

Leaving a record that survives you

Label both ends, in the same visit, with the same identifier: the panel name and circuit number on the equipment or its disconnect, and the equipment description and its location on the panel schedule. Add the date and the initials of whoever proved it. Where the proof was a clamp correlation rather than an interruption, write "correlated, not interrupted" on the schedule.

Photograph the panel directory after you update it and attach it to the job record. The next call on that building, by anyone in your shop, starts at Step 4 instead of Step 1, and that is the compounding return on ten minutes of writing.

References

• 29 CFR 1910.333(a)(1) and (b)(2) for energized work and de-energizing; 29 CFR 1910.147 for mechanical and stored-energy lockout, with its electrical carve-out at (a)(1)(ii)(C); 29 CFR 1926.417 for lockout and tagging of circuits in construction
• NFPA 70E-2021, 120.5, 130.5 and 130.7, applied through an employer electrical safety program or contract
• NEC Article 440 as adopted, in the edition in force in your jurisdiction, for the minimum circuit ampacity and maximum overcurrent device markings on air-conditioning and refrigeration equipment
• See related: Why a Transformer Configuration Changes What You Can Measure; What a Distribution System Is Actually Doing