How to Document a Circuit You Had to Trace
Why this matters
Tracing a circuit is expensive and the result evaporates. A tech spends hours establishing what feeds what, writes three words on a directory card, and eighteen months later somebody pays for the same hours again. Worse, the three words get trusted. A record that says "kitchen" without saying how that was established, on what date, or what was ruled out is not documentation, it is a rumour with a tech's handwriting on it, and someone will eventually open a device based on it and find it live.
Before the cover comes off
Decide, before you start, whether the trace actually requires the dead front off an energized panel. Most of it does not. A breaker toggle test with a plug-in indicator at the outlet, or observation of a fixed appliance stopping, is performed with the cover in place.
Where the work does require exposure to energized parts, that is energized work under 29 CFR 1910.333(b)(2), which permits it only where de-energizing would introduce additional or increased hazards or is infeasible due to equipment design or operational limitations, and it carries the approach boundary and arc-rated PPE requirements with it. Convenience is not infeasibility.
Where you can de-energize instead, do it, then lock and tag the disconnecting means and prove dead with the meter checked on a known live source, checked on the conductors, then checked on the known live source again. That proving sequence is NFPA 70E-2021, 120.5.
Step 1: Photograph the as-found record before you change it
Even a wrong directory is evidence. It tells you what somebody believed, and when your trace contradicts it you want to be able to show the difference rather than describe it. Photograph the existing card, the panel with the dead front on, and the breaker positions as found.
Skip this and you lose the ability to answer the question that comes up when something turns out to be mis-fed: was it like this before you got here. Silence answers that question badly.
Step 2: Trace with one method at a time, and log the method per circuit
The result of a trace is only as good as the method that produced it, and different methods fail differently. Record which one you used for each entry, because the next reader needs to know how much weight it carries.
- Breaker toggle with an indicator at the load. Direct and strong. Its failure mode is a load that was already off, or a shared neutral arrangement that makes two circuits look like one.
- Observed operation of a fixed appliance. Strong for that appliance, silent about everything else on the same conductor.
- Tone tracer. Useful where you cannot interrupt power, and it induces on adjacent conductors, so it narrows rather than proves.
- Continuity, de-energized. The strongest of all, and it requires the circuit out of service, which is often the reason it was not used.
- Inference from the drawing or from what somebody told you. Not evidence. Record it as inference and say who or what it came from.
A record where every entry says the same thing is a record where nobody logged the method. The mix is the point.
Step 3: Log the negatives
This is the entry that saves the next person and almost nobody writes it down.
When you confirm that a device is NOT on the circuit you expected, that finding is worth more than the positive one, because it contradicts the assumption a reasonable tech would make. "Garage receptacles on the north wall are not on the garage breaker, they are on the bedroom circuit" prevents somebody working a receptacle they believe is dead.
Positives tell you where to look. Negatives stop you from being wrong about where you already looked.
Step 4: Mark confidence on every entry
Three levels, stated in a legend on the record itself so it needs no explaining:
- Confirmed. Direct evidence, method logged.
- Probable. One indirect indication, no contradiction found.
- Unresolved. Traced and not established. Say what was ruled out.
An unresolved entry is a good entry. It tells the next person exactly how much work remains and what has already been eliminated, which is the difference between ten minutes and a repeat of the whole trace. An entry that quietly guesses to avoid a blank is the one that gets somebody hurt.
Step 5: Write the directory in the language the rule requires
When you rewrite a panelboard directory, Article 408 of NFPA 70 requires each circuit be legibly marked with a description specific enough to identify its purpose, and requires that the description not depend on transient occupancy, with unused positions identified as spare. So "bedroom 2 receptacles, north and east walls" satisfies it where a person's name does not, and a blank position gets marked spare rather than left empty.
Practically, that rule is also just good documentation. "Lights" and "misc" are the two entries that cause the most wasted time in the field, because they are true of almost every circuit in the panel.
Step 6: Leave the record in two places
At the equipment. A durable, legible directory inside the panel door, printed or written in a way that survives heat and handling, secured so it cannot fall into the enclosure or contact energized parts.
In the job record. A photograph of the completed directory plus the full trace log, including the methods, the negatives, and the unresolved entries. The card inside the door will not hold all of that and should not try to.
One copy is not enough for a specific reason: the card gets replaced during the next panel change and the detail is gone, while a job-record copy with no field copy means the tech standing at the panel does not have it.
Step 7: Date it, sign it, and state the scope
Date, initials, and one line naming what the trace covered and what it did not. "Traced all breakers in this panel on this date. Subpanel in garage not traced." Undated documentation cannot be aged, and a reader has no way to know whether it predates a renovation.
Worked: a 24-breaker panel, rebuilt
A service call on a house where nothing could be isolated with confidence. The panel had 30 spaces with 24 breakers installed. The existing directory carried 9 legible entries, 6 that had faded past reading, and 9 blank. Those add to 24.
The trace took about 2.5 hours, which across 24 breakers is roughly 6 minutes per breaker. Most of it was walking, not testing.
Method mix, logged per entry. Receptacle circuits were traced by breaker toggle with a plug-in indicator, cover on. Fixed appliances were traced by observed operation. Two runs that could not be interrupted during occupied hours were narrowed with a tone tracer.
Result. Twenty-one of the 24 breakers ended confirmed with direct evidence. Two ended probable: the tone tracer narrowed each to a single run but nothing was interrupted to prove it. One ended unresolved. Those add to 24.
The unresolved one, written out properly. A 20 ampere breaker that operated nothing anyone could find. The log recorded what had been ruled out: every receptacle and fixture in the occupied space had been accounted for on other breakers, and the conductor left the panel toward an exterior wall. It recorded what remained: an exterior or outbuilding load, or an abandoned run. It did not say "spare," because a breaker with a conductor on it is not a spare, and marking it as one invites somebody to reuse the position.
The negatives, which turned out to matter most. Three receptacles in the garage were confirmed to be fed from a bedroom circuit rather than the garage breaker. That is the finding a reasonable tech would never guess and would be wrong about by default.
The payoff, on the next visit. A different tech needed to isolate a bathroom exhaust fan. The directory named the circuit, the log said the entry was confirmed by breaker toggle with an indicator at the load, and he isolated it, locked it, and proved dead in about ten minutes rather than repeating the trace. He also read the garage negative before working a receptacle he would otherwise have assumed was on the garage breaker, and it was live.
The failure mode, if the record had been thinner. Write "garage" on the garage breaker and nothing else, and the same tech isolates that breaker, believes the north-wall receptacles are dead, and finds out otherwise with a hand on a terminal. The trace was correct in both versions. The difference was entirely in whether the negative got written down.
How to verify the record before you leave
- Re-test two entries at random against the written record. One toggle test and one observation. If both match, your logging discipline held during the walk, which is where errors enter, not in the testing.
- Count the entries against the breakers installed. Every installed breaker has a line, including the unresolved one and including any you did not test. A record with fewer lines than breakers reads as complete and is not.
- Read one entry as a stranger would. Pick your worst-worded line and ask whether somebody who has never been in the building could act on it. If it needs you standing there to explain it, rewrite it.
- Confirm the negatives are on the card, not only in the log. The negative findings are the safety-relevant ones and they belong where the person at the panel will see them, even if that means a short note pointing to the full log.
- Check the card is secured and legible with the door closed and reopened. A directory that slides behind the deadfront or shreds on the first close is documentation you will pay for again.
References
- NFPA 70 (National Electrical Code), Article 408, legible circuit directory marking, specificity of description, and identification of spare positions
- 29 CFR 1910.333(b)(2), de-energizing and lockout or tagging before work on or near exposed energized parts
- NFPA 70E-2021, 120.5, verification of an electrically safe work condition including the test-before-and-after instrument check
- See related: How to Map Which Loads Share a Circuit or Supply; How to Reconstruct a Diagram From the Equipment; Proving a Circuit Dead the Right Way