How to Verify Documentation Against the Equipment in Front of You
Why this matters
A document you have not checked against the equipment is not information, it is a hypothesis with a company logo on it. Prints get revised, models get built in option packages, serial breaks change wiring mid-production, and equipment gets modified in the field by people who never wrote anything down. Every one of those makes a genuine, current, correctly-printed document wrong for the unit in front of you. The cost of skipping the check is not that you get no answer, it is that you get a confident wrong one and act on it.
What "verified" has to mean before you use the word
Verified does not mean the document looks like the equipment. It means you have tested specific claims the document makes against specific facts on the unit, you know how many claims you tested, and you know what you did not test. Anything short of that is familiarity, which feels identical and protects nothing.
The whole procedure below produces one of two states, and they carry different working rules. Verified: you may read the document and act on what it says within the scope you checked. Unverified: you may read the document for orientation and hypotheses, but every point you act on gets metered first.
Step 1: Settle the electrical state before you handle anything
If verification means opening an enclosure and touching conductors or terminals, open the disconnect and apply your lock and tag first, because 29 CFR 1910.333(b)(2) requires the circuit be de-energized and locked or tagged before work on or near exposed energized parts. Prove it dead with the full sequence: meter on a known live source, meter on the conductors, meter back on the known source (NFPA 70E-2021, 120.5).
If verification is purely visual through an open cover with no reach inside, say so out loud and keep it that way. The failure here is the drift: a look becomes a point becomes a nudge becomes a hand on a terminal.
Step 2: Build the identity chain from the equipment outward, never inward
Transcribe the model and serial from the unit's own nameplate, with your own eyes, before you look at what the document says. Then compare.
Order matters and it is the most-skipped instruction in this article. If you read the document first, you are looking for confirmation, and a partial model match will satisfy you. If you transcribe first, you are comparing two independently obtained strings.
Never source the identity from the work order, the customer's file, the prior invoice, or the tag somebody zip-tied to the conduit. Those all descend from the same original entry, so agreeing with them proves nothing. If the plate is unreadable, that is its own procedure, and it comes before this one.
Step 3: Read the document's revision and applicability block
Two things live in the title block or the footer and both are load-bearing:
- Revision identifier and date. A document with no revision identifier at all cannot be positioned in a sequence, which means a newer one may exist and you would never know. That document is a hypothesis on arrival regardless of how well it matches.
- Applicability. Model range, serial range or serial break, and sometimes a date range. A serial break is a real thing: manufacturers change wiring mid-production and publish the effective serial. Your unit's serial being outside the printed range is a hard stop, not a technicality.
Step 4: Resolve the option package before you resolve anything else
Most family prints cover every option the model can carry, with notes reading "when equipped," "factory option," or "remove jumper when installed." Those notes are the difference between a print that matches and a print that appears not to.
Go through the print and list which optioned items it shows, then determine which of them your unit actually has. Everything absent from the unit and present on the print is explained; everything present on the unit and absent from the print is not, and is a finding.
Step 5: Run the three-anchor check
The anchor rule: three anchor points per document per enclosure, one drawn from each of three classes.
- Supply side. Something about how power arrives: the disconnect arrangement, the overcurrent device class and rating, the conductor count into the enclosure.
- Control side. Something about the control power: a transformer tap landing, a control voltage source, the terminal a specific control conductor lands on.
- Optioned component. Something whose presence or wiring depends on the option package resolved in step 4.
Accept the document as verified for that enclosure only if all three agree. The Boolean is AND, not OR. One disagreement drops the document to unverified for the whole enclosure, not merely for the circuit the disagreement showed up on, and the verification burden steps up from those three points to every point you intend to rely on, metered individually before you act on it.
The reason it is three from three different classes rather than three of anything: the supply side is the part of a design that changes least between revisions, so three supply-side checks will pass against almost any document in the family and tell you nothing.
Step 6: Scan for field modification before you accept a passing result
A document can be genuinely correct for the unit as built and useless for the unit as it now exists. The tells, in the order they are easiest to see:
- Conductor jacket, colour, or gauge that does not match the surrounding factory harness
- A conductor entering sheet metal through an opening with no bushing where factory conductors use grommets
- Mixed terminal styles on the same strip, fork on one landing and ring on another
- Splices inside an enclosure where the factory harness is continuous
- Handwritten tags, tape flags, or marker on the enclosure interior
- A component carrying a manufacture date later than the unit's own build date, which makes it a replacement and makes its wiring a place the document may no longer describe
Two or more tells clustered in the same section of the panel is where you look hardest, because that is a place someone worked.
Step 7: Mark the document with what you checked
On the print itself, or on the photograph of it you are carrying: the date, your initials, the three anchor points you used and their results, and one line naming the scope you verified. "Verified for the control enclosure, not for the heat section" is a complete and honest statement. An unmarked print gets re-verified by the next tech from zero, or worse, gets trusted because it looks handled.
A worked verification: three anchors, one disagreement
Rooftop packaged unit, family print pulled from the door pocket, revision letter and date legible, unit serial inside the printed serial range. Disconnect open, locked, tagged, live-dead-live proved on a known source before the cover came off.
Anchor 1, supply side. The print calls for a specific fuse class and rating in the disconnect's fuse block. The fuses in the block carry that same class and rating stamped on the body, and the block label agrees. Match.
Anchor 2, control side. The print shows the control transformer primary landed on the tap corresponding to the supply voltage present at the unit. Measured supply voltage falls in that tap's band, and the primary lead is on that tap. Match.
Anchor 3, optioned component. The print shows a factory jumper across two control terminals with the note "remove when the optional control is installed." On the unit the jumper is absent, and no optional control is installed anywhere the print says it would be. Disagreement.
Two of the three anchors agree and one does not, so under the rule as stated the result is unverified for that enclosure. Not "mostly verified." The whole point of an AND gate is that it does not average.
What that changed in practice: the job needed four points from that print, the contactor coil path, the fan interlock, one sensor pair, and the supply landing. Under a verified document those get read. Under an unverified one they get metered, one at a time, cold, before anything is acted on. The three anchors took about 0.2 hours. The stepped-up verification of those four points took about 0.5 hours. Both are absorbed effort hours on the same visit, which is the only reason they belong in the same sentence.
What the metering found: the missing jumper's function was present, carried by a field conductor routed the long way around the panel rather than across the two adjacent terminals. So the print was correct about the function and stale about the routing, and the unit was fine. The document ends the visit marked "unverified, routing deviation at the control jumper, function confirmed by meter," which is exactly what the next tech needs and is not what "print matches unit" would have told them.
What would have made this a stop rather than a proceed. If the disagreement had landed on a protective function rather than a routing detail, the finding is not a documentation note, it is a defeated safety, and the unit does not go back on until it is corrected. That case has its own article in this category.
Two ways this check quietly passes when it should not
Verifying against the paperwork instead of the unit. The model on the work order, the model in the customer's asset list, and the model on the print all frequently descend from a single original data entry. Checking them against each other is checking a copy against its own source. Only the nameplate is independent, which is why step 2 puts transcription first.
Verifying a section that was replaced wholesale. If the control section was swapped as an assembly at some point, everything inside it will be internally consistent and consistent with a print, just not necessarily this unit's print. The catch is the build date on the replaced assembly compared to the build date on the unit nameplate. If the assembly is newer than the unit, the print that covers the unit does not automatically cover the assembly, and you need documentation for both.
References
- 29 CFR 1910.333(b)(2), de-energizing and locking or tagging circuits before work on or near exposed energized parts
- NFPA 70E-2021, 120.5, verifying an electrically safe work condition, including the before-and-after instrument check on a known source
- Manufacturer documentation for model applicability, serial breaks, and option notes, which is where the applicability block in step 3 comes from
- See related: The As-Found Drawing and Why You Make One; When the Drawing and the Installation Disagree