What a Distribution System Is Actually Doing

Why this matters

A large share of the equipment complaints a shop gets paid to chase are written up as equipment faults and are really supply faults. A compressor that trips on the hottest afternoon, a control board that resets whenever a large motor starts nearby, a heat strip that measures right and never quite gets there: all three read as bad parts on the bench and as a soft supply on a meter. The distribution system is the part of the job nobody documented, nobody warrantied, and nobody put a probe on, and it is where the second trip lives.

The one job the system has

The distribution system exists to hold voltage inside a window at the equipment terminals while the current it delivers swings by more than a factor of ten between an idle building and a summer afternoon. That is the whole assignment. It is not asked to deliver a number, it is asked to hold a range under a load it cannot predict.

The window itself comes from ANSI/NEMA C84.1, a consensus standard that binds through your utility's tariff or a purchase specification rather than on its own authority, in whatever edition the utility references. For a 208 V nominal system its favourable-condition (Range A) utilization floor sits near 191 V, roughly 8 percent below nominal, with a slightly wider Range B for infrequent excursions. Two consequences follow immediately, and they are the reason this article exists: a voltage can be genuinely inside the window and still be the reason your equipment misbehaves, and a voltage that moves when load is applied is telling you about impedance in the path regardless of where it lands.

The path, and what each stage is trading

utility primary
   |
transformer changes voltage, adds impedance
   |
service panel overcurrent protection starts here
   |
feeder and subpanel
   |
branch circuit
   |
equipment terminals the voltage window is measured HERE

Every stage in that path buys the reader something and charges for it:

• The transformer buys a usable voltage and charges impedance. Its percent impedance sets both the fault current available downstream and how far the secondary sags when load arrives. A sibling article covers reading that off the nameplate.
• The conductors buy current-carrying capacity and charge voltage. Ampacity is a thermal limit set by the conductor's insulation temperature rating, corrected for ambient and for how many current-carrying conductors share the raceway, per Article 310 of the NEC in the edition your authority having jurisdiction has adopted. It says nothing about whether the voltage arriving at the far end is usable. The 3 percent and 5 percent voltage-drop figures techs quote sit in informational notes in that same code, which are advisory rather than enforceable requirements.
• The overcurrent devices buy conductor protection and charge nothing else. A breaker protects the wire, not your equipment, under Article 240 as adopted in your jurisdiction. This is why a breaker that trips is usually reporting a real condition on that conductor rather than being defective.
• Every lug, splice and set of contacts buys a connection and charges resistance in series with the load. At no load that resistance is invisible. At load it is the entire fault.

Five things the system does not promise

The fastest way to stop misreading a supply is to be clear about what it never agreed to do. There are five.

1. A fixed voltage. It promises a window under stated conditions, and the window is wide enough to contain equipment misbehaviour.
2. A clean waveform. Nothing in the path guarantees a sine wave free of harmonic content, and the loads inside the building are usually the source of what is there.
3. Protection of your equipment. Overcurrent devices protect conductors. Equipment protection is inside the equipment: overloads, internal thermal protectors, phase monitors.
4. An accurate drawing. Panel schedules are historical fiction in most buildings older than their last tenant.
5. Three equal, continuous legs. Balance and continuity are conditions to be measured, not features to be assumed. Two siblings cover what an imbalance and an outright open do to a motor.

Reading the supply from the equipment end

The equipment terminals are the only place the window means anything, and the readings that matter are taken with the equipment running, which is energized work. 29 CFR 1910.333(a)(1) permits live work 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 voltage measurement to diagnose a load-dependent fault is the case that qualifies. Work it inside your employer's electrical safety program: establish the arc-flash and shock boundaries and select PPE on the basis set out in NFPA 70E-2021, 130.5 and 130.7, which is a consensus standard binding on you through that program or your contract rather than on its own. Use a meter and leads rated CAT III or better at or above the voltage present, keep one hand out of the enclosure, and route leads clear of any moving fan before you close a door on them. The moment you stop measuring and start repairing, open the disconnect, lock and tag it under 29 CFR 1910.333(b)(2) for the electrical exposure, and prove the conductors dead using the live-dead-live sequence in NFPA 70E-2021, 120.5. Note the fork: 29 CFR 1910.147 does not govern this exposure, because (a)(1)(ii)(C) carves out work on conductors and equipment in electric utilization installations; 1910.147 is what you cite for the unexpected startup of the fan or compressor, and 29 CFR 1926.417 is the construction counterpart for lockout and tagging of circuits.

The method itself is two readings, not one. Take the voltage with the load off, then take it again with the load running and everything else in the building unchanged. The difference between them is the impedance of everything upstream of your probes, measured by the load itself. That is a better instrument than any drawing.

Worked example: splitting an 18 volt sag

A three-phase packaged unit on a 208 V nominal service, complaining only in the afternoon. Same two probes, two locations, three minutes apart, with no other load in the building changing state.

Start with the equipment end, because that is where the window applies. The sag is 212 minus 194, or 18 V, which is 8.5 percent of that same 212 V no-load reading. Against the 208 V nominal, the 194 V running figure sits 14 V low, or 6.7 percent below nominal, which is still above the roughly 191 V Range A utilization floor stated earlier. So the voltage is technically inside the window and the equipment is still starving, which is exactly the trap: a tech who takes one reading, sees 194 V, and calls it "in spec" has answered a question nobody asked.

Now split it. At the moment of load, the panel is at 203 V and the equipment is at 194 V, so 9 V of the 18 is being lost between the panel and the equipment terminals: that is the feeder, the subpanel, the branch conductors and every connection in between, and it is entirely on the owner's side of the meter. The panel itself fell from 212 V to 203 V, another 9 V, and that half sits upstream of the service: transformer impedance, service conductors, and the utility's own primary. Nine plus nine is the 18 V measured at the equipment, so the accounting closes.

The reason this matters commercially is that the two halves have different owners. Torquing every lug in the building fixes at most half of an 18 V problem. A utility that is asked to look at its transformer will measure at the service, see 9 V of sag it considers ordinary, and close the ticket. The honest report says the load-side share is 9 V of 18, that it is worth chasing because connections and undersized runs are correctable, and that even a perfect load side leaves the equipment at 203 V under load unless the source is addressed too.

One caution on the arithmetic: those two locations were read sequentially by one tech, so they only add up because nothing else in the building changed between them. If the readings are taken while other loads cycle, the split is invalid and you need two meters or a logger. In an occupied building at two in the afternoon that is the normal case rather than the exception, which is exactly when the complaint arrives, so the two-meter version is worth owning if you chase supply complaints at all.

What would change this read

A single-phase reading on a three-phase supply is not a supply read at all. If the three legs are not equal at the same instant, the "sag" you measured may be an imbalance you sampled at its worst leg, and the diagnosis changes completely. Take all three legs before you split anything.

A sag that is present with the load off, or that does not recover when the load stops, is not a distribution impedance problem. That is a source voltage problem or a tap problem, and it belongs with the transformer, not with the conductors.

If the sag appears only when a different piece of equipment starts, the impedance is real but the trigger is inrush, and the fix may be on the starting equipment rather than on the supply.

Verifying you actually proved it

The split above is a hypothesis until a deliberate change confirms it. Two checks, and neither requires the customer to spend anything before you know:

Move the load, not the meter. If a second, similar load exists on a different feeder from the same panel, run it instead and repeat both readings. An upstream share of roughly the same size on both feeders confirms it is upstream. A share that follows one feeder confirms the problem is in that feeder.

Read the connections thermally, under load, from outside the enclosure. A joint carrying a disproportionate share of the load-side drop runs hotter than its neighbours doing the same work. Do it with the covers on where the equipment allows an infrared read through a rated viewing port, which most light-commercial and residential gear does not have, so on that stock this check is simply not available at the boundary you are working in and the move-the-load check above is the one you have, and do it as a live-work task under the same 1910.333(a)(1) gate and the same 70E-2021 130.5 boundary determination described above, because removing a panel cover to see the joints is what turns a temperature survey into an arc-flash exposure.

Log both no-load and loaded voltages on every visit for a supply complaint. Two dated pairs of readings taken a season apart will tell you whether a building is getting worse, and that is the single most useful piece of evidence a shop can hand an owner who is deciding whether to call the utility.

References

• 29 CFR 1910.333(a)(1) and (b)(2), OSHA general industry requirements for working on or near energized parts and for de-energizing circuits; 29 CFR 1926.417 for the construction counterpart
• NFPA 70E-2021, 120.5 (verification of an electrically safe work condition) and 130.5 and 130.7 (risk assessment, boundaries and PPE), binding through an employer electrical safety program or contract
• NEC Articles 240 and 310 as adopted, in the edition in force in your jurisdiction, for overcurrent protection of conductors and for ampacity and its correction factors
• ANSI/NEMA C84.1 for standard nominal system and utilization voltage ranges, applied through a utility tariff or purchase specification
• See related: What a Transformer Nameplate Is Telling You; How to Work Out What a Piece of Equipment Is Actually Fed From