What a Secondary Voltage Depends On

Why this matters

A supply that measures perfectly at eight in the evening and starves equipment at two in the afternoon is the most commonly misdiagnosed condition in commercial service work, because both readings are true and only one of them gets taken. The usual response, tapping the transformer up, makes the afternoon reading better and the night reading worse by exactly the same amount, and a shop that does not understand why ends up owning a new set of failures at the other end of the day.

Three terms decide the secondary voltage

Secondary voltage at any instant is the primary voltage arriving at the transformer, divided by the turns ratio the tap position selects, less the transformer's own internal drop at whatever current is flowing at that moment. Three terms, and they behave completely differently:

• Primary voltage is not yours. It moves with the utility's loading and with the drop in whatever primary and service conductors sit ahead of the transformer. It is generally highest at night.
• The turns ratio is fixed by construction and adjustable only in the discrete steps the tap block offers, most commonly a small number of positions in 2.5 percent increments. It does not vary with anything.
• The internal drop scales with the current being drawn. It is bounded by the percent impedance on the nameplate, which for distribution-size dry types commonly falls in the 4 to 6 percent band, and it approaches that value at full load with a lagging power factor typical of motor load; a sibling article covers reading that off the plate.

The consequence that runs this whole article: only the third term varies with load, and a tap change touches only the second. A tap moves the entire voltage curve up or down. It cannot change the slope. If your secondary swings by a given amount between an idle building and a busy one, it will swing by the same amount after any tap change you make.

What a tap actually changes, and the trap in the labelling

Tap positions on a dry-type transformer are usually on the primary winding, and they are labelled with primary voltages: a 480 V unit will offer positions named for something like 504, 492, 480, 468 and 456 V. The label names the primary voltage that position is designed to receive, not the direction it moves your output.

So the rule that trips people: selecting a lower-numbered primary tap raises the secondary voltage for the primary voltage you actually have, because you have told the transformer to produce full rated secondary from less primary. Half the field mistakes with taps come from reading the tap block as an output adjustment.

Two things follow that are worth knowing before you touch one. Tap changing is not a live task on any transformer that does not carry an under-load tap changer, which almost nothing in a mechanical room does: open and lock the primary disconnect under 29 CFR 1910.333(b)(2), prove the windings dead with the live-dead-live sequence in NFPA 70E-2021, 120.5 using a meter and leads rated CAT III or better at or above the voltage present, and treat the secondary as a possible back-feed source until you have proven that side dead as well. And a tap change is a documented change: mark the new position and the date inside the enclosure, because the next person to read a strange voltage has no way to know the ratio is no longer nominal.

The call: right at eight in the evening, wrong at two in the afternoon

A two-storey commercial building, three rooftop units, all three tripping on the hottest afternoons and none of them tripping in the evening. Two techs had already been out. Both measured in the evening, both got a clean 207 V, and both wrote "supply OK, equipment fault."

The third visit measured twice in one day, at the transformer terminals rather than at the panel, with the same meter and the same leads. Every reading below was taken with the equipment running, which is energized work: 29 CFR 1910.333(a)(1) permits it 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 a load-dependent voltage complaint is that case, because the load is the condition being measured. Boundaries and PPE were established on the basis in NFPA 70E-2021, 130.5 and 130.7, a consensus standard binding through the employer's electrical safety program rather than on its own, and 29 CFR 1910.147 does not apply to this exposure because (a)(1)(ii)(C) excludes work on conductors and equipment in electric utilization installations. The construction counterpart, where the job falls under construction, is 29 CFR 1926.417.

The transformer is a 480 V to 208Y/120 V unit on its nominal tap, so the nominal ratio is 480 divided by 208, which is 2.308 to 1.

Reading it as two separate sags

Divide each primary reading by 2.308 to get what the secondary would be with no internal drop at all:

• At 8 pm: 478 / 2.308 = 207.1 V, and the meter read 207 V. With the building idle there is almost no current, so almost no internal drop, and the measured value lands on the ratio. That agreement is the check that the tap position is actually nominal.
• At 2 pm: 465 / 2.308 = 201.5 V, and the meter read 194 V.

So the afternoon secondary is 13 V below the evening secondary, which is 6.3 percent of that 207 V evening reading, and it splits cleanly into two causes with two different owners:

• The primary itself fell 13 V, from 478 to 465, which is 2.7 percent of the 478 V evening figure. That is upstream of the transformer and belongs to the utility and the primary conductors. Through the fixed 2.308 ratio it costs the secondary 207.1 minus 201.5, or 5.6 V.
• The transformer's internal drop accounts for the rest: 201.5 minus 194 is 7.5 V, which is 3.6 percent of the 208 V nominal secondary. For a unit whose plate impedance is in the usual 4 to 6 percent band feeding a lagging motor load, that is the transformer behaving as designed rather than a defect.

Add the two and 5.6 plus 7.5 is 13.1 V, against the 13 V measured, with the tenth of a volt being rounding. The accounting closes, which is what tells you no third cause is hiding in the numbers.

The tap change, and what it moved

The building's own electrician had been asked to fix the afternoon, and moved the transformer two taps in the raising direction, 5 percent total. Five percent of the 208 V nominal secondary is 10.4 V, applied equally at every hour because the ratio is a constant:

The afternoon complaint improved and the swing did not change by a single volt, which is the demonstration. What the building bought at 2 pm it paid for at 8 pm: 217 V is 4.3 percent over the 208 V nominal, and the line-to-neutral figure that all the small single-phase load in the building sees goes with it, 217 divided by 1.732, which is 125.3 V. That is at the top of the favourable-condition utilization range in ANSI/NEMA C84.1, a consensus standard that binds through the utility tariff or a purchase specification rather than on its own, in the edition the utility references. Motors built to NEMA MG 1, in the edition the manufacturer built to and binding through the motor's own literature rather than on its own, are generally rated to operate at plus or minus 10 percent of nameplate voltage, so they are not the equipment that complains first at the high end. Electronic drivers, small power supplies and lighting control gear carry their own narrower windows in their own literature, which is where a persistent over-nominal night-time condition tends to surface. Treat that as the reason to check those windows rather than as a proven cause of any particular failure.

Choosing the fix that changes the slope

Only three things change the swing rather than moving it, and they are worth naming in the order a shop can actually act on them:

Reduce the impedance on your side of the transformer. Retorquing loose lugs, replacing a corroded splice, and upsizing an undersized feeder all reduce the load-side share of the drop. This is the part a service shop can quote and complete. It does nothing for the 5.6 V that arrived through the primary.

Take the primary sag to the utility with data. A 2.7 percent primary fall between an idle building and a busy afternoon may be entirely ordinary for that service, and the utility will say so, but the request is a different one if you hand them two dated readings at the same terminals rather than a complaint.

Have the swing engineered out. A voltage regulator or a tap changer that operates under load is the device that genuinely changes the slope, and specifying one is an engineering decision based on a load study, not a field call. Say that plainly to the owner rather than implying a bigger transformer will fix it, because a larger unit at the same percent impedance sags by the same percentage at the same fraction of its rating.

Verifying with a two-reading log

One reading proves nothing about a load-dependent supply. The verification that settles these jobs is cheap and repeatable:

1. Read the same two points at both ends of the day and record the primary line-to-line, the secondary line-to-line and the time, on the same meter. Both readings must be at the transformer terminals, because a reading at a downstream panel contains conductor drop you have not separated yet.
2. Divide each primary reading by the nominal ratio and compare against the measured secondary. At idle they should agree within a volt or two; if they do not, the tap is not where you think it is and someone changed it without marking it.
3. Recompute the two shares and confirm they sum to the measured swing. If they do not, there is a third term, and the usual candidate is a loose connection that heats and grows during the loaded period, which is the one cause that gets worse rather than repeating.
4. Repeat once in the opposite season. A supply that swings 13 V in August and 6 V in March is telling you about the building's own load profile, and that is the argument that decides whether the owner spends money on conductors or on a regulator.

References

• 29 CFR 1910.333(a)(1) and (b)(2) for energized work and for de-energizing before work; 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
• ANSI/NEMA C84.1 for nominal system and utilization voltage ranges, applied through a utility tariff or purchase specification
• NEMA MG 1 for motor voltage tolerance, in the edition the motor manufacturer built to, reaching you through the motor's literature
• See related: What a Transformer Nameplate Is Telling You; What a Distribution System Is Actually Doing