Short answer: two legs of a 208Y/120 wye sit 120° apart, so the neutral carries the vector sum N = √(I1² + I2² - I1×I2), and 8.3 A with 2.1 A of 120 V load puts 7.5 A on the neutral, not the 6.2 A a split-phase instinct gives. A clamp that reads 5.9 A is not looking at a mystery feeder: lagging transformer current, a hidden 208 V load and clamp error near the bottom of the jaw range each account for part of the 1.6 A gap. The arithmetic takes two minutes, and three checks settle it on a live panel.
What is the neutral current on two legs of a 208Y/120 feeder?
Use the vector sum N = √(I1² + I2² - I1×I2), so legs reading 8.3 A and 2.1 A put √(68.89 + 4.41 - 17.43) = √55.87 = 7.5 A on the neutral, with both loads at the same power factor. The minus term is 2 × I1 × I2 × cos 120°, and cos 120° is -0.5, so the two currents partly cancel instead of adding or subtracting cleanly. The two single-phase reflexes bracket the answer wrongly: 8.3 A - 2.1 A gives 6.2 A and 8.3 A + 2.1 A gives 10.4 A, while the wye truth sits between them.
Two sanity checks pin the formula down. One leg loaded at 10 A returns all 10 A through the neutral. Two equal 10 A legs give √(10² + 10² - 10 × 10) = 10 A, so a wye neutral never drops to zero and, at equal power factors, never exceeds the larger leg. The same two leg readings give different answers under the assumptions people reach for:
| Assumption | Angle between the currents | Neutral from 8.3 A and 2.1 A |
|---|---|---|
| Split-phase reflex, legs treated as 180° apart | 180° | √(73.30 - 34.86) = 6.2 A |
| Wye, equal power factor on both legs | 120° | √(73.30 - 17.43) = 7.5 A |
| Lagging transformer on the light leg at 0.6 power factor | 173° | √(73.30 - 34.62) = 6.2 A |
Only the 120° row is the feeder's real geometry; the other two rows explain why an honest clamp reads lower.
Why are the two legs 120 degrees apart instead of 180?
A 208Y/120 service is three 120 V windings sharing one neutral point, spaced 120° apart, so any two legs measure 120 V × √3 = 208 V line-to-line and their currents cancel only partly in the shared neutral. A 120/240 split-phase service is the other animal: one winding with a center tap, two legs 180° apart, perfect cancellation, and a neutral that reads the difference. That is where the instinct comes from, and it is the wrong geometry here. In the wye, the two leg currents leave the panel 120° apart and add as vectors, so equal legs leave a full leg's current on the neutral instead of nothing. The 120° spacing is not a quirk of this feeder; it is the geometry the whole three-phase system is built on, and the page on why three-phase power runs 120 degrees apart covers the machine side.
Why does the clamp read less than the formula predicts?
A neutral reading of 5.9 A against a 7.5 A prediction leaves a 1.6 A gap, and three ordinary causes produce exactly that: unequal power factor between the legs, 208 V loads hidden in the leg readings, and clamp error below 10 A. None of these means the feeder misbehaves; the wye math holds. The load shifts give you the tell: as the two legs move toward equal, the neutral falls toward the shared leg value and never reaches zero. Two 5.2 A legs put 5.2 A on the neutral, so a feeder that evens out and still shows amps on the neutral is behaving exactly like a wye. Rank the causes by effort at a pool pit: the pool-light transformer's magnetizing current first, a still-energized 208 V element second, the clamp itself third.
How much does power factor move the neutral current?
If the light leg feeds a 300 W pool-light transformer whose magnetizing current runs at a power factor of 0.6, its 2.1 A lags its voltage by 53°, the angle between the two currents opens to about 173°, and the same leg readings predict 6.2 A instead of 7.5 A. The general form is N = √(I1² + I2² + 2 × I1 × I2 × cos θ), where θ is the true angle between the current vectors. Resistive lighting sits at 1.0 power factor; at light load a small transformer's magnetizing current dwarfs its real load, which is why its power factor lands near 0.6. Which leg carries the lagging load decides the direction: lag on the light leg opens the angle past 120° and lowers the neutral, while lag on the heavy leg closes the angle and raises it.
How do 208 V loads hide from the neutral?
A line-to-line load puts the same amps on both legs and nothing on the neutral, so one 5 A load across L1 and L2 makes each leg read 5 A while the neutral sits at 0 A. Feed those leg readings into the two-leg formula and it predicts a 5 A neutral that does not exist, because magnitudes alone cannot tell a 120 V amp from a 208 V amp. Pool equipment hides such loads well: an ice machine compressor, a heat strip or a transformer primary can bridge two legs while the appliance's controls run at 120 V. The check is fast: switch off every 2-pole breaker in the sub panel, re-read the three conductors, and the gap closes if a line-to-line load was inflating the leg currents.
Can I trust a clamp meter at 6 A?
At 2-6 A on a jaw built for hundreds of amps, treat the reading as good to a few tenths of an amp: typical handheld clamp specifications of ±(1.5 to 2)% of reading plus a few digits apply mid-scale, not at the bottom of the jaw range. Resolution, jaw seating and residual magnetism move the needle there, and an average-responding jaw misreads the peaky magnetizing current of a transformer, so use a true-RMS clamp on that leg. Error sensitivity is lopsided too: in this example the neutral follows the heavy leg almost one for one, so a 10% error on 8.3 A shifts the 7.5 A prediction by about 0.8 A, while a 10% error on the 2.1 A leg shifts it by only about 0.06 A.
How do I verify the neutral current on a live feeder?
Clamp both hots and the neutral together in one jaw and a correct feeder reads near 0 A, then put one known resistive load on the panel: a 1,500 W heater alone draws 12.5 A at 120 V and the neutral should read the same 12.5 A. Work the list and you settle the feeder in about ten minutes:
- All-three check. Both legs plus the neutral in one jaw; the vector sum of a circuit's conductors is near 0 A, and a larger reading means current is escaping through a path it should not have.
- Simultaneous legs. Read the two legs at the same instant with two clamps, or accept a few tenths of slip between one-at-a-time snapshots.
- True RMS on the transformer leg. An average-responding jaw misreads peaky magnetizing current.
- Clear the line-to-line loads. Open every 2-pole breaker in the sub panel before comparing the 120 V case against the formula.
- Resistive-load proof. The 1,500 W heater is 12.5 A at 120 V (1,500 ÷ 120), an easy number to hold a clamp to.
Measure live parts with rated test tools and rated PPE; NFPA 70E sets that practice. While the panel is open, confirm neutral and ground stay on their own bars downstream of the service; the subpanel neutral and ground bonding article covers what a stray bond does to those readings.
What does the two-leg result change in the panel and the box?
Size the neutral like a leg rather than like a difference, one leg at 60 A returning 60 A through it and two equal legs still putting a full leg's current there, and build the pool pit panel in 316L stainless rated NEMA 4X / IP66 by the manufacturer; not UL listed. NEC ampacity rules count this neutral as a current-carrying conductor, and the geometry above is why: it can sit anywhere from zero to the larger leg's current. A neutral splice works as hard as a phase splice, so a bad lug carrying 7 A heats like a bad lug on a leg; torque and inspect it the same way. Where panels and raceways share these return amps, the parallel neutral current in metal raceways article covers what shows up in the steel. The bond belongs at the service disconnect, upstream of the pit; the outdoor subpanel bond removal article walks the correction when one lands downstream.
The pit is also the harshest environment on the job: wet soil, washdown water and chlorine vapor against the box. Chlorides pit 304 stainless even at room temperature, so the material call is 316L; the pool chemical vapor enclosure article works that comparison, and what Type 4X adds over Type 4 is exactly this corrosion margin. Keep the rating whole at every opening: a gland, window or drain rated below IP66 sets the whole assembly at that lower number. Put every cutout on the drawing before the box is cut; the custom enclosure builder cuts 316L to your sizes from one piece, delivered in 10 business days from drawing approval.
Catalog dimensions to compare
Compare these catalog dimensions with your component layout. Open each enclosure for its rating, material and usable space, then check those specifications against your application.
| Enclosure | Material | Outside H x W x D / display | Rating |
|---|---|---|---|
| 12 × 10 × 6 in | 304 stainless | 12 x 10 x 6 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 20 × 16 × 8 in | 304 stainless | 20 x 16 x 8 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 24 × 20 × 8 in | 304 stainless | 24 x 20 x 8 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
Questions
Does the neutral carry current when the two legs are equally loaded?
Yes. Two equal 10 A legs of a 208Y/120 wye put 10 A on the neutral, because the currents sit 120 degrees apart and add to √(10² + 10² - 10 × 10) = 10 A. Only the 180-degree legs of a 120/240 split-phase system cancel, and there two equal 10 A legs give 0 A.
How should the neutral of a two-leg feeder be sized?
Size it like a leg, not like a difference. One leg alone returns its whole current through the neutral, and two equal legs still put a full leg's current on it, so a 60 A feeder needs a neutral rated for the full 60 A. NEC ampacity rules count this neutral as a current-carrying conductor.
Why does my clamp read less neutral current than the formula predicts?
Three causes, usually together: lagging magnetizing current on one leg (a small pool-light transformer at 0.6 power factor cuts a 7.5 A prediction to about 6.2 A), 208 V line-to-line loads that inflate both leg readings while adding nothing to the neutral, and clamp error at 2-6 A on a large jaw. Clamp all three conductors together; a correct feeder reads near 0 A.
Is a NEMA 4X enclosure enough for a pool pit panel?
Pick Type 4X in 316L stainless: chlorine vapor and washdown water pit 304 even at room temperature, while 316L holds against chlorides. Our boxes are rated NEMA 4X / IP66 by the manufacturer; not UL listed. Keep every gland and window at IP66 or better, because the weakest opening sets the assembly rating.
