Lighting fixtures used in a commercial apartment expansion project featured low PF ballast and used two phases sharing a neutral. The result caused excess current on the neutral and a potentially hazardous situation.

When the landlord in a new building agreed to build out two floors of space for a new tenant, he hired an architect and engineer for the design and a general contractor for the actual construction. There was no reason to expect problems. It was a nice-looking space for what would become the lead tenant's. Everything seemed to run smoothly, until move-in day.

Shortly after the tenant took possession of the space, an electrician noticed several lighting circuit conductors (particularly neutrals) were carrying a lot of current. Using his clamp-on ammeter, he found loads as highas 28A. Considering the circuits were rated for 20A (and should not exceed 16A continuously), this arrangement was a potentially serious electrical accident waiting to happen.

Everyone was sure the culprit was harmonics stemming from electronic 3-lamp ballasts. But neither the landlord nor the tenant cared about harmonics, power quality, or any other technical explanations. They just wanted to know if the building was going to burn down, whether to call their lawyers, and who they should sue. Luckily, the project manager had the foresight to call for help before the situation escalated.

The engineer who designed the system laid it out using building standard light fixtures. Using three lamps per fixture, one ballast operated the center lamp; the other operated the two outboard lamps.

By providing two circuits to each fixture, the engineer reduced the number of contactors required to control the inside and outside lamps separately from the building management system.

The power system was 208Y/120V, 3-phase, 4-wire. The building owner did not have enough fixtures on hand for the entire job. Thinking he was doing the owner a favor, the supplier furnished the additional fixtures with 3-lamp switchable ballasts, instead of the original two ballasts per fixture.

Our first step in finding the cause of the problem was to tabulate the loads on each circuit. Of 90 circuits, 15 had overloads. Four were overloaded because the engineer miscalculated. The remaining 11 circuit overloads appeared to be due to the difference between the ballasts originally specified and the ballasts actually provided.

The electrician questioned the basis of serving each ballast with separate circuits derived from different phases. He theorized that the time difference between the Phase A and Phase B sine waves would make current add on the neutral, possibly overloading it. A graphical analysis showed Phase A and Phase B would add current on the neutral. However, the combination of Phase A plus Phase B (fundamentals only, no harmonics) has the same amplitude as Phase C. With a balanced 3-phase linear load, the current on the neutral is zero.

With balanced linear loads on two phases, the current on the neutral is no greater than it would be with just a single-phase load. This analysis ruled out the theory that supplying the ballast with two circuits from different phases was the problem.

Investigating the specification for the ballasts that were actually supplied revealed they were not electronic. The ballasts consisted of a 1- and 2-lamp ballast, all in the same housing. Both ballasts were conventional magnetic units, but the 1-lamp ballast was a "normal" (i.e. low) power factor ballast. The current and power factor ratings for all ballasts were:

• 1-lamp operation = 0.87A at 0.40 PF

• 2-lamp operation = 0.77A at 0.85 PF

• 3-lamp operation = 1.44A at 0.85 PF

Part of the problem was the engineer expected the 1-lamp circuits to draw about half the current per lamp as the 2-lamp circuits. Since the 1-lamp ballast used a normal power factor design (low PF), the 1-lamp circuits (serving half the number of lamps) actually drew more current than the 2-lamp circuits. That explained many of the overloads.

Despite these discoveries, the question of harmonics remained. Fluorescent ballasts (even with magnetic ballasts) generate harmonics. This has been common knowledge since at least 1968, when the NEC began to require a full size neutral conductor on circuits serving discharge lighting (which includes fluorescent) and also electronic equipment.

To avoid problems of overloading the neutral on 3-phase, 4-wire systems, manufacturers of CBM-certified ballasts limit the third harmonic to 33% of the fundamental, so as not to overload the neutral.

Our case involved two phases sharing a neutral at each fixture. Fig. 2, on page 16 (not available online), shows what happens when Phase A and Phase B (each with a 30% third harmonic) combine on the neutral. The waveform seriously distorts. The peak to peak is 1.6 times the peak to peak value of the pure sine wave (fundamentals only).

A true rms meter would indicate the correct value of 0.82A. However, the electrician's average responding meter interpreted this current at 1.13A. This measurement error compounded the problems caused by the ballast substitution and the original miscalculation.

Although the remedy involved new light fixtures for this space, replacing these fixtures was less expensive than rewiring. Luckily, contractors still had other floors to complete in the building. They replaced the fixtures with the 3-lamp switchable ballasts (and low PF 1-lamp operation) from the two floors involved. These "incorrectly supplied" fixtures could be reused in a future build out on a floor not requiring 1-2-3 switching. Although they had to rearrange a few circuits to correct some other overloads, the overall fix cost only $7500.

Elovitz is an electrical and mechanical engineer and attorney with Energy Economics, Inc. in Foxboro, Mass.