A hot summer’s evening usually begets long lines at the local ice cream shop. Recently, at one such shop, the line grew extra long when the cash registers and credit card machines mysteriously powered off and on. The youthful operators attributed the problem to the use of all three soft-serve ice cream dispensers on busy days.


This particular facility had a 30-­kilowatt (kW) load fed from a 120/208-volt (V), 400-ampere (A) service. There are two service windows with three registers and one credit-card machine, three soft-serve ice cream dispensers, chest freezers, a walk-in storage freezer, and two induction/microwave ovens and toaster ovens. The three registers are across the storefront, fed from quad outlets in a daisy-chain circuit. The load is fairly consistent during the business hours; however, additional equipment may be used to make cakes and other nonimmediate service items.


One power quality (PQ) monitor was placed on the main service panel to monitor three-phase voltages and current. To record the single-phase voltage and current into the register/credit card machine, a second monitor was placed under the counter into the receptacle for the third station, which is farthest from the panel. Monitoring began approximately 6 p.m. on a Thursday to 6 p.m. Monday. Employees using credit- card machines or registers during that time reported no service interruptions. However, the data collected provided valuable clues as to “whodunit.”


The PQ monitor on the register circuit recorded 13 sags during business hours over the four days. Eleven of them were very similar in magnitude and duration (107V, seven cycles). Two of them were longer (109V, 13–17 cycles). This would suggest that a different type of equipment was the source of the voltage sag. The time interval between sags was not consistent enough to determine a pattern of operation of the equipment causing the sags.


Figure 1 shows the effect on voltage and current during the sags. This is typical of a rectified-input switching power supply found in single-phase electronic loads, such as information technology (IT) equipment. It also shows that it can sustain itself on the charge stored in the input capacitor for four or five cycles. Hence, these short-­duration events aren’t as likely to cause a problem with the equipment unless the sags were more severe due to coincident startup of other equipment operations or reduced nominal voltage due to loading within the facility or from the utility itself during peak-loading effects in the summertime.


All of those seven cycle sags had a similar profile. There is likely a load on the same circuit as the registers that comes on occasionally during business hours and results in a voltage drop due to the large increase in current. Two longer duration sags, similar to what was recorded on the service entrance, had a signature that suggested a load on a parallel circuit starting up.


The PQ monitor on the main service panel noted the three-phase voltages relative to ground as well as the current on the incoming conductors. The power demand (in watts) was fairly regular over the four days of monitoring. At nighttime, it averaged 5–10 kW, while in daytime, it averaged 15–25 kW, peaking at 27 kW. The power demand per phase was reasonably balanced, though phase C is more lightly loaded at times. The phase voltages (line-to-neutral) and phase-to-phase (line-to-line) voltages were fairly well regulated and balanced over the monitoring period, except for times where phase C averages 2–3V lower than the other phases, with a minimum value around 112V, or 8V below nominal. Phase C averages 30A higher than phases A and B during peak hours, which is likely a contributing factor to the voltage unbalance.


At the service entrance, a series of sags occurred on Sunday evening, all of similar magnitude and lasting 22–32 cycles. They occurred nearly 20 minutes apart each time, suggesting a periodic load turning on. The events occurred downstream, which means something in the facility is the cause of the sag. The offending equipment was a 240V load (phases A and C); only two phases had the large current increase.


These longer sags are what is most likely a similar event or disturbance to the cause of the random cash/credit transaction equipment misoperation. In this case, though, it wasn’t severe enough to cause a problem. The sags that originate from the 208V load that causes the disturbance on phases A and C has already occurred on the phase A conductor when another load starts up that causes the voltage to sag further at the cash/credit transaction equipment circuit, creating a lower remaining voltage.


A precise cause/effect determination cannot be made from this data, though it is highly likely that the events observed on Sunday evening are a probable source. The coincidence of multiple equipments drawing large currents at the same time results in sags of various depths and durations. The more equipment that coincidentally energizes to this higher current level, the more likely the misoperation can occur. Further testing was to be conducted to find the likely culprit behind the long lines on a hot summer’s evening.