Welcome to EC&M's monthly forum for your power quality questions. This month, Mark McGranaghan, vice president of Electrotek Concepts, and EC&M's editorial director, John DeDad address the topics of leading power factor and electrical noise.

Q. We suspect that the rack equipment in use at a small data center with a load of about 90kVA and a leading power factor (PF) of .89 has integral PF correction components. These components work in such a way that when the rack equipment is lightly loaded, the power factor goes leading. Why don't all UPS systems seem to handle this very well? Are you familiar with any similar situations at other sites? What was done to solve the problem? McGranaghan's answer It's very confusing to talk about leading and lagging PF when you're dealing with loads like computer power supplies. The terms leading and lagging can only be used for one component of the true PF: the displacement power factor (DPF). This component has to do with the PF caused by the fundamental frequency voltage being out of phase with the current. Only the in-phase components of the voltage and current produce real power. The out-of-phase components produce reactive power, which doesn't do useful work.

The PF is leading when the current leads the voltage, and the PF is lagging when the current lags the voltage. True PF is the total relationship between real power (P), or useful work, and apparent power (S), or total power based on the voltage and current magnitudes. It's expressed by the equation PF=P÷S.

True PF is also influenced by the harmonic distortion in the load current. That's the most important effect for loads like computer power supplies. These loads actually have a DPF very close to unity. However, it's often slightly leading because of the capacitors in the power supplies. For instance, the DPF could be .98 leading. Generally, this isn't a problem for UPS equipment. However, when you measure the true PF, you get a reading like .7 due to all the harmonic distortion in the load current. The meter will say .7 leading because the term leading refers to the DPF component.

The bottom line is that it's unlikely the customer has PF correction equipment at its loads — they just happen to be this way. UPS equipment is generally designed to handle computer power supplies just fine. At least the equipment should be able to handle these loads — they're the most common loads requiring backup power.

Q. We've been experiencing problems and shutdowns in our process control system. In discussions with PLC factory technicians, electrical noise and the resulting current ground loops are cited as probable causes. I thought our grounding system would take care of this. How is electrical noise generated and why does it affect our PLCs?


DeDad's answer Electrical noise is characterized by unwanted electrical signals superimposed on a useful waveform. Fig. 1 shows what a waveform with electrical noise distortion would look like. Basically, electrical noise is a repetitive voltage or frequency transient. Power electronic circuits, arcing loads like welders, and switching power supplies (usually associated with computer systems) are the culprits.

Electrical noise usually affects microprocessor-based equipment, such as microcomputers, programmable logic controllers (PLCs), gauges, workstations, monitors, displays, controllers, and data acquisition computers, because of the way these devices work — they operate sequentially, driven by internal clocks operating at millions of cycles per second. Instruction words are brought into the processor from memory and are then executed over the next few cycles. As each instruction is completed, the address of the next instruction is found in the memory. If one sequential address or the instruction it contains is slightly altered, the processor is unable to continue correctly with the programmed sequence, causing an error or halt to occur. It's this inability to tolerate any internal scrambling of these low-voltage bits that makes microprocessor devices so susceptible to problems from electrical noise.

There are two types of electrical noise: normal-mode and common-mode. Normal-mode noise, also known as transverse-mode noise, is the voltage noise that always exists between a pair of conductors. In a 3-phase connection, common-mode noise is noise-voltage appearing equally in-phase with each line-to-line or line-to-neutral voltage.

Common-mode noise appears between the grounding conductor (ground) and line conductors or between the grounded (neutral) conductor and the ground. The latter is usually termed N-G, or neutral-to-ground, voltage. Common-mode noise can destroy the integrity of low-level computer signals by driving stray currents through all the available grounding paths connecting raceways, equipment, and building steel. The result is a current ground loop through the signal circuit.

Fig. 2 shows a local ground loop where one circuit is directly coupled to another and is directly driven. The resistance of the section of the loop that is shared (RG) will determine how much voltage is driven into the target loop by the current of the source loop. Reducing the length of the shared part of the loop will reduce the amount of interference.

You'll find ground loops within a rack set because electronic return paths are shared. All too often there are no provisions for these return currents, so they flow on the only paths available: rack frame steel, bolted and painted rack-to-rack joints, safety ground wires, wireway and conduit surfaces, and anything else available. These chance connections are much worse than any rational intentional path would be. The inevitable system return currents develop voltages of mutually interfering noise over the high impedance of these chance returns. Ground loops are bothersome because the paths on which they flow are poor.