The how, what and where of power conditioning, part 1.

Depending on the situation and problem, power conditioning may not be the solution. And, in the majority of instances, it isn't.

At past EC&M Harmonics & Power Quality Conferences, we were asked a number of questions on selecting power conditioning equipment: "Where should I begin in determining how to provide conditioning?"; "What selection choices do I have?"; and "How do I determine how much to spend?"

We've addressed a wide variety of power quality "ailments" and suggested solutions for each here in PQ Corner. What's been missing is a comprehensive overview of power conditioning. Maybe we've been hesitant in this coverage because the emphasis on the word "power" in power conditioning might distract needed attention away from grounding, transient protection, or harmonic interaction.

Certainly, these latter subjects command a more prominent place in power quality analysis. In fact, power-related problems come in last in all the problem-solving sessions of which we've been a part. This is partially due to the greater number of solutions that key into grounding issues first, transient protection second, and harmonic-related issues third.

Industry-wide problem

There still are many who assume, at the onset, their problems are caused by their power source. Usually, these are people who have no power conditioning equipment but think they need it to protect their sensitive devices from a "disturbance event." And, it's very difficult to convince them that concerns other than power may be equally important. Our standard question to a client asking for an estimate on a UPS system is, "If you install this power device with battery backup and it does not solve your problem, what will you do next?" The standard response is astonishment: "How could this much expense not solve the problem?" Here's our chance to advise on grounding-related issues and other areas of power quality. But, it's only as a secondary discussion; the UPS system still has top interest.

As a result, we continually have to remind end users of what actually is happening to upset their sensitive loads and how these loads are behaving under the influence of various disturbances.

To verify if the disturbance is power-related, you first should determine whether it correlates with any power-related event at the facility or on the power lines serving the location. If the correlation is supported by a "time stamp" of power-related events, you can investigate the nature of the event and consider available alternatives. These may include separating the sensitive equipment from the disturbances or mitigating the disturbances with a power conditioner.

If, as is more often the case, there is no power-related event observed at the time of the upset, you should consider other conditions as prime candidates for the upset.

A case history example

A hospital was searching for answers as to why its new variable air volume (VAV) controllers were misbehaving when mysterious spikes hit its electrical system. (VAVs are the energy-efficient alternative to the old method of throttling and directing air flow by using mechanical vanes.)

We connected a monitoring device to the power distribution system and obtained the trace shown in Fig. 1 (on page 33). This appeared to be a very high-speed surge peaking at well over 200V when measured on a system that could stand only 20V to 30V peak before the circuit was damaged.

The client looked at this spike and immediately concluded that the electric utility had transmitted this disturbance on its power line. We explained that while there may be certain conditions that could be blamed on the utility, this certainly was not one of them. We knew there was no known power activity taking place while we were making the measurement, and we could verify this by contacting the utility. Sure enough, the utility verified that no storm activity, sagging, or circuit breaker operations took place at the time in question.

Let's look at the trace more closely to see if we can derive some information. First note the time scale: Not seconds, not milliseconds, but microseconds! We are looking at 128 microseconds, or one-eighth of one millisecond, for the entire trace as shown. Second, note that the "wavefront" of the spike rises almost straight up vertically. With the above analysis of the trace, it's obvious that it would be physically impossible for the energy represented by this spike to have traveled over any distance. If the spike did travel a distance, it would look more like the one shown in Fig. 2 (on page 33). We can see that time (represented by the sloped dotted line) would be needed for the transmission of this disturbance. This was the confirmation for rejecting the theory that the utility was the source. The problem source was obviously nearby.

Upon investigation, we found arcing in a motor connection on one of the hospital's elevators. To protect the VAVs, we not only recommended repair of the elevator motor connection but also suggested the use of in-line reactors or chokes ahead of the VAVs to soften any high-speed surges.

Note that what started out as a power supply concern turned out to be a solution targeted inside the facility.

Another case history example

A mechanical contractor installed micro-processor-controlled compressors, some with variable speed drives and some with across-the-line starters. Convinced that the power company was delivering "bad" power, the contractor was ready to order power conditioning equipment to smooth out any disturbances affecting the microprocessors.

To examine the site for the type of equipment upsets, we connected a disturbance analyzer on the incoming power lines. We noted only four or five incidents of increased voltage in a 2 1/2-week period, and then only 2% to 3% above our threshold. This certainly did not warrant a power conditioning device. And, we found that the compressor equipment and microprocessor controllers were erratic whether the drives were on or off.

Ground reference problem. With further investigation, we discovered one problem: The lack of a stable ground reference for the microprocessors. This showed up when we traced the supposedly continuous ground for both the power system neutral and the case of the microprocessor, the point where the processor obtained its signal reference. While a shielded isolation transformer was used, the connections were incomplete, leading to an inadvertent ground loop. This defeated the value of the transformer shield as a means of rejecting electrical noise. Remember, we're trying to establish an equipotential signal reference: The same potential for both the power system neutral and the safety (case) ground, with no impedance between the two.

Also, the facility wiring had grown "like topsey;" thus, we were working with power feeds from several different locations as well as ground wires going in different directions. With the implementation of bonding shield wires and ground locations, an improvement in noise rejection was realized.

Power interaction problem. Another problem involved an interaction produced by the input conversion apparatus on the variable speed drives. The drives were demanding 5th, 7th, 11th, and 13th harmonic currents that were flowing through an upstream distribution transformer and creating large voltage distortion at each of the harmonic orders. The voltage distortion was affecting the processor with nonsinusoidal voltages.

We recommended the installation of a tuned harmonic filter to remove the harmonic frequencies from the power lines at the panelboard. Once the 60-Hz sine wave was restored, the controls and the power equipment operated well.

In the next three installments, we'll discuss the following groups of power conditioning devices: Voltage regulators and stabilizers; stored energy and long-term power protectors; and finally, the newer additions to the power delivery/conditioning field.