There are critical questions that must be asked before a specification can be put together for conditioning equipment purchase.

How does power conditioning fit into our power quality picture? Well, in order for these devices to work well, we certainly need to know how they "fit in" with our facility's condition. In other words, we need to know the way in which our load devices draw current, what signal system protection is needed for proper operation, and what the conditioning equipment will do to the rest of our electrical distribution. This understanding will help us prepare specs for any required power conditioning apparatus.

Armed with this knowledge and capability, we can be certain that our proposed solution will actually do the job intended. And, we should be able to perform an acceptance test verifying this.

Remembering past lessons

Before we look at the critical parts of specifying and selecting equipment, let's remind ourselves of some of the lessons we've learned about the behavior of sensitive electrical/electronic loads and expected interactions from different combinations.

Any power apparatus placed into a circuit between the utility' source and the load device becomes a "soft" or small source. In other words, it takes the place of the larger, stiffer utility source. This means that we have decreased the stiffness and lowered the capacity of the original system while, at the same time, increased the impedance in the direction of the new "source." This doesn't depend on the type of power product used.

This leads to interaction as the currents of the load device now flow into the higher impedance source, possibly creating problems with the very device intended to "solve" the problems.

One way to avoid this impedance interaction is to use the low impedance, two winding, isolation transformer (approximately 4 % to 4.5 % impedance). Now, the source looks toward the load and doesn't "see" a high impedance; the same is true for the load looking upstream toward the smaller source. This application works well with line conditioners, regulators, motor-generators, and UPSs.

This type of transformer, having an electrostatic shield and used with secondary dedicated wiring, helps turn electrical noise back into the building, keeping it from interfering with the load's sensitive digital logic signal circuit. To do so, we place the power device in the primary and the sensitive load circuits in the secondary.

Before we specify this transformer, we first should be certain of what disturbances we are trying to prevent. By using a shielded transformer, we should be able to mitigate the "mystery" disturbances, or those that come and go at random times and circulate in the ground system.

Remaining should be line-to-line problems. Here, a voltage stabilizing, or "buffering" product, such as a motor-generator or even a stored energy device such as a UPS, may do the job.

In the Federal Information Processing Standards Publication (FIPS Pub 94), a section on electrical disturbances warns us of stability problems in series-connected voltage regulators. Looking at Fig. 1, we have an external voltage regulator's output (that of Regulator No. 1) connected to an internal voltage regulator's input (that of Regulator No. 2). This is not uncommon as many automatic data processing (ADP) units have their own internal voltage regulators fed from external regulators. But, there may be a stability problem.

If both regulators have similar time constants and high gain and phase shift characteristics, the resulting interaction can create a phenomenon called "oscillatory response." This phenomenon can range from damped oscillation, at a minimum, to a disastrous "flip-flop" of the wave-shape, at the most severe case.

Other spec points

Certainly voltage, current, and kVA/kW ratings are important parameters in selecting the proper device. But, there are other questions that must be answered.

* What is the voltage range required on the input of the device? Are we dealing with a modest 12% to 15% adjustment or are there much wider swings of voltage?

* Can we count on the voltage regulation in the load device to handle the narrow band of voltage while we supply the wider swings with external voltage stabilization? (Sometimes, an adjustment of the power transformer primary voltage taps can help keep incoming power closer to the center of the required range, thus eliminating the need and expense of an additional device.)

* Does the device work only on linear or inductive loads, or can we count on handling nonlinear currents?

* Will we need to oversize the product so that it can adequately serve our load? (Not too large for high energy losses; not too small such that the device goes into "current limit" if a step load or inrush is seen.)

These same questions also can be asked (and expanded upon) in specifying any power conditioning device.

There are other valuable spec characteristics we should consider.

Dynamic parameters. We should know the amounts of inrush currents or periodic step changes in the load current. One such group of devices that come to mind here are X-ray and imaging equipment.

We should also consider what levels or changes in system impedance occur when this special equipment is served.

Power factor. We must consider the total power factor (PF) of our system. After all, we should be energy conscious and a PF in the low to middle 90's is our best assurance that we're properly utilizing our transformer and wiring capacity. While adding the required capacitance, however, we must consider those loads demanding a high content of frequencies other than the fundamental (60 Hz).

The following questions should be asked here.

* What's the displacement PF?

* What's the contribution to total PF made by the distortion content and crest factor (CF) of the load current? (Available now are conditioning devices designed for a CF of 3.0, instead of the 1.414 associated with a pure, undistorted sine wave.)

* Does our conditioning device produce distortion at its input side, causing problems for our distribution system? As shown in Fig. 2, an engine generator is being asked by a UPS for a distorted wave-shape; when the power source is small, as is this case here, it can't supply the demands of the rectifier, in this case, a phase-angle fired circuit. The end result is a refusal to accept the UPS, and failed backup protection.

Conclusions

Remember to seek out high efficiency products. Don't get trapped by standard load levels of 75 % and 100% load efficiencies; remember, we don't normally operate in those ranges. You'll find that most of our applications end up running at only 40% to 65% of the nameplate rating of the conditioning device. Thus, you should ask suppliers to maximize the efficiencies in the 40%-to-65% range, and that they make the high efficiency stable across the load range.

One final word to those who must use battery-supported power conditioning services: whatever your choice, solid-state or rotary, be sure to specify at least two strings of batteries. In this way, you'll be able to avoid loss of your "lifeline" in a UPS system. In a single string, you won't be able to easily exam inc each cell periodically to take action before a single cell failure. Many UPS users report that this oversight causes a "surprise" shutdown at a time when the system is most needed.

SUGGESTED READING

EC&M Books:

Practical Guide to Quality Power for Sensitive Equipment. Practical Guide to Power Distribution Systems for Computers. For ordering information, call 800-654-6776.

Standards:

Federal Information Processing Standards Publication (FIPS) 94.

To order, write to the National Technical Services, U.S. Dept. of Commerce, Springfield, VA 22161.

RELATED ARTICLE: TERMS TO KNOW

Crest factor. The ratio of the peak value of a waveform to its rms value. The crest factor of a pure, undistorted, sinusoidal waveform will always be 1.414 (1 divided by .707). The crest factor of a complex waveform, however, maybe 1.414, or it may be some other value. For triplen waveforms, this value will vary from 2.0 to 3.0, and has been recorded as high as 4.0.

Harmonics. The distortion of the main 60 Hz current (or voltage) sine wave whereby additional currents or voltages (harmonics) that are multiplies of the fundamental 60 Hz current (or voltage) are introduced into an electrical system.

Nonlinear load. Also referred to as nonlinear impedance. Any type of electrical equipment that changes or modifies the voltage or current waveform to one that is not sinusoidal. The result is a complex waveform, consisting of a fundamental (60 Hz) component plus any harmonic components.

This type of load consist of a solid-state device that switches the 60 Hz fundamental current instantaneously at various points on the sine wave to obtain the required modified current. The portion of the current in given sine wave that is not passed through the device to the equipment is reflected back into the AC system, creating harmonics. Examples include solid-state devices such as inverters, rectifiers, electronic ballasts, PCs, UPS systems, variable speed drives, etc.

Power factor (PF). The degree to which current is out of phase with the voltage. There are various ways to determine PF, one being the ratio of the circuit or active power (watts) to the total apparent power (volt-amperes).