If you were in charge of dozens of computers that processed data for a major bank, how would you ensure their continuous operation? Probably your first job would be to determine the risks that could affect them. You might arrange for a degree of redundancy within the computer system or a means of net-working to another company’s computers in the event of an emergency. One of the biggest risks on your list would be the AC power to which your computers are connected.
According to studies by IBM and Bell Labs, referenced by American Power Conversion (APC), power quality is deteriorating. If the computers are directly connected to the AC supply, you can be relatively sure of being down for more than the five minutes per year that corresponds to 99.999% (five nines) reliability.
The causes of voltage transients on the AC power lines are many. In addition to lightning, the list includes “flashovers, network faults, switching on/off heavy plant [machinery], and current transients from old equipment such as welders. As a result, the AC line must be considered as having these transients as a routine expected and statistically predictable phenomenon.”1
And that’s just transients. Hot weather last summer resulted in record demands on the electricity grid, causing dropouts, brownouts, and blackouts. Your computers would find dropouts of more than a few cycles untenable.
A popular way to overcome unreliable AC power is to install an uninterruptible power supply (UPS). As the name implies, the output from a UPS is continuous even though the AC input may not be. In larger systems, redundant UPS modules can provide virtual certainty that no data will be lost should the system have to be shut down.
Statistically, AC line interruptions usually are short enough that your computer system will operate from the UPS’s batteries without missing a beat until the AC power is reestablished. Longer power outages may force you to shutdown the system, but the UPS guarantees that it will be a controlled and orderly shut- down. Some installations use a UPS to bridge the short time from AC line failure to when an auxiliary diesel or turbine generator is fired up, giving true independence from AC line vagaries.
Special Testing Required
To provide the peace of mind customers expect when they buy a UPS, the supply must be thoroughly tested when manufactured. Any design or component weaknesses should be found via design-verification testing and corrected before the product reaches full-scale production. In production testing, manufacturing process defects should be the main reasons for failure.
Final testing must balance the need to confirm correct supply operation while meeting high production rates at low cost. Consequently, automated control of real-life line and load conditions is preferred.
According to Peter Swartz, president of NH Research, “Our investigation of UPS testers indicated that the major weakness was inadequate loading instrumentation. Manufacturers had to choose between fixed resistors for each size of UPS or a combination of an AC-to-DC power supply connected to a programmable DC load. To remedy the situation, we developed an AC electronic load that provides programmability of both the power factor (PF) and the crest factor (CF) in addition to voltage and current.”
Herman van Eijkelenburg, vice president of product development at California Instruments, agreed. “Resistive load banks don’t actually reflect the behavior of real-world, nonlinear loads. All UPS loads these days use switching power supplies, which often exhibit high CF currents and PFs less than unity.”
DC power supply testing applies a load to the supply output and measures the supply’s performance as the load is changed. Other things like the supply input voltage, temperature, and perhaps humidity also can affect the output. But, the typical measurements include efficiency, output noise, line and load regulation, and transient response. If multiple outputs are involved, cross regulation becomes important, and input-dropout/brownout conditions also may be explored.
AC supply and UPS testing is more complicated. First, the load has become more complex because it now has frequency and phase characteristics. Secondly, in the case of the AC output of a UPS, there is a local area network (LAN) interface to consider. Also, coordination is necessary between the software and hardware in the UPS and the test system.
As an example of AC load complications, the relationship between PF and CF for the California Instruments Model 3091LD AC load is shown in Figure 1. The company’s accompanying description helps clarify the parameter interactions within the load during different operating modes:
“When operating in the constant current or the constant power mode, the load supports CF control by narrowing the conduction angle of the current waveform to match the requested CF. Thus, the peak current is increased while retaining the rms current level. While the apparent power (VA) remains constant, the true power decreases. This results in a reduced PF. Consequently, as CF is increased, the PF automatically decreases. The load further controls PF by shifting the current with respect to the input voltage (displacement PF). Both leading and lagging PF control is available. A phase shift of the current is only possible if the CF is higher than 1.414. Thus, CF and PF control ranges are coupled as shown in Figure 1.”
- Green, L., “Safe Inputs?,” EE-Evaluation Engineering, October 2000, pp. 94-98.
Published by EE-Evaluation Engineering
All contents © 2000 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.