Power Supply System Integration Part 2: Battery Backup

Utility blackouts and brownouts are now a fact of life, and industry sources cite an average of 100 anomalies in utility power per month. In essence, the cost of downtime to a telecom system provider, an Internet hosting system, or a medical process control is so high, that it's necessary to mandate a backup energy source, regardless of its price.

Employing redundancy by paralleling multiple identical units in the N+1 configuration does not preclude power loss due to the failure of the input power source. The use of batteries as a backup source of energy provides a cost-effective and reliable redundancy to the primary source, regardless if it's ac or dc. When selecting the optimal backup energy source, it's important to consider initial cost, and the total life cycle outlay — in a reality where the price of electricity is steadily increasing.

Protection of sensitive equipment by use of ac stand-alone uninterruptible power supplies (UPS) provides an easy solution; however, this solution is far from being reliable, cost-effective, or power efficient. An integrated dc UPS (DUPS) offers a substantially lower cost, more reliable, and power efficient solution.

The time it takes the UPS to provide useful power depends on its rating, the capacity of its internal (or external) batteries, and the magnitude of the load. Common backup times offered by most standard UPSs range from 5 min to 4 hr, but for the most part are 5 min to 15 min. This backup is enough for short power interruptions, or just to facilitate an orderly shutdown of the system. However, in many parts of the world, the dropout of utility is a daily occurrence lasting for an extended time. In parts of the United States blackouts last more than an hour as the utility administers a controlled blackout to cope with peak demands.

To survive extended blackouts, the ac or dc UPS battery backup solution is not adequate, requiring the use of motor generators or chemical energy sources such as fuel cells. Such a backup isn't limited in time, since refueling these larger systems is possible while operational. Because these systems are beyond the focus here, we will discuss electronic solutions to source redundancy. We can classify the various backup systems as:

Local backup of a small-scale system by an ac UPS. This will provide 5 min to 15 min backup for a stand-alone ac UPS. You can add extra backup time by connecting a larger external battery bank.

System embedded dc UPS. This is where a battery bank provides power, which directly uses or converts this dc to another dc level, as required to run the host system or the load. In this solution, no dc to line frequency ac conversion takes place, and the dc backup apparatus is part of the host system power supply rather than a stand-alone UPS. The backup time depends on the battery capacity, and extended time will mandate large battery banks. This solution requires proper planning prior to the system design, since the DUPS must be part of the host system and is not an add-on later.

Backup of large systems consuming power in the multi-kW range. This requires a large ac UPS for the initial portion of the utility drop (lasting a few minutes to a few hours), and thereafter an automatic replacement of the UPS by a motor generator (MG) or chemical energy plant. The MG may produce ac output to replace the utility or dc to charge the battery of the UPS and keep it running for the extended time. You can find both types of such large systems commonly used today.

In the majority of cases, the end user makes the decision between these systems and undertakes the cost responsibility to employ an ac UPS. This deprives them of options, since they cannot reconfigure the systems (to include other means for backup), and thus must purchase a stand-alone UPS to locate between the utility and the host system. This seems to be an easy and straightforward solution; however, the use of a stand-alone ac UPS doesn't result in an optimal solution to the utility backup problem in terms of overall efficiency or maintenance cost.

The far better solution satisfies the requirements for the uninterruptible power by DUPS within the system itself. This embedded approach requires the system architect to take responsibility for implementing the DUPS as a part of system design, or at least planning for such an option should the application require its inclusion in the future. This approach pays well at the end in a more efficient, less costly, and far more reliable system for the end user.

There's a severe incompatibility problem between an ac UPS and its loads, if the load is a switchmode power supply (SMPS). These power supplies are highly nonlinear loads, which demand current in high pulses causing output distortion of the UPS. It also affects the ability of the UPS to deliver its full rated power.

Knowledge of the ac UPS and its performance attributes will contribute to understanding the complex issues and limitations at hand.

A UPS is a stand-alone unit containing a battery, power inverter, and control circuitry, as well as alarms and communication capability. A classical UPS takes input utility (or generator) power and provides ac output voltage similar to the utility in terms of voltage and frequency. The size of the internal battery, the UPS rating, and conversion efficiency determine the backup time when the utility fails. Commonly, you will find UPS equipment rated in VA (volt-ampere), not watts, because the load may contain a reactive component. This is also a sort of specmanship game played by all the producers of UPS. The bigger the battery (in terms of ampere-hour capacity) and the smaller the load, the longer resultant backup time will be. Often manufacturers specify backup time both at half load and at full load.

Two types of UPSs are available in the consumer and industrial electronics markets: The first is standby or line-interactive, and the second is online or double-conversion.

Standby UPS

The standby UPS delivers the utility voltage as is — or after some conditioning — to its output and from there to the load via relay contacts or a solid-state switch. Once it detects a utility voltage failure (drops below some threshold), the relay transfers the load to an internal inverter, which converts the voltage of the internal battery to ac voltage of magnitude and frequency similar to the utility. The battery now starts a slow discharge and its voltage gradually drops. If the utility voltage reappears quickly, the load transfers back to it by the relay. At this time the load loses power, but prior to cut off the alarm circuitry within the UPS alerts the user by visual alarms and audio beeps starting at one every few seconds and progressing to rapid beeps as the UPS reaches cut off. The UPS delivers relay contacts or electric signals to the user — facilitating orderly shut down. These alarm signals communicate to the host system via RS-232 or RS-485 protocol, enabling automatic control over the host system's operation. Sophisticated UPSs provide a screen display (in an MS Windows environment), which appears on the host system monitor, advising the user how much time remains on battery. Some “Smart Software” is also available to program the orderly shutdown of the host system without human intervention — a useful feature if the system is unattended when power failure occurs.

The majority of standby UPSs in the market are underrated in power — intended for a short operational time on their inverter. They also deliver a quasi square-wave ac voltage to the load. This allows a substantial cost reduction, since it's much easier to convert battery voltage (12V, 24V, or 48V) to quasi square-wave than to a pure sinewave. This also contributes to making the unit smaller, lighter, and more efficient, and doesn't seem to bother some appliances — including personal computers. Because they are relatively inexpensive, portable, and simple to use, the standby UPS is most suitable for consumer applications, such as to back up computers, monitors, and printers. Most also include protection against utility transients and lightning. Certain models provide voltage regulation on an ongoing basis by some tap control on an input autotransformer. All of these features help to enhance computer protection. Fig. 1, on page 36, shows a 500VA UPS used for computer back up.

Despite these advantages, however, the standby UPS suffers from a major disadvantage. A relay makes the transfer of load from utility to the internal inverter, and creates a disruption of 5 ms to 10 ms in current flow to the load. Fortunately, most personal computers (and some electronic systems) can tolerate such a disruption without a detrimental effect to performance. Therefore, inexpensive standby UPSs ranging in price from $50 to $120 are used by millions of users of personal computers all over the world.

Standby UPSs are available in the range of 300VA to 500VA for office use, and up to 2KVA to 3KVA for workstations and large servers. Some are available with a step wave shape resembling a sinewave. Large standby UPSs employing ferroresonant topology provide emergency lighting for buildings.

Online UPS

In the online UPS, (Fig. 2, on page 39) the internal inverter operates at all times and its output feeds the host system. As a result, the inverter must work nonstop even while utility voltage exists. In addition, online UPSs usually provide sinewave voltage on their output terminals, and that obviously complicates its circuitry and adds to its volume and cost. Therefore, the price of an online UPS is three to five times that of a standby UPS with the same power rating. The internal inverter runs on the battery continuously, and the battery is charged from a high power converter (charger) within the unit while running on utility voltage. Hence “double-conversion UPS” is frequently used to describe an online UPS.

A simpler method runs the inverter on the rectified voltage of the utility — like an SMPS — instead of on the battery. This enables the charger to be much smaller, since it slowly charges the battery over 10 hr to 12 hr. When the utility drops, the battery takes over the dc rail directly or via dc-dc converter, causing the operation of the inverter to continue without interruption. The two main features of the online UPS are its sinewave output and its true uninterruptible operation. Some online UPSs contain an input or output power transformer, which provides galvanic isolation from utility — a desirable feature for safety and noise reduction. However, due to cost considerations, most low power (1kVA to 5kVA) UPS, especially those made in the Far East, have no isolation, and you use the input neutral line also as the output neutral.

Online UPSs cover the power range of 500VA to several hundred kVA. Most have an internal battery adequate for 5 min to 15 min interruption. Some high power UPSs are equipped with a separate cabinet for the battery, and provide backup time in the hours. For those requiring very long backup (12 hr to 48 hr), the battery bank becomes huge in size, compared to the UPS itself, and a generator UPS combination may be a better choice. Some online UPSs include input power factor correction and have elaborate front panel display and communication software. This allows remote monitoring of the UPS and permits gathering of useful data on ongoing basis.

Many online UPSs have a solid-state transfer-switch (SST), which enables load transfer to the utility if the UPS fails. The transfer also permits maintenance work on the UPS while temporarily feeding the load from the utility. The transfer switch enables the load to be on the utility at start, and only after the initial current surge subsides does it transfer the load to the inverter. A phase-lock-loop (PLL) circuit within the UPS ensures “seamless” transfer of the UPS by this SST between utility and inverter and vice versa. The PLL circuit “locks” the frequency and phase of the UPS to the utility at all times.

The main advantage of the UPS is being a stand-alone unit, which can be added arbitrarily only to those systems or loads that cannot tolerate a utility interruption. The end user, who must pay its cost, (about $1 per VA) makes the decision whether to use the system and chooses the type, model, and make. Also consider the outlay, otherwise the resulting UPS could be incompatible with the system's needs.

UPS Nonlinear Loads

Most end users are oblivious to the problems encountered when an online UPS interacts with nonlinear load, such as a switchmode power supply. Fig. 3, on page 42, illustrates the interface between an online UPS and a switchmode power supply. The switchmode supply within the host system consumes its current in high amplitude pulses, and as a result overloads the UPS, distorts its output and degrades in performance. This isn't the case if the SMPS has power factor correction (PFC); and in the United States most SMPS don't have PFC.

Repeated measurements under practical conditions show that in non-PFC SMPS, the current peak to rms ratio reaches a crest factor in the range of 2.5 to 3.5. This high crest factor imposes a difficult load for the UPS.

Manufacturers specify UPS capacity in volt-amperes, and set the UPS current limit with resistive loading. This affects the user in two ways: first, the actual power (in watts) that the online UPS can deliver is 75% to 80% of its VA rating. Second, the UPS cannot support a nonlinear load like a SMPS to full power capacity. In fact, it can support an SMPS only rated to 30% to 40% of its capacity. A 1kVA online UPS will only be suitable for backing up a system energized by a 300W to 400W (non-PFC) switchmode power supply. Attempts to draw higher power by the SMPS will severely distort the output of the UPS and may result in shutdown. Some UPSs support nonlinear loads to various degrees, but the above rule is safe to use as a guideline.

The whole idea of generating a high-purity sinewave within the UPS, and then rectifying it within the SMPS is wasteful. A UPS may be deployed to run non-electronic loads, such as lights or motors. However, this is a minor portion of the market for online UPS. Many applications back up computers and telecom devices, as well as instrumentation systems used for critical applications. In all of these applications, the immediate load for the UPS is the SMPS within the systems. Peak current and inrush current are associated with this load at startup. This current of 40A to 80A per SMPS will cause the UPS to shutdown due to overloading, or the UPS must transfer the load to utility to draw its inrush.

For more information on this article, CIRCLE 333 on Reader Service Card