Uninterruptible power supply

An Uninterruptible Power Supply also commonly referred to as a UPS is an electrical system that provides instant electrical power in the event of an emergency. The UPS system comes online when the primary source of power is disconnected. However, a UPS is different from an emergency power system or a power generator because it provides instantaneous protection from interruptions caused by your main power system or the grid as it is commonly referred to. The system works in supplying energy which was previously stored in large led acid or maintenance free batteries or a flywheel. That said the overall run time of a battery based UPS is short but enough for homes and offices to kick start their standby power generator or initiate a proper shutdown to protect their equipment.

An Uninterruptible power supply is most commonly used to protect electronics like telecommunications equipment, data centers and computers or any other electrical equipment which in the case of failure may cause damage like fatalities, injuries, data loss or disruption to the business. A UPS unit ranges in size from ones designed to protect one computer with a 200 VA rating or larger units like the whole of a data center or building. Today, the world’s largest UPS is a 46 megawatt, BESS or Battery Electric Storage System located in Fairbanks, AK. This mega UPS powers the whole of Fairbanks as well as rural communities during power outages.

Common Power Issues

The basic role of a UPS is to work as a substitute to the main power source when the main input source fails. That said many types of UPS systems are designed to be able to correct numerous utility power problems like:

  • Abnormal voltage or a spike in the voltage
  • A sustained or momentary reduction in the voltage
  • Correct noise, which is a high frequency transient or an oscillation which is caused by nearby electrical equipment
  • Instable mains frequency
  • Frequent harmonic distortion which is caused by a departure from the regular sinusoidal waveform

UPS systems can be placed into two distinct categories based on which problems (mentioned above) they address. Some manufacturers will categorize their products based on which types of power related problems they can fix.


There are three general categories of a modern UPS system i.e. standby, line interactive and on-line. Online UPS systems use a method called “double conversion” which accepts an AC input, and rectifies it to DC which is then passed to a rechargeable battery. When the backup system kicks in the system inverts the battery’s power back to 12V/230 VAC to continue powering equipment. A line interactive UPS system works by maintaining the inverter directly in line and will redirect the battery’s power (DC current) from normal charging mode in order to supply current in the event that the mains lose power. A standby also known as “off-line” system works by directly powering everything via the input power with the backup system being invoked when the main utility power goes down. Generally, UPS systems below 1kVA are line interactive or they are standby both of which are comparatively cheaper.

Larger units often require the use of a dynamic uninterruptible power supply. This type of UPS makes use of a motor/alternator which is connected directly to the mains through a choke. A flywheel stores all the energy and so when the mains fail the Eddy current flowing will maintain the power as long as there is energy in the flywheel. At times a DUPS will be combined with a diesel generator which comes on after a short delay, this is often labeled as a diesel rotary uninterruptible power supply.

Recently, a fuel cell based UPS has been introduced which runs on hydrogen fuel cells. The technology is still in its infancy but holds the potential to provide long backup times in a small package.

Offline / standby

An offline standby UPS usually has a protection time of up to 20 minutes. Its capacity is non expandable. This type of UPS offers very basic features usually just surge protection and backup from a lead acid battery. The UPS in this case is directly connected to a point from where utility power lines enter into a home or office. When the voltage from the utility company falls below a certain level the UPS will turn on and the internal DC –AC inverter will start working to convert DC into AC current. When the power is back to normal it switches back to its AC to DC conversion to charge the battery. The delay is usually just 25 milliseconds depending on how much time it will take for the standby UPS to detect that power has been lost or dropped. That said this type of system is designed to power certain types of equipment like personal computers without causing a drastic dip which could ruin the circuitry.

Click to see hi-res image



The typical protection time of a line interactive UPS is between 5 to 30 minutes but the capacity can be expanded to many hours. A line interactive UPS works a lot like a standby UPS but unlike a standby UPS it has a multi-tap autotransformer. This special transformer can add and subtract power from the coil which helps to increase or decrease the overall magnetic field and also the output voltage of this transformer. A multi-tap autotransformer is also called a Buck boost transformer.

Click to see hi-res image

A line interactive UPS is designed to be able to tolerate under voltage, over voltage and brownouts without having to consume the limited battery power it is hooked on to. The compensation system works by automatically choosing different taps on the built in transformer. So, depending how the system has been designed it will change the autotransformer tap which causes a very short power output disruption which may cause the UPS alarm to go off for a while (if it has a power loss alarm).

Today even the cheapest UPSs have this built into them since it can take advantage of the components which are already part of the system. The transformer which converts between battery and line voltage should have two different winding ratios: the first ratio will convert the battery output into the line voltage and the second will convert the line into a slightly higher voltage than the battery i.e. around 14V to charge the battery. That said it is much easier to do all the switching directly on the line voltage end of a transformer because it has a lower current.

In order to gain from the buck/boost feature, you just need two different switches so that AC input can be connected with either one of the two primary taps of the transformer, while your load is connected over to the other which leaves the primary winding to work as an autotransformer. A battery can still charge when in “Bucking” overvoltage but when it is boosting some under voltage the battery will not charge because the current in the transformer is very low.

The beauty of an autotransformer is that they can easily be engineered to cover an array of input voltages but this may require more taps and so the complexity will increase as well as raise the price of the UPS. That said it is common for many autotransformers to just cover a range of between 90V to 140V prior to switching over to battery power if the voltage goes too high or low.

If the voltage is low a UPS will use up more current than usual so it will require a high current circuit as opposed to a normal device. For instance, a 1000 watt device which is at 120 volts, a UPS will easily draw 8.33 amperes. If there is a brown out causing the voltage to drop to 100 volts, the ups will draw around 10 amperes in order to compensate. This very same principle will work in reverse, in the case of an overvoltage condition a UPS will require far less current.

Double conversion / online

Online UPS systems are great for environments where there should be electrical isolation so that sensitive equipment is not affected by power fluctuations. Even though this type of UPS was once only reserved for large installations i.e. over 10 kW, numerous advances in technology now allow for it to be available to the average consumer requiring 500 watts or so. However, the initial cost of this type of UPS is higher but the overall cost of ownership is lower owing to much better battery life. An online UPS is necessary in a noisy power environment, when outages and anomalies are very frequent and so protecting sensitive equipment is needed or when you need to get power from an extended run generator.

While the basic technology of an online UPS is very similar to a standby or a line interactive it still costs a lot more. This is mainly because it has a much larger AC to DC charger / rectifier which has been designed to run continuously with a cooling system. Often referred to as a double conversion UPS owing to its direct inverter that runs even when it is being powered by normal AC electricity.

Batteries are always connected to the inverter in an online UPS and so having switches that transfer power is important. When there is power loss the rectifier will drop from the circuit and everything will be powered by the battery. When power from the grid returns the rectifier resumes taking on the load and charging the batteries. That said the charging current is often limited in order to prevent the high power rectifier from heating the batteries which causes the electrolyte to boil away.

The big advantage of this type of UPS is that it provides a sort of electrical firewall that separates the utility company’s power and your sensitive equipment.

Other UPS designs

Double conversion on demand

Often also referred to as hybrid designs they don’t have any official name other than double conversion on demand, a term used by Eaton and HP. This type of UPS is designed for high efficiency applications that also require the same level of protection as offered by a standard double conversion UPSs.

This hybrid as we like to call it operates as a standby UPS when the utility power is within a certain predefined range. This enables the UPS to output high efficiency ratings. When the utility power goes beyond or under the predefined range the UPS will automatically switch to double conversion operation. When in this mode the UPS is able to adjust according to voltage variations without needing to use a battery. It can filter line noise and also control frequency. Some latest examples of this type of UPS is the HP R12000, HP R8000, and the new Eaton Blade UPS.


A Ferro resonant UPS operates a lot like a standby UPS. The major difference here is that they are online and that a especially designed ferro-resonant transformer filters the output. The beauty of this transformer is that it can hold energy for a long enough time so that the UPS can switch from existing line power to the backup battery. This helps to eliminate any transfer time. The majority of these types of UPSs are around 88% efficient but offer great isolation.

The transformer of a ferro resonant UPS has three main windings, one is used for the mains power, the other for rectified battery power and the third is used for AC power output.

There was a time when this was the prevailing type of UPS but is limited to around a 150kVA range. Probably, this is why these units are still mainly used in an industrial environment. That said these are great UPS types because they don’t interact directly with power correcting tools.

DC power

An uninterruptible power supply which has been designed to power DC equipment is a lot like an online UPS except that an inverter is not required. However, if the voltage of the UPS’s battery is matched exactly with what the device requires a power supply for the device will not be required. Since, a few power conversion steps are removed efficiency as well as run time improves.

The majority of systems used in the telecommunications industry utilize 48 VDC of power mainly because of fewer safety regulations like having junction boxes and a conduit installed. That said DC has been the dominant power source in the telecommunications industry for a long time while AC has been used in computers as well as servers.

Various experiments with a 48 VDC power source for servers have been performed over the years with the hopes of it being able to reduce the overall likelihood of failure and the resulting cost of equipment. But in order to supply the same level of power a higher current will be required, perhaps in the range of between 115V and 230 V, which requires larger conductors and lots of energy is lost because of heat being given off.

A fair majority of PCs can easily be powered with 325 VDC, because ATX switching power supplies work by converting AC input into around 325 VDC. A single unit with a voltage selector switch, as well as the 115V setting allows for the voltage doubler to put the first half of the AC wave into a capacitor, and the following half into another. In this mode only half of the bridge rectifier is used and runs 2x current through it. A 230V setting just rectifies your AC using a full bridge rectifier and also puts it into both of the capacitors. Both capacitors are wired in series. This makes the power supply safe to run on around 340 VDC but the selector will need to be set in the 230V position. This will not work when DC power is switched in the 115V position with around 162 VDC applied. This is because only one capacitor is charged. That said if 325 V is put, the fuse will blow in addition to the surge suppressor. Most Active PFC power suppliers are auto ranging and don’t have a voltage selector switch. These usually just have one capacitor for input which is charged to around 400 VDC with the aid of a boost mode supply which is built as part of the circuit. That said it is very uncertain how active PFC and auto ranging power supplies may respond to using DC power when they all expect AC power 50-60 HZ. One classic example of this is a laptop computer with a DC UPS that is built in.


Rotary UPSs use the inertia of a spinning flywheel in order to provide a short term back up ride through when the utility power is off. In addition, the flywheel also works as a buffer against power issues like sags and spikes, because these short glitches in the power do not affect the overall rotational speed of the flywheel. That said this is probably one of the oldest UPS designs which predates integrated circuits and vacuum tubes.

A rotary UPS can be labeled as an on line UPS because the flywheel spins when power is flowing through it. But unlike a regular battery backup type UPS, a flywheel based one can only provide protection for around 20 seconds max prior to the flywheel slowing down and halting the power. This is why this type of UPS is often used in conjunction with a diesel generator, which will turn on within these 20 seconds.

A rotary UPS is usually used for applications that require over 10,000 watts worth of protection because that is the only way to justify its expense and also benefit from the advantage of this system. Many times a large flywheel or probably a few wheels increases the backup time and the capacity of the system.

Since, flywheels are a source of mechanical power, using an electric motor is not necessary it may not even need a generator that will work as an intermediary between it and a regular diesel engine generator in order to provide uninterruptible power. That said a transmission gearbox is used, which helps to directly start the generator via the inertia of the flywheel, once the generator starts it will continue to spin the flywheel. A number of flywheels can be hooked up in parallel via mechanical countershafts which does away with the need of a separate motor and generator for each and every flywheel.

Rotary UPSs are designed to provide a high current output as compared to their electronic counterparts. They are also able to provide a better inrush current for various types of inductive loads like a motor startup or a compressor load and even medical MRI equipment. These systems can tolerate various short circuit scenarios which are 17 times larger than what an electronic UPS can handle which in most cases will fail after the fuse blows.

The life cycle of a rotary UPS is a lot greater than an electronic UPS, the former providing a life cycle of over 30 years. That said in order to properly maintain a rotary UPS periodic downtimes are required. Things like ball bearings need to be replaced every few months. Battery based UPSs do not require any downtime because the batteries can easily be hot-swapped. The latest rotary units use various techniques like magnetic bearings as well as air-evacuated enclosures which increase its standby efficiency and reduces its need for maintenance to a great extent.

Usually, a high mass or heavy flywheel will be used in conjunction with a large motor generator. There are a number of common configurations:

  • A motor that directly drives a generator connected mechanically
  • A synchronous motor and a generator which are wound in different slots of one rotor and a stator
  • A hybrid rotary system which is designed like an online UPS but with an added flywheel instead of a battery. In this system a rectifier will drive the motor which will spin the flywheel, while the generator uses this flywheel to turn on the inverter.

In the case of the third point (above), the motor generator can either be synchronous/ induction or synchronous / synchronous. In the case of point no. 2 and 3 the motor side can be driven directly by an AC source of power (usually when the inverter has been bypassed), probably a 6 pulse inverter or a 6 step double conversion type motor drive. In point no. 1 the system uses an integrated flywheel which provides a short term burst of energy which does away with the need for batteries, allowing for enough time for an external or a coupled generator to start and come online. In point 2 and 3 batteries or a free sanding flywheel is used as a source of short term energy.



In a large business environment reliability is of the greatest importance, here having just one huge UPS can also mean a single point of failure which will disrupt everything. That said in order to provided added reliability, a number of smaller UPSs and batteries can be integrated to provide a redundant power protection system which is equivalent to a large UPS. The “N+1” really means that if the load can be supplied by an “N” module, the installation will have an “N+1” module. This means that one module failing will not have a big impact on the operation of the system.

Multiple Redundancy

There are many computer servers that will offer the option of a redundant power supply, which in the event of a power failure either one or more of these power supplies will power the load. In the computer industry this is a critical point because every power supply should to be able to power the whole server on its own. The Redundancy is then further enhanced by adding each power supply into a fresh circuit (another circuit breaker). This redundant protection can be further extended by adding each power supply to a UPS of its own. This will provide a solid double protection from both the failure of the power supply and the UPS which ensures continued operation. Many times this is referred to as a 1+1 or a 2N redundancy. If your budget does not permit you to purchase two UPS units then one power supply can be plugged into the mains power while the other goes into the UPS.

Outdoor use

A UPS system which is designed to be placed outdoors needs to have specific features that will help it withstand the weather without affecting performance. Humidity, snow, rain and storms are just a few natural elements that the system needs to withstand, something that the manufacturer has to plan for. Outdoor operating temperatures will usually range from -40 to around + 55 °C in the summer.

An external or outdoor UPS can either be host, ground or pole mounted. When placed outdoors the UPS needs to be able to withstand extreme cold which in some cases means that it should include a battery heater, or some other form of heating. In addition it also needs to have a cooling system like a fan or air conditioning system.

Internal Systems

A UPS system can easily be designed so that it can fit within the chassis of a computer. There are two main types of internal UPSs i.e. a regular UPS which has been miniaturized enough to fit in a 5.25 inch drive bay. The other is a re-engineered power supply which will use dual AC or DC power as its inputs with a built in control unit which switches between the two.

Machine Standards

Measuring Efficiency

Efficiency of a UPS is measured differently by every manufacturer and there are a few reasons for this too. Many manufacturers will claim that they have the highest efficiency, often using a table of figures and criteria to prove their figures. That said the industry norm as agreed by many experts is between 93% and 96% when a UPS is operating optimally. So, in order to get these figures a UPS manufacturer will test their UPS in an ideal scenario. However, on site figures are closer to 90% because of fluctuating power conditions. This is because we will never have a perfect scenario owing to voltage sags and fluctuations, plus the declining efficiency of a UPS battery.

Uninterruptible power supplies are often backed by a warranty which has varied quite a bit over the years. The warranty usually depends on if the unit is a single phase or a three phase one. However, many companies don’t compete on warranty, because their focus is on maintenance contracts and efficiency. Standard warranty is between 1 and 2 years, and could also be limited to certain parts of the machine. This could exclude more expensive parts like batteries. That said companies selling a 3 phase UPSs offer a 2 year warranty if not more.

Difficulties with generator use

Variations in Frequency

The overall frequency and the voltage which is produced by the generator will depend on its rotational speed. This speed is regulated by a ‘governor’. ‘Governors’ are either mechanical or electronic. The job regardless of its type is to keep the frequency and voltage constant, while the load changes. The governor does not instantly respond to changes in load which can pose a problem if for instance the surge from an elevator startup causes the frequency to drift for a few seconds while the governor adjusts to the latest load on it, this will affect all the other devices connected to the generator. Numerous radio transmission sites will have a backup generator, which in case of AM radio transmitters the load will change constantly and instantly with the level of the signal. So, here the governor needs to constantly try to correct the voltage and its frequency. A UPS can be incompatible with a diesel generator or even a bad mains supply if it has been designed to only run on exactly 50 Hz. The UPS can remain on battery and unwilling to connect to the input which is beyond what it’s programmed to handle.

This problem of an input frequency does not seem to be an issue with a Double Conversion / online UPS. This UPS should adapt to almost any frequency thanks to it’s built in clock source which generates the required 60 Hz or 50 HZ frequency.

Power Factor

The biggest problem with combining of a double conversion UPS with a generator is the overall voltage distortion mainly created by the UPS. A big rectifier works as an input for the double conversion UPS, and so the current drawn by the UPS is mainly non-sinusoidal. This often causes the voltage taken from the mains or a generator to also become non-sinusoidal. This type of voltage can cause problems with almost all types of electrical equipment, which includes the UPS too. Also, a lot more power will be lost within the wiring supplying the UPS because of spikes in the current. This type of noise is often measured as “Total Harmonic Distortion of the current” aka THD(i). Many classic UPS rectifiers have around 25%-30% level of THD(i). in order to reduce this distortion heavy mains wiring will be needed which is twice the size of the UPS!

There are a number of solutions which could reduce THD(i) in a typical double conversion UPS:

Passive correction: This is a classic solution which uses passive filters to reduce overall THD(i) to 10% when at full load. These tend to be reliable but they are big and will work only when there is full load. But these tend to present their own unique set of problems when used with generators.

Active power factor correction: This is mainly an air filter. By using a device like this the THD(i) will drop to around 5% when in full range of power. A rectifier is the latest technology used in a double conversion UPS but this rectifier does not consist of diodes and thyristors. In a double conversion UPS that has an IGBT rectifier as well as an inductor the THD(i) is around 2%. This eliminates having a huge generator and transformers. It helps to save space, cost and losses.


PM or Power Management requires the following:

  • A UPS needs to report its current status to a computer that it powers through a communications link i.e. GPRS, USP, Ethernet, Serial port, Firewire etc
  • There needs to be a subsystem built into the OS that generates reports and sends notifications. It needs to also send power management commands. While some UPS manufacturers make their communications protocols public others use proprietary protocols.

The most basic control methods are mostly designed for one to one messaging i.e. from one source to another. For instance a UPS may connect to one computer to provide a status report about its functioning, and may allow the computer to control the system. In the same way the USB protocol is intended to connect a computer to numerous devices.

There are situations where having a single big UPS communicate with many protected devices is necessary. For the traditional USB or serial control, it may use a signal replication system which for instance may allow one UPS to easily connect with five other computers using either USB or serial connections. However, this splitting is just in one direction i.e. from the UPS over to the devices in order to provide status info. Returning control signals may only be allowed from one system to the UPS.

Use of an Ethernet has certainly become a common place since the early 90s, today it is common for control signals to be set between multiple computers all of which use a standard Ethernet connection via TCP/IP. The control information and status is usually encrypted so that a hacker cannot gain access to the UPS and send a command to shut it down.

The overall distribution of the UPS’s status and all the control data will require that all intermediary devices like Ethernet switches or even their multiplexers be powered by either one or more of the UPS systems so that the UPS alert can reach a target system when the power goes down. In order to avoid total dependency on Ethernet, the UPS can be directly connected to the main control server via GSM or GPRS. An SMS or GPRS packet data can then trigger the software of the UPS asking it to shutdown the PC in order to reduce load.


The size of a battery connected to a UPS is what governs its overall backup time. Most manufacturers will provide run times in minutes usually printed on the packaging. Large systems like ones used to power data centers require more detailed calculations of inverter efficiency, load, and the characteristic of the battery for max endurance.

Common characteristics of a battery and testing loads

Each time a lead acid battery is discharged or charged it will affect the chemicals reacting inside of the battery i.e. the electrolyte and electrodes. Over time, the charge that the battery is able to hold at its interface also known as “interface charge” which will spread by diffusion of the chemicals inside the battery throughout the overall volume of the chemical material.

Once the battery has been discharged completely it can be recharged using a fast charge method which takes a few minutes, but the effect of this is that during its long charge cycle it will develop a charge only near the interface. The voltage of the battery can rise in order to be close to what the charging voltage is so the charging current will decrease. After a couple of hours the interface charge spreads to the entire volume of the electrode and the electrolyte, which lowers the charge to the extent that it is insufficient to provide adequate backup.

Because of this interface charge short UPS self test functions that last a couple of seconds is not able to accurately reflect the runtime capacity of a UPS, but rather a recalibration or a rundown test may be required.

It is important to note that deep discharging to test the battery can be damaging to it because the chemicals in a discharged battery can start to crystallize into more stable shapes which means that they will not dissolve when the battery needs to recharge this reduces the capacity permanently. Often referred to as sulfation in lead acid batteries it also affects nickel cadmium and lithium batteries. This is why the rundown test should only be performed every six months or every year.

Testing strings of batteries

Huge multi-kilowatt uninterruptible power supply systems that have easily accessible but large battery banks are able to isolate and test individual cells within every battery string. This string can consist of either combined batteries (like 12V lead acid cells) or chemical cells which have been wired in series. A single battery can be discharge tested by isolating and then installing a jumper in place, while the other batteries are charged and still able to provide backup.

You can also measure the chemical and electrical characteristics of each cell in a string, using an intermediate sensor which is installed on each cell to cell junction and are monitored both collectively and individually. Battery strings can also be wired in parallel and series for instance you have two sets each with 10 cells. However, when wired in such a manner it is important to monitor the flow of current between the parallel strings, as the current may circulate between strings in order to balance the effects of a week cell or a dead cell having high resistance. For instance, a stronger string can easily be discharged through weaker strings until the voltage imbalance is equalized.

Parallel and series battery interactions

Battery strings which have been wired in series and parallel may develop strange failure modes because of multiple parallel strings interacting with each other. The most common of these failures stem from defective batteries in a string which can affect the operation and the lifespan of new and good batteries in the other strings. This same issue applies to instances where a series –parallel strings are being used, not only in UPS systems but also in electric vehicles.

When you have a series-parallel battery arrangement with good cells and any one cell dies or is shorted:

  • The cell which has failed will reduce the overall maximum voltage for the whole series it is strung with.
  • All the other cells in the series wired in parallel will now start degrading by discharging via the degraded string until their present voltage is the same as the degraded one. This could overcharge the dead cell leading to the electrolyte boiling out or gassing from all the good cells in the degraded cell’s string. That said now the parallel strings cannot ever be fully recharged as the voltage will simply bleed off via the string that has a bad cell.
  • Some charging systems may try to gauge the capacity of the battery string by measuring its overall voltage. But because of the depletion, the charging system may detect an abnormality as a state of discharge, and will try to charge the strings leading to overcharging and extensive damage to all the cells which are a part of the string.
  • If a lead acid battery is used all the cells in the bad battery parallel string will start to sulfate owing to their inability to charge them fully. This will result in the storage capacity of these cells being damaged for good.

By far the only way to prevent this from happening in a series-parallel string is by not using a parallel string to begin with. Use separate charge controllers and also inverters for each string.

Old and new battery in a series – interactions

A single string of batteries which has just one old battery in it can cause a number of adverse interactions. Older batteries have a reduced storage capacity and so they discharge faster than new batteries, they also charge faster than new batteries.

In a mixed string of old and new batteries the voltage in the string will drop when the old batteries deplete while the new batteries may have charge. The new cells will continue discharging via the string but because of low voltage the energy will not be useful and often wasted on the depleted cells.

For batteries that have to operate within a certain discharge window, some new cells with larger capacity will cause all the old cells to discharge beyond their safe bottom limit which further damages the old cells.

Old cells will recharge more quickly, which leads to a quick rise in the voltage near to the full charge state. But it reaches this state prior to the new cells being fully charged. The charge controller will easily detect a rise in voltage of the nearly fully charged batteries and reduce the flow of current. Now the new cells which have larger capacity will charge at a slow pace, as a matter of fact it will be so slow that the chemicals may crystallize prior to being fully charged. This reduces the capacity of the new cells until it matches the old cells.

This is why some large industrial UPS systems require a periodic replacement of the whole battery array that consists of many expensive batteries mainly because of these damaging interactions.