Written By: Charles F. Kerchner, Jr., P.E.

Copyright 1985
Charles F. Kerchner, Jr.
All rights reserved
Last revised 19 Aug 1988

The answer in most cases in microcomputer applications is no. Sine wave output UPS systems are really only necessary for continuous on-line UPS systems and certain directly supplied AC motor driven disk drives. Since most, if not all microcomputer systems are operating the CPU and disk drives off an internal DC power supply, sine waves are not necessary for emergency backup UPS systems. Let me explain why!

There are basically three waveform types used with UPS systems for use with microcomputers. They are square wave, sine wave, and quasi-sine wave pulse width modulated (PWM) stepped rectangular wave. All three types of waveforms must be tightly controlled as to frequency and should also have some form of maximum voltage "governor" limiting device to limit the maximum output average or RMS voltage to safe levels. This is necessary to prevent overheating of the computer power supply, especially for continuous on-line UPS units. While this is not as critical for emergency standby units it is a very desirable feature.

The cheapest waveform to provide is the square wave. Next in price range come your quasi-sine wave pulse width modulated stepped rectangular waves. And finally, you get to the higher priced sine wave units. Sine wave units use the same principles as square and stepped waveform units but they add an additional filtering device or transformer on the output to convert the waveform to an approximate sine wave.

Some people (those who sell only sine wave units) disparage all other type waveforms with scare tactics similar to some people I have been exposed to who sell fire alarm units door-to-door. This type waveform selected really depends on factors such as what type load will it be used with, is it a continuous on-line unit or an off-line emergency standby unit, and how much am I willing to pay to protect my system from crashing.

The backup requirements for modern microcomputer power supplies, which in turn supply DC voltage to power the CPU and the floppy or hard disk drives, is a lot different than the backup requirements for a main-frame computer or a disk drive running off a synchronous AC motor. Most, if not all, micro computer disk drive motors are DC driven and use phase locked loop (PLL) technology to maintain frequency and speed control and therefore do not need sine waves. Also the requirements are a lot different for waveform shape and tolerances if you are going to run your system off the UPS continuously on- line for 8-24 hours per day as opposed 2-10 minutes in an emergency condition to prevent a system crash due to momentary or temporary power failure.

As everyone knows the power generated and supplied by your local utility is a sine wave. This is because it is generated by rotating AC machinery and sine waves are a natural product of rotating AC machinery. Just because sine wave AC is provided by your utility at your outlet does not make it the only nor the best waveform to use to backup your computer. There are other factors to consider as outlined previously. In fact, for computer power supplies most engineers would tell you it would be better if smooth DC came out of the wall outlet instead of AC sine waves. Sine waves are great for power companies to make and transmit power over great distances but DC runs modern microcomputers. Interestingly enough it turns out that square waves, and quasi-sine wave pulse width modulated stepped rectangular waveforms, make better sources for rectification into smoother more ripple free DC voltage than do sine waves. The reason is that these "flat-topped" waveforms as I call them have a higher average output voltage value and the output voltage is at peak value longer than for "round-topped" sine waves. All engineers know that the charging of a DC power supply occurs at the peak of the waveform. Thus, since flat-topped waveforms are at the peak longer they keep the DC supply input fully charged longer and thus the DC output is smoother. This reduces ripple and improves the system power factor. This can be easily demonstrated by attaching an oscilloscope on the output of a DC power supply and observing the ripple with a sine wave input and then a square wave or stepped waveform input, all of equal RMS value. The DC is smoother with the flat-topped waveforms than for round-topped sine waves.

Other people have disparaged flat-topped waveforms saying they run off frequency and cause overheating. As mentioned earlier either type waveform can be off frequency and thus cause overheating. Frequency control is very important and is a separate unrelated parameter and has nothing to do with the waveform shape. All good units have tight frequency control (within 0.5%) regardless of waveform. So this argument is just a red-herring.

Another criticism is lack of control of the RMS output voltage. This is an important parameter. Low cost square wave and sine wave units both have unregulated outputs. They run wide open with the output value dependent on the level of the inverter battery. This problem is solved with pulse width modulation in flat-topped waveforms and with voltage regulating transformers with sine wave units. So blanket criticism concerning output voltage control based on waveform shape is not valid. This is another red-herring.

Another argument heard is that concerning harmonics and audible noise. It is true that flat topped waveforms make more harmonics and audible noise due to the fast rise time of the waveforms. However, good units use high frequency EMI filters on the input and output to remove any potential interference. The higher audible sound may be objectionable with continuous on-line units running all day long but it has no real effect in emergency applications using standby UPSs on modern microcomputer systems. The audible sound is most likely coming from the computer's internal AC driven cooling fan, not the computer's DC power supply. This will not hurt the fan either for short term emergency use.

An advantage often overlooked by critics in addition to the fact that flat- topped waveforms make better DC and improve the system power factor, is that flat-topped waveforms are more easily and quickly created. To create the sine wave one must first create a flat-topped wave then convert it to a sine wave. Thus, the transfer time on flat-topped waveform units can be faster since it takes longer to create a stable sine wave than a flat-topped waveform. And transfer time is a critical parameter for standby units especially for modern microcomputers which unfortunately do not have much reserve capacity or coasting time built into them. Thus, the transfer must be done as quickly as possible. Another plus for flat-topped units. Of course, if you have an on-line unit, transfer time is irrelevant. And since most on-line units are sine wave units so they can run AC motors, fans (with no hum), etc. in addition to computer power supplies, people who manufacture these would like to disparage flat-topped waveforms so you will buy the much more expensive sine wave units and since they transfer more slowly they'll then talk you into moving up to a continuous on-line unit. Hold on to your wallet!

So in summary, which waveform is better for you depends on what you want to use it for and whether it is for continuous or emergency use. Keep the above discussion in mind and you will not be talked into an expensive device you don't really need. You may find the quasi-sine wave unit is just fine for your application.

Copyright ©1985
Charles F. Kerchner, Jr.
All Rights Reserved
Last Revised - 19 Aug 1988
Return to URL: