Power-Line Conditioners

In this section, we will introduce another device that is used to solve power problems—the transformer. But before we get deeply involved in a discussion of transformers, we must take yet another look at terms we are using.

The term, power-line conditioner, is a term that might be applied to most any product that is connected to the electrical line. Filters, surge suppressors, UPS, and voltage regulators are all power conditioners of one type or another. We can further define conditioning as "changing" or "modifying" something.

On the other hand, there are products that do condition power all the time but perform other enhanced functions as well. Voltage regulators, e.g., not only condition power, but they also adjust for variations in voltage, while UPS products provide power when the utility supply fails.

In this section, we will be talking about products that only condition power and that always condition power. This may seem a bit confusing. Why are we going to this extreme to explain this fairly simple point? Simply because the uncomplicated, bare-bones, no-frills power conditioner is the single wisest investment that the average user can make. it's a far better product than a surge suppressor, although it costs more. and it will solve the vast majority of all the power problems that we will ever encounter, far less expensively than many of the other products we will discuss.

And, yet, the simple power conditioner is probably the least popular of all power products on the market. Why? Because what they do is not simple to explain. They don’t just suppress spikes. They don’t regulate voltage. They don’t provide backup power. So what do they do? They condition power, and that doesn’t paint a dramatic mental picture for marketing executives to sell to the public. In fact, some executives have changed their terminology and now use the word “filter” to better describe the product’s function. But “filter” implies something that can act only on an engineered band of noise frequencies. Power conditioners act to eliminate all noise frequencies, and that’s the beauty of the product. Before we can explore these seemingly mystical qualities, we need to understand the heart of the technology involved.

Isolation Transformers

Transformer. Now there is a word we have heard a time or two. We know from Section 2 that transformers are amazingly efficient, around 96 - 97% efficient. We also know that “power in” equals “power out.”

We made a fleeting comment in Section 6 about a transformer’s propensity to pass normal-mode noise. But if a transformer will pass noise, why call it an isolation transformer? Remember that in Section 2, we discussed the construction of a transformer as being one coil of wire inside another, both wound jelly-roll style.

These two coils, the primary and the secondary transformer windings, are physically separated from one another by air and other insulating material. This type of construction inherently isolates the primary from the secondary. So, in that sense, all of these type transformers are isolation transformers. But we are talking about noise isolation. This type of isolation calls for some special design considerations.

Capacitive Coupling

Because of the proximity of the primary and secondary windings, a trans former will display a characteristic called interwinding capacitance. This is the effect of the facings of the windings displaying capacitive-like qualities at certain noise frequencies. This stray capacitance can be illustrated as one capacitor that bridges the distance between the primary and secondary windings (ill. 9-1).

ill. 9-1. Stray, interwinding capacitance exists inside a transformer. (A) Multiple stray capacitance. (B) Stray capacitance simplified to one value of capacitance.

High-frequency noise can, therefore, use this path to travel from the primary to the secondary of the transformer. In order to prevent this, or at least attenuate it, we can significantly reduce the interwinding capacitance by installing a Faraday electrostatic shield between the primary and secondary windings (ill. 9-2).

ill. 9-2. The addition of a Faraday electrostatic shield greatly reduces the interwinding capacitance.

Wait! That doesn’t seem right. The shield appears to produce twice as much capacitance. After all, our diagram shows twice as many capacitors. The answer to this contradiction is found in the mathematics of electricity. It would seem logical to assume that the total capacitance would equal the total of each individual capacitor being added to the next, or

C_T = C₁ + C₂ + …

In fact, that is the formula for finding the total capacitance of capacitors in parallel. The formula for finding the total capacitance of capacitors in series is

C_T = 1/C₁ + 1/C₂ + …

In other words, we are using the reciprocal of each individual value. Now it's apparent why we can state that the addition of a shield greatly reduces capacitive coupling.

Well, if one is good, might not two be better? In fact, this is usually the case. A second shield tied to the primary winding acts as a low-impedance path for common-mode noise to travel from one shield to the next. This path diverts common-mode noise away from the winding, thus eliminating common-to normal-mode conversion by the transformer. Double-shielded isolation transformers are a popular product in the three-phase power world. Large products can accommodate the additional material and associated costs much better than a product which is oriented toward PCs. In fact, some of the transformers for PCs have no shield at all. While this holds the top-end normal-mode attenuation to around -20 to -40 dB, they do have some thing that helps just as well, which we will discuss shortly. Certainly, a double-shielded isolation transformer would be a rare find in the PC world.

Let’s take this one step further. What about triple shielding? Again, this is something a specifying engineer might call for in the plans for a large three-phase installation. it's never found in PC products. and there is considerable controversy as to whether three shields actually help reduce noise coupling much more than double shielding. Most manufacturers of large, computer-grade transformers don't offer a triple-shielded device, except on a custom basis.

The shield, itself, is most often a solid copper sheet—a little thicker than aluminum foil—that is wound, jelly-roll style, between the primary and secondary of the transformer. Notice in ill. 9-2 that we show the shield grounded. This has the effect of completing a noise-current path for any noise originating on the primary side of the transformer (ill. 9-3). This keeps much of the noise away from the secondary and the sensitive load. Not everything is perfect, and it's possible for some noise to transfer to the secondary even with the addition of a shield. We will deal with this problem later.

Common-Mode Noise

A glance at ill. 9-3 shows that the noise the shield is designed to deal with is mainly in the common mode. Any time that ground is a path for noise, we are talking about common-mode noise. The addition of a shield will raise the noise attenuation characteristic of a transformer to about -60 dB. This means that a 1000-volt impulse will appear as 1 volt on the secondary. There are manufacturers who make claims of much higher noise attenuation. and , in fact, even a standard shielded transformer will display a wide range of attenuation, depending on the frequency of the signal applied. This device is better at lower frequencies than at higher. Overall, -60 dB is a good average and more than adequate for the PC environment. A company that is making wild claims about noise attenuation, over -100 dB, is making a dubious and almost unprovable claim.

ill. 9-3. The addition of a shield produces a loop for the flow of noise current, without its being transferred to the secondary.

There is another valuable feature of transformers that will totally eliminate common-mode noise as a concern. it's a National Electric Code safety requirement that one leg of an isolation transformer’s secondary be grounded and identified as a neutral. ill. 9-4 shows how this is accomplished. This is the neutral-to-ground bond we have been talking about all through this guide. A power conditioner allows the ground to be reestablished close to the computer. Now, neutral is solidly referenced to ground through the impedance of the bonding strap and not through any stray capacitance that may exist in the load. Any common-mode voltage that might appear from either side of the transformer has its current short- circuited to ground through the bond. and , of course, any voltage developing across the bond will be extremely low.

ill. 9-4. The neutral-to-ground bonding point is established on the secondary winding, thus eliminating common-mode voltage drops.

ill. 9-5 shows before and after photographs of common-mode noise. Obviously, there exists significant noise on the line, but the addition of the power conditioner has eliminated it. As we have stated, common-mode noise is the most serious problem that PCs must deal with. Tying the neutral and ground together near the computer is the answer.

ill. 9-5. The addition of a power conditioner with the neutral and ground bonded together virtually eliminates common-mode noise. (A) Before conditioning. (B) After conditioning.

Before we leave this subject, let’s take another look at ill. 9-4. We see that the neutral and ground are tied together. We also see that the shield and core of the transformer are tied together. This is single-point grounding. Another ground element that we don't see in the diagram is the chassis of the power conditioner itself. it's clear that all chassis are tied to ground. But we bring this up to emphasize what the designer of such a device is trying to do. Every effort is being made to hold ground and anything referenced to it to the lowest possible impedance with respect to one another. This eliminates stray paths through the case of the unit, through the load, or differentials between the primary and secondary windings. We have established a clean, single point of ground reference for anything connected to this device.

There is another advantage to this design setup. Often, printers and other computer peripherals are a considerable source of common-mode noise or ground-loop voltages. This grounding of the neutral line is a direct path for those signals, thereby keeping them from propagating to other loads.

Normal-Mode Noise

Normal-mode noise is noise that appears from the line to neutral. The 120- volt supply voltage is a normal-mode signal. it's transferred from the primary windings to the secondary by currents induced by the changing magnetic field. It stands to reason that normal-mode noise would behave in much the same way.

An increase in any current flowing in the primary windings will result in an increased current flow in the secondary windings, regardless of the frequency of the applied voltage. The transformer principle applies to all frequencies, not just 60 Hz. We must realize that the ability of a transformer to pass normal-mode signals does vary as the frequency of the applied signal varies. After all, a supply transformer is engineered to pass 60-Hz power with efficiencies of about 97%.

The efficient transfer of energy at other frequencies will be different. and this frequency response will not be uniform as the frequency increases. Because of this, we can expect some distortion in the noise-impulse wave form as it passes through the transformer. But, the overall magnitude of a normal-mode noise impulse will be relatively unaffected by the insertion of a transformer in the circuit.

Before we talk about how to solve this problem—the problem of normal- mode noise passing through a transformer—let’s talk about another source of normal-mode noise: conversion. By conversion, we mean common- to normal-mode conversion. ill. 9-6 shows common-mode noise appearing on the primary windings of a transformer while it becomes normal-mode noise on the secondary windings. How can this be?

ill. 9.6. A transformer converts common-mode noise into normal-mode noise.

In Section 6, we discussed how common-mode noise appears at the terminals of a transformer. Since the impulse is common to both the line and neutral, these two impulses appear across the transformer 1800 out of phase. Ideally, this means that a cancellation will take place in the primary coil. Theory is one thing, reality is another.

Because of imperfections in the construction of transformers, reactive as well as capacitive discontinuities exist. This affects the speed of the noise travelling through the primary windings, as well as the distribution of current along the wires. These imperfections result in an imperfect “bucking” or “cancelling” of the common-mode signals. The resultant current from these variations produces a differential from the line to neutral in the secondary windings. This action causes common-mode voltages in the primary to produce normal-mode voltages in the secondary windings.

Lumped Capacitance

Whether the normal-mode noise on the secondary is coupled directly from the primary or converted by the transformer into normal-mode noise, there are two things that can be done to dramatically eliminate it. The application of lumped capacitance, connected from line to neutral on the secondary side, can provide a lower impedance path through the capacitive elements than through the load. Also, the application of an MOV from line to neutral will assist in handling high-energy impulses that might pass through the transformer (ill. 9-7). We also see a second shield introduced specifically to eliminate common- to normal-mode conversion. This shield, however, is almost never to be found in products designed for the PC market.

ill. 9.7. The addition of an MOV, along with lumped capacitance, will dramatically attenuate normal-mode noise and high-energy normal-mode impulses.

Let’s see how this combination of MOV and lumped capacitance works to attenuate normal-mode noise. ill. 9-8 shows a photo of normal-mode noise. Wow, that’s a lot of noise. Actually when the photo in ill. 9-8 was taken, the gain on the oscilloscope was turned up as far as possible. Or, put another way, the sensitivity was turned to the highest setting. The noise at this setting literally fills the screen. Next, we put the power conditioner into the line, using the same settings. Nothing else was changed. Notice the dramatic difference (ill. 9-9).

ill. 9-8. Normal-mode noise fills the oscilloscope screen.

The power conditioner has obviously done several things to the noise. Most obvious is the magnitude of the noise, which has been reduced significantly.

ill. 9-9. The same noise shown in ill. 9-8 after a power conditioner, using the circuit design shown in ill. 9-7, is put in the line.

This is the most important effect of the power conditioner. Secondly, all the high-frequency components seem to have been stripped away. and lastly, the noise that is left has a much rounder look to it. Another way of describing this might be to say that the slope of the noise is less steep.

These last two effects are a result of the transformer working with the capacitors, much in the same way that a filter works. The inductance of the transformer windings has slowed down the rise time of the noise, while the capacitors have provided a return path. This action shunts the noise away from the load as we see it on the screen.

Clearly, care must be taken to select the proper values for the capacitors. it's not unheard of for these capacitors to form resonances with the load elements, causing disruptive problems, including the burnout of capacitors in the load devices. The capacitors work together with the inductance of the transformer. As frequency increases, capacitive reactance decreases. This means more and more noise sees the path through the capacitors as a much lower impedance path than the path through the load. Meanwhile, the inductive reactance associated with the windings in the transformer increases, opposing the flow of high-frequency current. This combination, if engineered properly, can be an effective normal-mode filter. This is why some companies call their product a filter instead of a power conditioner.

There are more benefits to this lumped capacitance than merely the reduction of normal-mode noise. Should common-mode noise appear on the live side of the transformer secondary, the lumped capacitance provides a low impedance path to ground. This works both ways. The common-mode source may indeed be upstream from the conditioner. This is the main perceived mission of the conditioner, to block noise from moving downstream. There might also be significant common-mode noise generated on the load side. The capacitors complete a noise-current loop, in much the same way as the neutral- to-ground bond, again keeping the noise from propagating to other devices.

Other Benefits

We stated earlier in the section that power-line conditioners will solve most of the power problems that a PC will experience. As we will see more vividly in the next section, a simple transformer will not compensate for line-voltage drops or power outages. We will also see why we needn’t be concerned about “sags” except in regard to certain types of equipment. “Outages” will be covered later.

It is the software disruption and hardware damage that most users are worried about. The power-line conditioner is a complete, portable, friendly, electrical environment for the personal computer. Because of its efficiency, the power-line conditioner gives off very little heat—certainly very little acoustical noise. Conditioners come suitably packaged for the home or office environment, and , for the most part, they are UL listed. More on that subject in a moment.

Another benefit of the power-line conditioner is that it's a low-impedance device. What this means is that, to the load, it appears as if the transformer is not even there. When the load’s instantaneous current demands an increase, the transformer delivers the required current. An isolation transformer has an impedance of from 3% - 6%, with 5% being about normal. How this percentage is derived is beyond the scope of this discussion. But there are devices on the market with ratings of around 15%. What this really means is that a low-impedance device does not impede the passage of current through it. An isolation transformer can pass currents that are 500% to 1000% of its rated load for very short periods of time. Devices with higher internal impedances can't do so, and may even damage themselves or starve the load for current if they tried.

In Section 6, we discussed the current demands of a switching power supply. It draws all the current it needs at the peak of the sine wave. Earlier, we discussed inrush currents. Motors starting and stopping, devices being turned on and off, and a R/W head seeking across a disk—all these things combine to make a computer system an ever-changing, dynamic load. The power source servicing this system must be able to deliver current on demand. This task a low-impedance isolation transformer can do admirably.

UL Listed Devices

For years, Underwriter’s Laboratories, Inc. (UL) has been testing equipment to certain performance standards. The UL listing and the associated label have long been a sign of safety for the user. We certainly don't want to argue UL’s ability to certify product safety—safety can't be sacrificed.

But, as of July 2, 1987, Underwriter’s Laboratories took a giant step forward. It was then that a brand new standard was introduced—the UL 1449 Standard for Transient and Surge Suppression. UL’s new standard starts where the IEEE 587 standard leaves off. It uses the IEEE standard as a starting point and then clarifies many of the areas left open previously.

The waveshape that is used is a 6000-volt impulse with a 1.2-us rise time. This then decays over 50-us. For plug-in-type devices, 500 amperes are used. After this initial hit, a series of smaller impulses are fed to the units.

The standard covers areas of ground continuity and safety as well as performance. UL 1449 is a welcome step forward and a criteria the cautious buyer should be aware of.

Power Distribution

Our discussion so far has assumed a particular kind of power conditioner. Most small power conditioners come in sizes like 200, 300, 500, 800, 1000, 1500, and 2000 VA (volt-amperes). These units generally plug right into a standard wall receptacle. They often will have two or more receptacles for power distribution. Each unit has a particular rating, either in VA, as stated above, or in amperes. it's important that the rated load not be exceeded either by plugging too many devices into the conditioner or by using a lot of extension cords. Several units can be used where more power is desirable. Each unit should be tied together with the others, by using a heavy bonding strap to equalize any chassis voltage differentials.

Often, as the size of the unit goes up, the number of receptacles on the back panel go up. Because of the larger size of the cabinet, as well as the ampacity of the unit, more outlets can be accommodated. Also, with increased ampacity comes the capability of mounting 20-ampere receptacles on the unit. Most manufacturers have a variety of combinations that are standard. We mention this subject for the simple reason that proper power distribution is vital for the optimum use of a line conditioner. As stated in Section 5, it's desirable to avoid ground loops by using a common power strip. In the case of power conditioners, all peripherals should be plugged into the unit. This may not be possible, and when it's not, a grounding strap should tie all chassis to a single-point ground.

There is another type of power conditioner that has no power distribution. These devices follow the same design criteria that we have laid out in this section. However, they are meant to be “hard-wired” upstream from a power panel. This means that every branch circuit fed from the panel board will enjoy fully conditioned power. Typically, these devices are in commercial, as opposed to consumer, enclosures and are meant for installation in equipment rooms and electrical closets.

To take advantage of a commercial-grade power conditioner, it's necessary for each branch circuit to feed some type of computer or related equipment. It would not do, for instance, to condition the power for a panel that fed both office electronic equipment and some type of corrupting equipment, like copiers or air conditioners. Lithe panel feeds devices that are common power polluters, the computer equipment must be serviced from another panel, or the conditioning should take place downstream from the panel.

These types of conditioners are available in sizes from about 2 kVA to hundreds of kVA. In the office environment, their application could be used to support networks or large numbers of work stations. We will explore this concept where other products are concerned at a later time.

The concept of conditioned power is one that we will find throughout the rest of the guide. We will see other ways of doing it using different technologies. Or, as will be found in the next section, we will see this basic concept expanded to include other more exotic features.

ill. 9-10. A power-line conditioner has attenuated the 900-volt impulse to 7.5 volts, but has not converted the noise from normal to common mode.

But, before we leave this section, you may have wondered how a power-line conditioner handles our 900-volt impulse. ill. 9-10 shows a photo of this. No normal- to common-mode conversion has taken place, and the device has attenuated the impulse to 7.5 volts.

NEXT: Voltage Regulators

PREV: Surge Suppressors

All related articles   Top of Page   Home