Transient Suppression Devices

There are a number of transient suppression devices on the market today. By this, we mean the discrete components that are the primary defense against damaging voltage impulses.

IEEE 587 and ANSI C62.41 STANDARDS

It turns out it's much easier to design a product to suppress something if you have some notion of what that something is.

IEEE stands for the International Association of Electrical and Electronic Engineers and is often referred to as “eye triple E.” ANSI stands for the American National Standards Institute—called “anssee.” These groups have come up with a description of what an impulse in a low-voltage (120-V AC) circuit might look like. This is based on a lot of research, interpolation, and field experience. ill. 7-1 shows the famous IEEE 587 “ring wave.” This is also referred to as ANSI C62.41, but we will ignore the politics of transient voltages.

ill. 7.1. IEEE 587 0.5- 100-kHz ring wave.

Description was only one priority for the working group that took up the subject of transients. The other priority was to establish the ring wave as a standard for testing. The ring wave is not the only waveform that the group of engineers came up with. But the common household or office convenience receptacle falls into the ring-wave category. Based on research and testing, the ring wave has been born out as a reasonably good likeness of the typical impulse that might appear in those electrical outlet locations where PCs are to be found connected.

The wave (ill. 7-1) is expressed in percentages—0.1 of the peak voltage, 0.9 of the peak voltage, and so forth. So what is the peak voltage? IEEE makes an attempt to predict what the peak voltage might be. This is reasonable since not every event is going to release exactly the same amount of energy; but experience has shown that they do tend to fall within certain boundaries. The IEEE standard says that since building wiring is known to experience insulation breakdown and sparkover at 6000 volts, we can expect events of no greater magnitude than 6 kV. Notice also that the standard predicts that each peak in voltage oscillation will be 0.6 of the previous peak. This describes an oscillating event that gradually dissipates.

The oscillating nature of the wave, plus the possibility of 6-kV peaks, are critical criteria for evaluating equipment. When a product claims to meet IEEE 587 standards, or when the product literature advertises that the device will withstand repeated hits of a 6-kV IEEE 587 ring wave, you can be sure that the manufacturer is aware of proper engineering design criteria. More on that later.

ANSI and IEEE have addressed two other waveforms. We find them shown in ill. 7-2. it's not terribly important to master the significance of these waveforms since they describe what an impulse might look like at the building service entrance. These impulses are unidirectional, as opposed to being oscillatory like the ring wave. The impedance of the building wiring will cause one to literally become the other.

ill. 7-2. Unidirectional (ANSI C62.41) waveshapes of outdoor and service entrance impulses. (A) Voltage waveform. (B) Current waveform.

Another reason why we should not be too concerned about these particular waveshapes is that they are to be found only at the building service entrance. PCs are connected to the line at such a point that the characteristics of the impulse waveshape have changed considerably. We are not just restating our previous point. A product designed for the unidirectional waveshape should not be used in conjunction with a PC. A manufacturer who would do this is using improper design criteria.

Furthermore, the place to deal with the unidirectional waveshape is at the building service entrance, not anywhere else. We will talk about these devices briefly in a few moments. But, let’s clarify this discussion further by asking what kinds of events produce this unidirectional impulse? The answer is: Major events that occur along the power distribution system. Lightning would be the most common event.

Thus, lightning arresters are the most common device that we think of when it comes to keeping this enormous release of energy away from sensitive equipment. As we shall see, lightning arresters, as a form of protection for computer equipment, are not only an inappropriate application of technology to a problem, but they create a silly sense of security for the user.

Filters

Transient suppression devices come in two basic forms. The first form is made up of Ls and Cs. Sounds like alphabet soup, but what we are really talking about are inductors and capacitors arranged in a circuit, with that circuit being used to block or attenuate the propagation of high-energy noise into the critical load. Shortly, we will talk about another kind of suppression device—diverters.

Remember from an earlier discussion that we mentioned the subject of resonance, and we talked about how the reactance of a capacitor or inductor changes as the frequency of the signal applied across it changes. This property of coils and capacitors makes them ideal for the purpose of blocking certain noise frequencies. But, we are not going to get heavily involved in the theory behind this. Let’s just consider a simple two-element filter (ill. 7-3).

In this case, we want to permit the normal 120-volt AC line current to flow to the “critical load,” but we don’t want the noise that might appear between the line and neutral to enter the load and cause problems. At some frequency, we know that the coil (L1) will develop a high impedance. This type of coil or inductor is commonly called a choke. So, we select this choke to have a particular reactance at the noise frequency.

ill. 7-3. A two-element filter using a coil (inductor) and a capacitor.

As noise tries to pass through the choke, an opposing current builds up that is 1800 out of phase to the noise. This opposition to the noise current attenuates the noise and dissipates its energy as heat. But just to make sure, we have chosen capacitance (C1) to have a low impedance to the noise, so it will see the path through C1 as a virtual short circuit, while the line current will see it as an open circuit. Thus, any remaining noise is given a path away from the load. This kind of filter is called a low-pass filter because it passes low frequencies while attenuating the higher frequencies.

It may have occurred to you that building a filter for all the various frequencies which might be considered noise would be very difficult. and , in fact, it's . it's far easier to design a filter to block a certain narrow band of frequencies than it's to build one that must work well over a broad band. This is one of the problems with filters. We will see this again in a later section. However, a well-engineered filter can be a power problem solver. The effects of a filter on that band of noise to which it's most closely tuned to is indeed dramatic.

Diverters

If we look back at our filter circuit of ill. 7-3, we might be able to make the argument that the capacitance (C1) does more diverting than attenuating. and we would be correct. There are several additional devices that perform this same function and which are designed to handle varying degrees of energy. We will look at each device, one at a time. Then, the stage will be set to take a look inside of those products in which they are used. Diverters are divided into two more categories: crowbars and clamps.

Crowbars

A spark gap is the driving concept behind all crowbar-type devices. But before we find out how a spark gap becomes a crowbar, let’s think for a moment about the definition and use of a crowbar. When we wedge an iron (construction-type) crowbar between iron, wood, or cement materials, there are periods of time when we apply more and more pressure. Finally, the materials give and separate, and our leverage pays off. Crowbar-type electronic devices that use spark gaps work in much the same way. Pressure in the form of a voltage buildup appears across the spark gap. When enough force (electromotive force) is applied, the gap ionizes, forming an electrical path.

The typical crowbar-type device in present use is a gas tube. Inside a sealed chamber, electrodes are placed in close proximity. This gap is filled with a special gas that is selected for its ionization properties. If the gas tube is connected between a line and ground, high-energy impulses will cause a high voltage to appear across the terminals of the gas tube. At some point, this potential will ionize the gas and a spark will bridge the spark gap, allowing energy to flow to ground. This is exactly how a lightning arrester works.

ill. 7-4. The gas-surge arrester, just one gap-type device, has electrical parameters that are a function of gas pressure and content, gap distance, and electrode size, content, and mass.

From the standpoint of a PC, there are a couple of inherent problems with crowbar devices. The first problem you may already have guessed. it's called the volt-time response. This refers to the amount of time it takes for the voltage buildup to trigger the arc mode of the gas tube. Typically, this is in the millisecond range. The problem is that during the initial rise time of the impulse, no protection is being offered the PC.

Ionizing the gas inside the tube is not a perfect process. This leads to varying voltage levels that might be needed to fire the device—sometimes higher, sometimes lower. Lower voltage ratings are difficult to produce with consistency due to the small gap that must be manufactured. The gap- or arrester-type device is well suited to handling large amounts of energy, far better than handling small amounts of energy.

After the high-energy impulse has dropped below the device’s firing threshold, the gas between the electrodes of the tube may remain ionized. Thus, a path to ground would remain, causing power to continue to flow away from the load. In other words, power from the steady-state voltage source will follow the path of the impulse. The clear result of this is a temporary power outage. Due to the nature of the device itself, it will often cause an outage or a significant sag as a result of its operation. At some point during the sine-wave period, the device will extinguish itself. it's not unusual, however, to find the hot gas and electrodes reigniting during the next voltage peak.

Still another problem for PCs is that the functioning of a gap device can cause an impulse of dangerous magnitude to occur. Arcing, sparking, and firing are not the kinds of events we would typically encourage near a computer. Also, the functioning of the gap device may couple energy into other conductors, which might direct this energy into some sensitive equipment. Remember that earlier we mentioned the importance of designing products in anticipation of impulse waveshapes. Crowbars bring this issue into focus. The unidirectional waveshape is usually an event of high magnitude moving in one direction. This type of transient is ideal for a gap device. The much-lower-magnitude ring wave and its characteristic environs don't match well with the functioning of spark gaps. A ring wave calls for a different approach. Gas-discharge tubes are useful for the high-current type of waveshapes found at the building service entrance. They are not as effective when used to protect low-voltage, low-impedance circuits.

Clamps

Now we come to the good stuff. Clamps are the most common devices used in protective circuitry. They come in two basic varieties: zener or avalanche diodes and metal-oxide varistors (MOVs). The two discrete components have differing properties, but overlapping functions. We will look at avalanche diodes first.

Avalanche Diodes

“Diode” is a term we have heard before. it's at the very heart of the rectifier circuits we discussed earlier. Remember that it has an anode and cathode, and when properly biased will allow current to flow. When it's reverse biased, no current will flow.

This is where the zener or avalanche diode stands alone from other types of diodes. Yes, it's true that when it's reverse biased, no current will flow. But this is only true to a point. That point is called the breakdown voltage. What this means is that at some reverse-bias voltage, the internal PN junction of the diode breaks down. Now, we don’t mean it breaks down physically and causes some kind of harm to the junction. What breakdown means in this context is that the physical behavior of the electrons suddenly change, allowing current to flow across the PN junction. This action is due to special doping of the semiconductor material during the fabrication process. ill. 7-5 shows this at work. When the voltage across the diode reaches the breakdown magnitude, current flows through the diode. The diode acts as a voltage divider and a low-impedance path. The bulk of the noise current then flows through the diode.

ill. 7-5. When the noise voltage potential reaches the breakdown voltage, the zener diode becomes a low-impedance path.

Two unanswered questions may still be hanging in the air at this point. The first relates to ill. 7-5. We have shown polarity signs at the input of the circuit. If we assume an alternating current source, clearly then, the diode would only protect during the half cycle when it's reverse biased. We have not shown the complete protective circuit for simplicity’s sake. The answer to having protection when the polarity is reversed is to put two diodes back-to-back.

The other question may be as to why are zeners called avalanche diodes? Avalanching is the process of charge carriers (which are accelerated to a high velocity by the buildup voltage applied to the device) colliding with valence electrons. This all has to due with the physics of the crystalline structure of the P region, the N region, and the junction between the two. These valence electrons collide with more valence electrons, producing a snow-balling effect— thus the term, avalanche. This process generates a nearly instantaneous high current flow across the junction, thereby creating a low-impedance path. Holes are involved in this process also, in the same way.

ill. 7-6. Zener diodes are constructed using a large-area PN junction which has been affixed between two heat sinks to protect the junction during impulse conduction.

We are now going to leave the zener diode, but we will talk about its relative advantages and disadvantages later. We want to wait and compare it to the next device being discussed. Comparison between the two devices will make the benefits and weaknesses of each stand out even more clearly.

MOVs

As we stated earlier, MOV stands for metal-oxide varistor. Varistor is an amalgamation of the words variable and resistor. A varistor is a variable resistor. This should give some clue as to its operation. Although different in construction, MOVs are voltage-dependent devices which operate in a fashion similar to back-to-back zener diodes.

The origin of the MOV is a story as serendipitous as the discovery of penicillin. Supposedly, the electrical properties of silicon-carbide grains were discovered when a grinding wheel accidentally became part of a circuit on a disorderly workbench. For years after, varistors made from silicon carbide were formed in the shape of tiny grinding wheels. It seems the fabrication of varistors is as much an art as it's a science.

Modern varistors are made of zinc oxide, along with bismuth, manganese, cobalt, and other metal oxides. The structure of an MOV is a matrix of zinc-oxide grains bonded together in such a way that the boundaries of the grains have the characteristics of a PN-junction semiconductor (ill. 7-7). Like the zener diode, these junctions block the conduction of electricity at low voltages and allow conduction at higher voltages.

ill. 7-8 shows a varistor at work. While the voltage applied by the transient has much to do with the operation of a MOV, the current affects its clamping voltage to some degree. The important fact about varistors is that they clip off the voltage impulse at a certain level and conduct the impulse current away from the sensitive load.

ill. 7-7. The structure of an MOV is a matrix of zinc-oxide grains bonded with a bismuth material.

ill. 7-8. Diagram showing a varistor-clamped voltage impulse.

Zeners and MOVs Compared

So, which device is better, a zener diode or an MOV? During the period of impulse conduction, large amounts of power are absorbed by the device. This might last from only nanoseconds to milliseconds, but current flowing through resistance creates heat. The life of the suppressor is dependent upon how it tolerates heat—can it dissipate the heat before the heat dam ages the crystalline structure that is doing the work.

Junction burnout has long been a problem where zener diodes are concerned. The PN junction must have a way of directing heat away from it. MOVs, on the other hand, distribute heat evenly throughout the device and are inherently better able to withstand energy over repeated numbers of events. The manufacturers of zener diodes are aware of this and have developed products that compete favorably.

Zener diodes have an offset advantage to the heating problem. They have lower clamping voltages. Obviously, lower clamping voltages mean that less current will be allowed to flow to the sensitive load. There is some argument that zeners have a faster response than MOVs, with the response time of some zeners being claimed to be on the order of one picosecond. This is labeled as theoretical. MOVs are rated at down to 500 picoseconds.

Other considerations are leakage current, aging, and capacitance. All of these are points of contention between manufacturers, and each manufacturer has developed design strategies to alleviate technological weaknesses.

It is natural for each manufacturer to think their device is by far the most effective in all cases. The fact remains, however, that most designers tend to think of the MOV as a more rugged device at higher energies, while the zener diode is a much more versatile device in low-power circuits. it's for these reasons that MOVs are used far more often in transient suppression products for personal computers.

MOVs at Work

Let’s take a closer look at the clamping action of an MOV as it might appear in a surge protector type of product. The typical MOV would be rated to clamp voltages above 130-volts rms. This is an important rating. The 130- volts rms means that impulses will be clipped above 184-volts peak. (130-volts rms times 1.41 equals 184-volts peak. Normal AC power is 120-volts rms or 170-volts peak.) One might be tempted to subtract one from the other and conclude that the impulse might be limited to only 14 volts.

Of course, this might be true if the impulse occurred right at the peak of the sine wave. But, as ill. 7-9 shows, if the impulse occurred at the negative peak, the impulse might have a magnitude of 368 volts. In fact the manufacturer’s specification for this particular device claims this voltage to be as low as 340 volts. These values will vary slightly depending on the varistor and the impulse current and waveshape, but the principle we have illustrated is an important one.

ill. 7.9. How an MOV clips impulses.

Another important consideration is that negative-going impulses are not clipped. As was shown earlier, many impulses seen inside a building have a ringing waveshape. We have no guarantee that every transient event will be positive in polarity. The vast majority of damaging events, however, have their beginnings as positive, unidirectional impulses. This does not mean that energy which is coupled, reflected, and induced will always meet this neat scenario, and will allow the suppressor to act as if it were operating under laboratory conditions.

MOVs in Circuits

Like avalanche diodes, MOVs are connected to the computer in parallel. This allows the MOV to act as a low-impedance path. Undesired currents are short- circuited through the MO keeping the currents away from sensitive circuits.

In ill. 7-5, we saw how a zener diode might be connected across a load, from line to neutral. But we know that there are actually three wires connecting our PC to the outside world. How might this effect the connection of transient suppressors across the lines? ill. 7-10 shows three alternative connections, and will be our introduction into the next section on Surge Suppressors. We see three distinct options for connecting MOVs across lines. Let’s explore each and see what we might accomplish in the way of protection.

ill. 7-10. How MOVs can be connected to protect a PC.

MOV 1 is connected between line and neutral. We might remember from an earlier discussion that normal-mode noise appears from line to neutral. Clearly, what MOV 1 will do is provide a low-impedance path to complete the noise circuit from line to neutral. By diverting the impulse current through the MOV, we are protecting the power supply of the PC.

MOV 2 is placed from line to ground. If it were used without MOV 1, no protection would result unless the impulse was of the common-mode variety. In that case, we would need MOV 3 since common-mode noise appears between both the line and ground, and neutral and ground. If the event were normal mode, MOV 2 would not be activated because it would not see any voltage differences from line to neutral.

Obviously, MOV 3 is needed to complement MOV 2, and form total common-mode protection. As we will see in the next section, many design engineers think of the ground wire as a pipeline to the electrical cesspool, and the use of a single MOV from line to ground is not uncommon. Electricity, even transient voltage impulses, travels in circuits. We can see that the neutral-to-ground terminals of the computer are connected across the current path flowing through MOV 3. This current will cause a voltage drop across the impedance of that circuit. We can see there is no alternative path for this current should the computer somehow bridge the gap from neutral to ground. All of this voltage will appear across the stray capacitance of the various components and boards inside the PC. and , of course, if a ground loop is present, we may have unseen paths that lead through boards, buses, and cables before returning.

Using all three MOVs is the most desirable design. This gives protection from both normal-mode and common-mode events. However, the let through voltage of an MOV is far too high, possibly by several hundred volts, to really provide common-mode protection. and , as we shall shortly see, if we use an MOV from line to neutral, we must have them from both neutral to ground and from line to ground. Why? Because MOV 1, while it absorbs some energy, converts one form of noise to another. Let’s move on to Section 8, and we shall see this process in action.

NEXT: Surge Suppressors

PREV: PC Power Supplies

All related articles   Top of Page   Home