Fire Hazards of Surge Suppressors

Copyright 2001-2002 by Ronald B. Standler

Table of Contents

            MCOV Values
Coordination of Arresters and Suppressors
Reaction of Manufacturers
My Suggestions
            surge suppressors hidden inside walls
Annotated Bibliography


Electronic equipment containing transistors, and particularly equipment containing integrated circuits (e.g., personal computers, television receivers, videotape recorders), are vulnerable to damage by transient overvoltages (also called "surges") on the ac supply mains. In the absence of a surge suppressor, such transient overvoltages may have peak voltages as high as 6000 V, which can destroy transistors, integrated circuits, and other electronic components. These transient overvoltages are commonly caused by lightning and switching reactive loads on the ac supply mains. As a result of the threat and vulnerability, it is good engineering practice to plug every personal computer or other electronic system into a surge suppressor.

In the late 1980s fire investigators became aware that surge suppressors could cause fires. Such knowledge resulted in fire departments often identifying surge suppressors as the cause of a fire. For example:
  1. The University of Washington posted a webpage about the fire hazard of surge suppressors, which mentions that "two families on Bainbridge Island [in Puget Sound] lost their homes due to fires caused by Multiple Outlet Power Surge Suppressors" during the one year ending Feb 1995. Another copy is posted by the Bellingham, Washington fire department. This webpage has been summarized at the website of the safety office at the University of California at San Diego.

  2. The U.S. Department of State issued a pamphlet in Sep 1998, titled Surge Suppressors Can Cause Fires after a varistor caused a $ 150,000 loss at the U.S. Embassy in Grenada and "a number of fires" elsewhere.

  3. The U.S. Department of Energy's facility at Hanford posted a Lessons Learned with a photograph of a surge suppressor that "started a small fire" at the Stanford Linear Accelerator Center on 28 Aug 1999.

  4. In January 2003, an employee of the Greensport Yard in Houston arrived at work to find the office full of smoke from a burning surge suppressor, with a second surge suppressor of the same model hot and ready to burn. This incident is reported by the Environmental Health & Safety Office of New Mexico State University, and at An Adobe PDF version of the report is available from a fire department in Pennsylvania.
These few webpages are just the "tip of the iceberg" of a much larger problem. Few people who have a burning surge suppressor post a webpage about the problem. Indeed, the true cause of many fires in homes and businesses is probably never identified, because – after the fire – everything is black and items in plastic enclosures are melted by the heat of the ensuing fire, making disassembly and diagnosis difficult.

The scope of this essay only includes surge suppressors used in the USA on single-phase nominal 120 V ac electric service, which is commonly used in homes and offices in the USA.

I am posting this essay to warn the public about fire hazards of surge suppressors, and to make a few practical suggestions to avoid such fires. While there is a bibliography of technical papers at the end of this essay, I have attempted to minimize the amount of technical information in the text of this essay.


Metal-oxide varistors (MOVs), a surge-protective component, were invented in Japan by Matsushita Electric Corp. in 1968 and extensively marketed in the USA from 1973 to the late 1980s by General Electric corporation, and subsequently by Harris, now Littelfuse. There are now at least a half-dozen manufacturers of metal-oxide varistors in the world. The development of metal-oxide varistors coincided with the widespread adoption of vulnerable electronic equipment (e.g., personal computers, television and stereos containing integrated circuits, etc.) that needed protection from surges on the ac supply mains.

Varistors connected to the ac supply mains are normally nonconducting. When a transient overvoltage occurs, varistor(s) become conducting and divert surge current away from the vulnerable equipment downstream, as well as limit the magnitude of the surge voltage.

More than 109 varistors were sold worldwide in 1989. Varistors are included inside surge suppressors, which are mostly commonly packaged as an outlet strip, but can also take other forms. Typical surge suppressors contain between one and three varistors per suppressor, so there were at least 108 surge suppressors sold each year worldwide in the late 1980s.

The varistor itself is a ZnO ceramic, which does not burn. However, continuous conduction of the varistor can cause the epoxy coating on the outside of the varistor to burn like a candle. At somewhat lower power dissipation in the varistor, the varistor may become hot enough to melt plastics in the vicinity of the varistor. Under some conditions, flammable plastics might be ignited by a hot varistor or by the burning epoxy coating on a varistor.

Recommend Higher MCOV Values

If all varistors connected to the ac supply mains were rated to be nonconducting during ac voltages that are at least 1.4 times the nominal line-earth voltage, and preferably 2.0 times the nominal line-earth voltage [i.e., a maximum continuous operating voltage (MCOV) rating of at least 170 V ac, and preferably 250 V ac, for varistors used on nominal 120 V ac service], then varistors would rarely cause fires.

However, nearly all manufacturers of surge suppressors in the USA use varistors with a maximum continuous operating voltage (MCOV) rating of only 130 V ac in suppressors for use at nominal 120 V ac service. This is the smallest MCOV rating that will not conduct on normal utility voltage. The advantage of using varistors with this smallest MCOV rating is that it also gives the lowest voltage protection level (sometimes called "clamping voltage" in the USA) during a surge. Arguably, the lowest voltage protection level during a surge puts less stress on the equipment downstream from the surge suppressor.

The Underwriters' Laboratory standard 1449, titled "Transient Voltage Surge Suppressors", assigns ratings to surge suppressors based on their voltage protection level, with 330 V being the lowest – and presumedly the best – level. Some engineers have criticized UL for including this performance rating in their tests, which are supposedly confined to safety considerations. However, without this UL voltage protection rating, manufacturers would still be clamoring for the lowest possible voltage protection level, and the claims between competing manufacturers would be at different peak surge currents or different surge current waveshapes, thus precluding meaningful comparisons by consumers.

A varistor with a MCOV rating of 130 V ac typically conducts a current of 0.001 A dc at a voltage, VN, of 200 V dc. VN is generally considered the boundary between the nonconducting and conducting states of the varistor: the varistor is nonconducting at magnitudes of voltages less than VN. The nominal 120 V ac sinusoidal voltage used in the USA has a peak voltage of about 170 V. A comparison of these two voltages, 170 and 200, shows that there is a margin of only 30 V dc between the peak of the normal ac mains voltage and the voltage at which the varistor begins to conduct. This margin of 30 V dc is too small a safety margin.

There are numerous disturbances of the utility voltage that can cause the peak voltage to exceed the nominal value of 170 V:
  1. Loss of the neutral conductor on common 120/240 V ac systems can cause a temporary overvoltage on half of the branch circuits inside a building. During such a temporary overvoltage, the rms voltage of the 60 Hz sinusoidal waveform is abnormally high for durations for minutes, or even hours.

  2. There are long-duration surges (i.e., surges with durations of milliseconds) that have but the capability of transferring large amounts of energy to conducting varistors. Opening of fuses in the utility distribution system is one cause of such long-duration surges; another cause is switching of power-factor correction capacitors in the utility distribution system.

  3. There are temporary overvoltages (also called "swells") during which the rms voltage of the 60 Hz sinusoidal waveform is abnormally high for durations between a half-cycle and a few tens of minutes. Common causes of such temporary overvoltages include sudden load shedding in the utility distribution network, malfunction of voltage regulators at the utility's substation, and dropping a 25 kV distribution line onto a 12 kV distribution line (e.g., when an automobile hits a pole that supports overhead lines, or when ice on lines causes the upper line to break).

Varistors are designed and intended only for protection against transient overvoltages with a duration less than a millisecond, not for regulation of the continuous, sinusoidal utility voltage! Choosing varistors with larger MCOV ratings helps protect the varistors from conduction, and subsequently burning, during temporary overvoltages.

For many years, designers of surge suppressors simply assumed that the lowest voltage protection level would give the best protection to the equipment, without being aware that there were other considerations (e.g., coordination of arresters and suppressors, fire hazards of surge suppressors inside buildings). In 1990, Smith and Standler did experiments in a laboratory that showed that a small sample of consumer appliances required peak open-circuit surge voltages of more than 2000 V to cause immediate damage. As a result, differences in voltage protection level of surge suppressors between 330 and 500 V may not be significant.

The selection of the MCOV rating of the varistor is important, not only for ability to avoid fires caused by disturbances of the voltage on the ac supply mains, but also because of the need to coordinate a surge suppressor with an upstream surge arrester. Coordination is discussed in the next section of this essay.

It has become apparent that one might need to choose varistors with an MCOV rating of 250 V ac to survive most abnormal increases of rms voltage (i.e., "temporary overvoltages" or "swells") on ac supply mains with a nominal voltage of 120 V ac. Such a large MCOV rating of the surge suppressor can still be accompanied by low voltage protection levels inside the building, as discussed later in this essay.

Coordination of Surge
Arresters & Suppressors

Surge protection takes two forms:
  1. a surge arrester is connected at the main circuit breaker panel and provides protection for the entire building from surges traveling into the building on the electric utility's wires (e.g., lightning strikes, switching reactive loads, etc.)

  2. a surge suppressor connected between the wall outlet and vulnerable equipment (e.g., computer), often in an outlet strip or hidden inside the equipment itself.
Both surge arresters and surge suppressor are members of the class of surge-protective devices (SPDs). In general, a surge arrester is designed to divert larger surge currents, and absorb more energy from a surge, than a surge suppressor.

Engineers who design surge-protective devices seem to assume that their products will be used alone in a building. However, good engineering practice is to use a surge arrester at the point where the utility's wires enter the building and a surge suppressor at a wall outlet or inside vulnerable equipment. One will then need to coordinate the arrester and suppressor, so that they work well together.

The conventional practice in the USA, from the 1970s to the mid-1990s, was to use a surge arrester with an MCOV rating of 175 V ac (i.e., the lowest rating in ANSI/IEEE C62.1-1984 and ANSI/IEEE C62.11-1987) together with surge suppressors with MCOV ratings of 130 V ac. Unfortunately, this conventional practice resulted in poor coordination, and often the surge suppressor (with small surge energy rating) attempted to protect the surge arrester (with a large surge energy rating). This conventional practice arose because surge arresters were designed and manufactured by engineers who specialized in SPDs for high-voltage utility transmission and distribution lines, where concerns about long lifetime were part of the design and testing. On the other hand, surge suppressors were often produced with little or no engineering design and testing, by simply including a varistor with a 130 V rating into an outlet strip, to get the lowest possible voltage protection level.

As pointed out by Scuka in 1987, and independently in Standler's 1989 book at page 296, the best design is to have an arrester with a voltage protection level lower than the voltage protection level of the surge suppressor.

In a paper at the 1991 Zürich EMC Symposium, Standler advocated a surge arrester with an MCOV rating of 150 V ac together with surge suppressors with MCOV rating of 250 V ac, for use on mains with a nominal voltage of 120 V ac.

There are four advantages of such coordination of a surge arrester and surge suppressor:
  1. Lower total cost of surge protection. One relatively expensive surge arrester is recommended at the point where electric power enters the building, and inexpensive (less than US$ 5 each) surge suppressors are recommended at many wall outlets inside the building where vulnerable electronic equipments are connected. Because of the presence of a properly coordinated arrester upstream, a surge suppressor can use an inexpensive varistor with a diameter of 7 or 10 mm, instead of conventional practice of more expensive varistor(s) with a diameter of 14 or 20 mm.

  2. Better electromagnetic compatibility because there will be smaller radiated magnetic fields inside the building from surge currents.

  3. Lower surge voltage at wall outlets that are not protected by a surge suppressor.

  4. Lower probability of a surge causing a suppressor to explode.
R.B. Standler, 1992 IEEE EMC Symposium at page 198;
R.B. Standler, 1991 Zürich EMC Symposium at page 522.

By using an arrester with an MCOV rating of 150 V ac, one can obtain a low voltage protection level inside the building for all surges that enter from outside the building, and also for surges with both an origin inside the building and durations longer than a few microseconds. Surges inside the building with durations less than a few microseconds can be blocked by common low-pass filters in electronic equipment, as well as diverted by suppressors with an MCOV of 250 V ac.

Although not common practice, it is possible to design arresters that will interrupt 60 Hz current to downstream loads when the arrester fails, so the arrester can protect the equipment (and surge suppressors) inside the building even after the arrester has "failed". For example, the varistor can be included inside a thermal circuit breaker package, so heat from the failing varistor will trip the breaker.
Edward K. Howell, Combination Circuit Breaker - Lightning Arrestor, U.S. Patent 4,168,514,   18 Sep 1979.

The large surge currents in a direct lightning strike, or the large 60 Hz fault currents on the ac supply mains, can cause a surge-protective device to explode. It is better that a surge arrester explode, instead of a surge suppressor, because the arrester is likely to be located inside a metal circuit breaker panel or in a utility room, where the explosion is less likely to injure people and where there is less flammable material nearby. In contrast, surge suppressors are typically located under desks in rooms where people work.

Reaction of Manufacturers

In the late 1970s, General Electric was aware that some of their surge suppressors were burning. In reacting to these failures, General Electric made two changes during the late 1970s in their surge suppressors for use on nominal 120 V ac:
  1. included a thermal disconnector that would disconnect a hot varistor from the ac supply mains, hopefully before the varistor's epoxy coating began to burn, and
  2. increased the maximum continuous operating voltage (MCOV) rating of the varistor from 130 V ac to 170 V ac, to provide a greater margin of safety between the peak of the sinusoidal mains voltage and the minimum voltage at which the varistor begins to conduct appreciable current.

Despite the design of General Electric's surge suppressors in the late 1970s, most surge suppressors sold in the USA continue to use varistors rated at 130 V ac. Furthermore, despite the design of General Electric's surge suppressors in the late 1970s and several patents in the early 1980s, most surge suppressors sold in the USA before about 1998 had no thermal disconnector.

A thermal disconnector (sometimes called a "thermal cutout", or colloquially called a "thermal fuse") is a device that interrupts the flow of current when the temperature of the device exceeds a rated temperature. Thermal disconnectors commonly contain a wax pellet that melts at a certain temperature, allowing a spring to open electrical contacts. Thermal disconnectors have been available for many decades, they are not a new component.

The second edition of Underwriters' Laboratories standard 1449, which became effective in February 1998, requires that a surge suppressor either fail in a safe manner, or survive connection to twice the nominal mains voltage for seven hours. The easiest way to pass this test, and still have a low voltage protective level during a surge, is to include a thermal disconnector with the varistor(s).

Why did UL take so long to revise its standard to prevent fires caused by surge suppressors? The revision of each UL standard is guided by an "industry advisory group" that is composed of employees of manufacturers of products being tested to the standard. I know from my work in ANSI/IEEE standards during the late 1980s and early 1990s that manufacturers of surge suppressors were vehemently opposed to tests in any performance standard for surge-protective devices for failure modes, tests which might have exposed fire or explosion hazards and found their products to be unacceptable.

A particularly horrifying fact is that many commercial surge suppressors in the USA put the thermal disconnector and varistor in series, so that — after the disconnector opens — the vulnerable equipment downstream from the suppressor is exposed to whatever voltage killed the varistor. Vulnerable electronic equipment (e.g., computers, television receivers, audio equipment) can be damaged by temporary overvoltages. The manufacturers of surge suppressors seem to be more concerned about protecting cheap varistors than protecting expensive electronic equipment! It would be much better engineering practice to connect the thermal disconnector upstream from both the varistors and the protected equipment, so that the equipment is always either protected or disconnected from the utility power.

In the late 1980s, when I was actively involved in development and approval of ANSI/IEEE standards for surge-protective devices, I heard many anecdotal reports of fires caused by surge suppressors. However, the manufacturers of surge suppressors did not disclose their failure rates and did not disclose any information gleaned from their examination of failed surge suppressors. Therefore, engineers, such as me, who were not employed by a manufacturer of a surge suppressor had no idea how prevalent the problem was or what was causing the problem.

My work as an expert witness in one products liability case involving a fire caused by a surge suppressor gave me the opportunity in March 2001 to examine thousands of pages of the importer's documentation about failures of their products. This one importer certainly knew that many of their products were sometimes causing fires. Incidentally, I was interested to read several consumer complaints to that importer involving surge suppressors used underneath aquariums. When the surge suppressor began to burn, the heat from the fire shattered the glass in the aquarium and the falling water extinguished the fire!

The process of writing and approving engineering standards for surge suppressors was dominated by manufacturers of these surge suppressors. Such domination is easy to understand:
  1. The business of these manufacturers is directly affected by a few standards. In contrast, engineers representing users or the public interest were concerned with thousands of standards, and both small travel budgets and limited time prohibited most users and public interest representatives from attending meetings to develop standards for surge suppressors.

  2. Most of the engineering expertise in surge-protective devices, as well as most surge test laboratories, is concentrated in manufacturers of surge-protective devices, which puts engineers who represent the public interest at a disadvantage in technical discussions.

  3. Finally, the possibility of litigation for "restraint of trade" under the anti-trust statutes scared many officers of the IEEE Standards committees and subcommittees from advocating any draft standard that might hurt a manufacturer's sales. These officers had no real understanding of anti-trust case law, but were reacting to their fear of litigation.
From 1985 to 1990, financial support from the U.S. Military and a local electric utility for my research in surge-protective device applications permitted me to be the only professor to be involved in development of ANSI/IEEE standards for low-voltage surge-protective devices. When this financial support was annihilated at the end of the cold war in 1990 and a simultaneous recession that caused utilities to decrease funding for research, I continued to attend meetings at my own expense for five years, then abandoned my work in ANSI/IEEE standards. This is just one of many examples of how drastic decreases in financial support for scientific and engineering research in the USA by both the U.S. Government and utilities have harmed the public.

My Suggestions

As a result of concern about the possibility of fires, some users might be tempted to discard all of their surge suppressors. Discarding all surge suppressors is probably an overreaction – the risk of damage to computers from surges is much greater than the risk of fire from a surge suppressor. In many situations (e.g., businesses, physician's offices, and attorney's offices) it is essential that computers operate continuously and that old data files and documents be quickly retrievable. Without surge protection, computers will be much less reliable.

In saying that the probability of a fire is less than the probability of a surge that would damage an unprotected computer, I am not saying that fires are an acceptable outcome. Products, including surge suppressors, should always fail in safe ways when exposed to foreseeable events. When a surge suppressor causes a fire, the victim's insurance company will pay for the damage, just as with fires from other causes. If the insurance company identifies a surge suppressor as the cause of the fire, the insurance company may file litigation for products liability against either the manufacturer or importer of the surge suppressor.

There is nothing inherently wrong with using metal-oxide varistors in surge suppressors, and a surge suppressor with a given energy absorption rating will be much less expensive when varistors, instead of silicon semiconductors, are used. The problem with varistors in surge suppressors comes from poor design:
  1. too low an MCOV rating,
  2. no thermal disconnector,
  3. cheap plastic enclosure that can burn, and
  4. no coordination with an arrester upstream (or worse: no surge arrester at the circuit breaker panel)

I make the following general suggestions:
  1. Surge suppressors manufactured after January 1998 that have passed the tests in the second edition of Underwriters' Laboratories (UL) Standard 1449 may be safer than earlier models.

  2. When choosing surge suppressors, do not purchase models with varistors that have minimal MCOV ratings (e.g., 130 V ac for applications on nominal 120 V ac service), not only because of the fire hazard, but also because of the difficulty of coordinating such surge suppressors with a surge arrester upstream. This recommendation translates into an avoidance of surge suppressors containing varistors that have a UL 1449 rating of 330 V. (Unfortunately, specifying a UL 1449 rating of 400 V or more does not eliminate all varistors with an MCOV of 130 V ac, as varistors with low MCOV values can have high voltage protection levels during a surge if the varistors have excessive lead length.)

  3. It is possible that some protection from ignition or melting of the plastic enclosure of a failing surge suppressor could be obtained by placing the surge suppressor in a heat-resistant glass container (e.g., baking dish) away from flammable objects (e.g., paper, curtains, etc.).

  4. Surge suppressors and surge arresters may conduct very large surge currents (e.g., from a lightning strike), which poses the risk of explosion. My personal practice, since I began designing surge suppressors in the mid-1970s, is to put varistors in a metal enclosure, preferably steel, with plenty of open space surrounding the varistor inside the enclosure, to contain any fire or explosion.

  5. Install a surge arrester inside the circuit breaker panel, upstream from all surge suppressors inside the building.
There is no guarantee that following these suggestions will prevent fires, nevertheless these suggestions seem reasonable to me.

surge suppressors hidden in walls

In the late 1980s, several manufacturers of electrical receptacles (e.g., Leviton, Pass & Seymour) began producing receptacles containing internal metal-oxide varistors for surge suppression.

I am particularly concerned about varistors inside such wall outlets, because many of these wall outlets are connected inside a plastic box that is nailed to a wooden beam. If the varistor burns, it might ignite the wooden beams that are hidden inside a wall.

In contrast, if a surge suppressor inside an outlet strip begins to burn and people are present in the room, people might notice the smoke and either unplug the surge suppressor or spray a fire extinguisher on it. Such quick intervention is not possible for surge suppressors that are concealed inside walls.

I am not aware of any fires caused by varistors inside electrical receptacles, but I do want to call attention to the possibility of such fires, so that fire investigators can consider this possibility.


During the 1980s and during most of the 1990s, surge suppressors were sold in the USA with very little engineering design. Many of these suppressors were nothing more than a varistor inside an outlet strip or other enclosure, with no testing for coordination with arresters and no testing for fire hazards during temporary overvoltages, such as would result from disconnection of the neutral wire in 120/240 Vac electric systems that are common in offices and residences in the USA.

As a consequence of the lack of careful engineering design and testing, at least tens of millions of surge suppressors sold in the USA prior to 1998-99 are hazardous.

In my opinion, many surge suppressors sold in the USA before 1998-99 have a design defect, as that term is used in products liability law. The most common specific design defects are both:
  1. use of varistors with a MCOV rating of 130 V ac, and
  2. no thermal disconnector when the varistor is inside a plastic enclosure.

My opinion is not the result of applying knowledge in the late 1990s to products designed and manufactured during the 1980s and early 1990s. As mentioned above, General Electric took appropriate steps in the design of its surge suppressors in the late 1970s. And, as mentioned below, there are several archival papers published in proceedings of international engineering symposia from 1989 to 1992 that discussed problems with MCOV ratings that were too low or the lack of a thermal disconnector.

Furthermore, some manufacturers and importers of surge suppressors were well aware of the fire hazards in their products, because of consumer complaints and warranty claims to those manufacturers and importers. Yet these manufacturers continued to sell products that I believe were defectively designed, and these manufacturers also resisted efforts to include safety tests in ANSI/IEEE engineering standards.

Annotated Bibliography

David Birrell and Ronald B. Standler, "Failures of Surge Arresters on Low-Voltage Mains," IEEE Transactions on Power Delivery, vol. 8, pp. 156-162, January 1993.
In the late 1980s, General Electric became aware that a few (e.g., less than 0.02%) of their surge arresters were exploding. This paper is apparently the first publication in the peer-reviewed, archival engineering literature that specifically discusses fire or explosion hazards of surge-protective devices for use on 120 V ac power systems. Dr. Peter Hasse wrote a comment to the Birrell/Standler paper in which he mentions that surge arresters manufactured by Dehn + Söhne in Germany have contained an internal thermal disconnector since the late 1950s.

K. Eda, "Destruction Mechanism of ZnO Varistors Due to High Currents," Journal of Applied Physics, Vol. 56, pp. 2948-2955, Nov 1984.
Review of failure mechanisms of varistors.

François D. Martzloff and Thomas F. Leedy, "Selecting Varistor Clamping Voltage: Lower is Not Better!" Eighth International Zürich Electromagnetic Compatibility Symposium, pp. 137-142, March 1989.
First published criticism of choosing the lowest possible MCOV rating for varistors. Martzloff and Leedy gave no numerical recommendations in their paper, and do not mention fire hazards.

Viktor Scuka, EMI Control in Low-Voltage Power Installations, Seventh International Zürich Symposium on EMC, paper 79M4, March 1987.
This is apparently the first publication in archival engineering literature that a surge arrester should have a lower voltage protection level than the downstream surge suppressors.

Steve B. Smith and Ronald B. Standler, "The Effects of Surges on Electronic Appliances," IEEE Transactions on Power Delivery, vol. 7, pp. 1275-1281, July 1992.
This paper is apparently the first publication in the peer-reviewed, archival engineering literature that specifically discusses the ability of unprotected electronic equipment to survive surges in a laboratory. Surprisingly, consumer electronic equipment was able to survive surges with peak voltages of 2000 V, which suggests that it is not necessary to have the lowest voltage protection level from a surge suppressor.

Ronald B. Standler, Protection of Electronic Circuits from Overvoltage, Wiley-Interscience, 434 pp., 1989.
Standler, on pages 290-291, suggests that the minimum MCOV rating of a varistor be at least 1.25 times the nominal system voltage. Standler's recommendation translates to a MCOV rating of at least 150 V ac for service at 120 V ac.

Ronald B. Standler, "Use of a Metal-Oxide Varistor with a Series Spark Gap Across the Mains," IEEE International Symposium on Electromagnetic Compatibility, Washington, DC, pp. 153-158, August 1990.
Mentions that the quest for a lower voltage protection level of surge suppressors "is a quest akin to seeking the Holy Grail." Mentions on page 155 the varistor failure mechanism of a linear V-I characteristic that is not commonly recognized.

Ronald B. Standler, "Coordination of Surge Arresters and Suppressors for Use on Low-Voltage Mains," Ninth International Zürich Symposium on EMC, pp. 517-524, March 1991.

Ronald B. Standler, "Calculations of Lightning Surge Currents Inside Buildings," IEEE International Symposium on Electromagnetic Compatibility, pp. 195-199, August 1992.

Ronald B. Standler, "Design and Performance of Surge Suppressors," IEEE International Symposium on Electromagnetic Compatibility, pp. 363-368, August 1993.
Criticizes common designs of commercially-available surge suppressors, discusses failure modes, leakage current to earth (which could be an electric shock hazard to people), and criticizes hyperbole and errors in manufacturers' specifications for surge suppressors.

Ronald B. Standler, Coordinated Electric Surge Suppressor with Means for Suppressing Oscillatory Transient Overvoltages, U.S. Patent 5,398,150,   14 March 1995.
Discloses design of an inexpensive surge suppressor that is well coordinated with an upstream surge arrester.

M.F. Stringfellow, "Fire Hazard of Surge Suppressors," Power Quality Conference, 1992.
Stringfellow found that fires in surge suppressors were commonly caused by the loss of a neutral wire in common center-tapped 120/240 V ac service. The resistance of the loads on the branch circuits cause the 240 V between the pair of utility conductors to be divided. The resistance (e.g., 30 ohms) of the loads on the branch circuits with lower voltage limits the current in the varistors on branch circuits with higher voltage. This current limiting produces a scenario in which varistors become hot enough to ignite the plastic housing of the surge suppressor, without becoming hot enough to destroy themselves (e.g., by a lead wire coming unsoldered from the side of the hot varistor) and without drawing enough current to trip circuit breakers upstream.

this essay is at
version 26 Feb 2002, links updated 25 March 2005

Standler's brief credentials in surge protection

Standler's homepage