Archive

Posts Tagged ‘metal oxide varistors’

Surge Suppressor Fires, Circuit Breakers, and Varistor Conduction

May 29th, 2013 2 comments

A customer called MIDWEST because some of their surge suppressor strips had caught fire underneath office workers’ desks.

The surge suppressors had integral circuit breakers in them for over current protection. However, despite the smoke and fire, the internal circuit breakers never tripped. The surge suppressors were still supplying power to the workers’ desktop computers.

Carefully taking apart the surge suppressors, MIDWEST discovered that extensive damage had been done inside the strip. The actual electronic component inside the strips that is meant to mitigate high voltage line spikes is called a metal oxide varistor. Most of these had burned up inside the strips. In addition, the heating caused massive damage to the printed circuit boards. The metal oxide varistors had greatly overheated, and both the varistors and the circuit boards had mostly been baked into carbon. This carbon was not a perfect short circuit, but had sufficient impedance to limit the current flow below the 15 amperes of the circuit breakers. But it had sufficient conductivity to generate large amounts of heat inside the strips.

From this and other evidence, MIDWEST determined the varistors had suffered an overvoltage condition, probably twice or three times the normal line voltage of 120 volts. This caused the varistors to conduct slightly, and dissipate power. It is very roughly estimated that the component dissipated between 5 and 20 times its nominal rated power dissipation of one watt, or 5 to 20 watts.

Experience has shown that power dissipations of less than roughly 5 times rated do not cause catastrophic damage. Above 5 times rated, and components begin to release the “magic smoke” that manufacturers seem to build into electrical components. Dissipations over about 20 times rated tend to cause violent disintegration of the component, accompanied by immediate flames, possible arc flash, and most noticeably, a very loud acoustic signature.

The varistors have an estimated 200 – 300 Volts RMS on them; therefore conduction currents were probably between 10 milliamperes to 100 milliamperes. Obviously, the 15 ampere circuit breakers would not trip from this conduction current. The overvoltage and the heating probably lasted several minutes.

It might seem strange to realize that the varistors never experienced any high voltage spikes, which is their usual reason for being. A sufficiently high voltage spike normally causes the varistors to explode with the sound of a gunshot, and vaporized zinc oxide and copper becomes spalled in the area around the varistor. That is, the internal bulk varistor material and copper electrodes vaporize in an arc flash, and deposit themselves in tiny balls on adjoining surfaces.

But no such spalling or deposited material was discovered upon disassembly of the melted plastic housings. Instead, the 5 to 20 watts dissipated in the 1 watt body of the varistor simply cooked itself, and everything around it into carbon. Then the carbon supported current flow.

In such a situation, surge suppressors have resulted in a number of fires throughout the US.

Fire Hazards of Surge Suppressor Strips with Integral Circuit Breakers

September 27th, 2010 2 comments

MIDWEST’S Engineering Department has had several customer’s ask to investigate some fires that have occurred in their offices.  These fires were caused by commercial surge suppressor strips.  Each strip had integral circuit breakers.  In each case, neither the integral circuit breakers nor the distribution panel’s circuit breakers tripped.  

 

National Fire Protection Association (NFPA) has researched and documented innumerable cases where the surge suppressors have caught fire.  In many cases, large fires have resulted; thus the NFPA’s concern.  The surge suppressor strips typically contain a circuit breaker and three or more metal oxide varistors.  Metal oxide varistors are non-linear circuit components that conduct large currents when the applied voltage exceeds a certain value, usually 130 Volts AC RMS.  

 

In the intended application, the varistors conduct in the presence of large voltage transients, such as from a lightening strike.  But a large transient of Kilovolts is not the only situation that can cause the varistor to conduct.  A small overvoltage can also cause conduction.  Of course, small is relative.  In this context, small might be a voltage increase such that the applied voltage is double or triple the normal 120 Volts, and can exist for hours.  Such events can result from wiring failures in the power distribution system.  In this case, the varistors will conduct small currents, perhaps just a few amperes.  However, the double or triple voltage will still be dropped across the varistor.   So, since power P = E * I, it is obvious that hundreds of watts will be dissipated as long as the overvoltage exists.  Since varistors are typically only rated for about 1 watt continuous, severe overheating will occur.  Such severe heat generation will result in destruction of the varistor and cooking of circuit boards.  These elements are often reduced to just carbon, which continues to conduct, and continues the generation of heat inside the surge suppressor.  The result can be a fire.  Undiscovered, the building can burn down.

 

It is salient to note that if just a few amperes flow, then neither the internal circuit breaker nor external circuit breakers will trip, as these are probably rated at 15 or 20 amps.

 

Underwriter’s Laboratories has tried to address this issue in the following revisions of UL 1449:

  • 2nd revision of UL 1449, titled ““UL Standard for Safety for Transient Voltage Surge Suppressors”.  Compliance with this revision is mandated as of February 2007.
  • 3rd revision of UL 1449, titled “UL Standard for Safety for Surge Protective Devices”.  Compliance with this revision is mandated as of September 2009.   This revision has also become an ANSII standard.

Surge Suppressors, Circuit Breakers, and Red Herrings

September 20th, 2010 No comments

A large municipality had a very frightening situation recently.  It seems that a dozen surge suppressor strips (with internal circuit breakers) underneath office workers’ desks had overheated to the point of starting small fires internal to the strips.

 

MIDWEST was contracted to solve the problem.  MIDWEST’s Engineering Department installed Dranetz line monitoring equipment (Dranetz is the gold standard of line monitors). 

 

The results revealed a very, very interesting situation.  The line monitoring equipment discovered two anomalies. 

 

First, the 120 volt line was continuously about 8% too high;utilities guarantee a high line tolerance of 5%.  Thus the customer’s line voltage was 3% higher than the utility allows.

 

Secondly, some large short term voltages were measured on the customer’s neutral to ground by the Dranetz.   

 

In addition, forensic analysis was performed on the surge suppressors.  It was found that in all instances, the internal voltage spike protecting component, the metal oxide varistors (MOV) had burned themselves up, due to long term overheating.  In doing so, much of the underlying printed circuit board was also burned up and carbonized.  This carbonized material continued to support a small current and generate heating in the strips.  However, the internal circuit breakers never tripped.  The fact that the circuit breakers never opened is a very critical clue.

 

At first glance, the “obvious” conclusion one could draw is that the 8% high line condition caused the varistors to conduct, causing the overheating.   

 

But, the obvious conclusion would be wrong.  The 8% high line condition is just a red herring in this investigation.  Referring to the V-I characteristic curves of the varistors, an 8% overvoltage would cause negligible current flow.  In order for the 1 watt rating of the varistor to be exceeded, double or triple the nominal voltage would have to be applied for at least several minutes; the varistors would conduct around an ampere or less.   This is far less than the 15 ampere rated current of the circuit breaker.   

 

The surge suppressor strips’ melted plastic housings, and the extensive baking of the printed circuit boards to carbonization indicated heating on the order of minutes.  

 

On the other hand, a typical large voltage spike of several thousand volts tens lasting tens of microseconds tend to explode varistors, creating a plasma ball, and spraying superheated varistor material and copper everywhere around the sacrificial varistor.   The plasma ball would act like a dead short circuit, causing an internal arc flash of hundreds of amperes.  This would have tripped the circuit breakers.  A typical varistor explosion usually sounds like a gunshot; however, no such noises were heard by personnel.  The author has personally witnessed these explosions during destructive testing of varistors under controlled conditions.  The surge suppressor post-mortems’ also revealed no deposited spray of bulk varistor material or of copper.

 

The only conclusion that could be drawn is that double or triple the nominal line voltage actually was existent for several minutes, rather than a one-time voltage spike.  This conclusion is supported by the anomalous Dranetz neutral to ground measurements.

Poor Connections Nearly Start Fires

September 14th, 2010 No comments

A recent customer called MIDWEST, saying that they had a dozen surge suppressor strips that almost caught on fire.  The strips had been powering computers for a large municipality.  Much to the consternation of the secretaries, the strips started smoking ominously right next to their feet under their desks.  These were high quality surge suppressors, each of which contained  circuit breakers, and three metal oxide varistors.     Very strangely, neither the circuit breakers inside the surge suppressors nor the panelboards’ circuit breakers tripped.  The computers that were plugged into the surge suppressors just continued to crank away, despite the smoke and incipient fires.

 

MIDWEST engineers dissected the burned surge suppressor strips, and found that the metal oxide varistors had catastrophically overheated due to a long term overvoltage condition, rather than a one-time voltage spike. This long term overheating baked the printed circuit boards. The boards had carbonized, thus supporting extraneous current flow and copious quantities of heat and smoke.  The current was below the trip level of all the circuit breakers, so they had not opened and not interrupted the current flow.

 

MIDWEST’s Engineering Department installed line voltage monitoring equipment at the main distribution panel’s circuit breakers.  After a couple of weeks, sufficient data was collected that indicated the probable cause was failures in the grounding and neutral connections relating to the 300 KVA distribution transformer.    

 

MIDWEST’s Engineering Department’s decades of experience and proven expertise in power electronics was essential in diagnosing the problem.