An SPD installed on the high voltage side of a transformer sees mainly currents from direct lightning strikes coming in off the power lines. It is exactly that kind of "direct lightning" that the 10/350 waveform was meant to simulate and that spark gap SPDs were supposed to be capable of handling. So it seemed like a perfect application for spark gap SPDs and in fact it was promoted as such for decades. (IEC Publication 99-2 first "recommended spark gap lightning arrestors for international use" in April 1960.) Customers purchased literally tens of millions of spark gap protectors to protect transformers in this way. It made sense to do it if you could trust the hype that promised spark gaps would protect against direct lightning.
By the 1980s experience from the field told a different tale--a tale mirrored in the experience of low voltage applications. The slow response time, short operating life and high let through currents plus general electrical instability of the spark gaps were causing lots of trouble to transformer components. Moreover, because the operation of the spark gaps appeared as short circuits in the system, the spark gaps were tripping upstream circuit protection devices. The reduction in reliability this caused to electric utility networks could no longer be ignored. By the late 1980s it became obvious MOV arrestors were superior in many important ways and they started to be used in most new installations.
We asked Jonathan Woodworth to comment about these phenomena. Mr. Woodworth is one of the world's foremost authorities on transformer surge protection and has been involved in IEC and IEEE standards organizations for over 30 years. (You can see his website here: www.arresterworks.com.) Here's what he had to say.
Mr. Woodworth, we’ve read that back in the 1980s gapped arrestors were being broadly used to protect the line-side of pole-mounted transformers but they’ve now been mostly replaced by the MOV variety. Is that an urban legend, or is that really the case?
The legend is real. The generation of arrester just prior to the solid state MOV type arrester was referred to as a Gapped Silicon Carbide (SiC) arrester. The term silicon carbide is used because this is the material that acted like a current limiter once the gaps sparked over. I would say that the present ratio of MOV to Gapped Sic for station class arresters is 80 - 20. For the 100 million distribution arrestors in use in the USA, maybe 60% of the gapped arrestors have been replaced with solid-state MOV arrestors so far.
Why were they all replaced? Was there a problem with the spark gap arrestors?
This type of arrester allowed for let through current from the power system that degraded the arrester and reduced its life. This design also allowed more of the surge through to the protected equipment. The spike on the front of the outgoing surge was a more stressful hit on the protected transformer. There is no doubt that higher-level spikes on the ongoing surge were present from this generation of arresters. The first station class MOV type arrester hit the market in the late seventies, and the first distribution arresters hit the market in the early 80's. The last silicon carbide arrester produced in the US was in 1994. I was the engineering manager at Cooper Power Systems at the time and saw this happen.
From what you say, the MOV technology became the technology-of-choice for protecting distribution and station transformers because it provided better protection and had a longer lifetime than the SiCs. Are there any other reasons?
That is a good summary. However at the same time there was a school of thought that the spark gap arrester failures were due to not enough energy handling capability and it was thought that MOVs offered higher energy capability, so this was another reason for the universal acceptance of MOV.
François Martzloff has been a Fellow of the IEEE since 1983. At the end of 2012 he was awarded the IEEE Lifetime Achievement Award for "a lifetime of integrity, leadership, and mentorship in standards development for surge protective devices and power quality fostering technological innovation, excellence, and benefit to mankind." We asked him for any comments or observations he had about the operation of spark gap protectors: Here's what he had to say.
Mr. Martzloff, what is your experience in investigating damage to transformers and motors where spark gap protectors had been installed?
I’ve been involved in investigating failures of transformers and motors—"François has scope, will travel"—and sure enough in most cases I’ve found that they had spark gaps installed. If you had an MOV arrestor, you wouldn’t have that damage. The problem is the gap.
Was there some particular fault of the spark gaps that caused that damage?
There’s a fundamental, I guess it’s IEEE 301 or 401 or R&D in transformer manufacturing, that tells you when you have a steel core with a winding over it, because of the distribution of capacitance between windings between each turn and the grounded core, the voltage is not distributed evenly. The first few turns get easily 30-50% of the total voltage. So those turns are stressed more than the other turns. (Smart transformer designers recognize that and provide more insulation on the incoming part of the winding.) You have your transformer or motor, can be big or small. A relatively steep waveform would produce this relatively uneven distribution which would stress the first few turns. In order to reduce that hazard, you put a capacitor in parallel that makes sure that the rate of rise of the incoming surge is sloped off. That's a workable strategy to ensure you don’t get that uneven distribution.
There is some irony in attempting to improve the protection intended by a surge capacitor at the terminals of a motor or transformer by adding an arrester that includes a gap: the incoming surge, gently sloped by the combination of the series line impedance and the parallel capacitance, is then abruptly changed to a steep change (collapse) of the voltage, resulting in that potentially destructive uneven distribution of the stress among the first few turns of the winding. When the surge is big enough the gap fires and therefore the voltage collapses. The collapse of voltage is no different than a fast increase. It matters not that the effect is to reduce the voltage (collapse it), the change is abrupt – hence unevenly distributed. So putting the arrestor and it firing defeats the purpose of having capacitors sloping it off. In other words, with the arrestor and the capacitor next to it, you have built and installed next to the motor a fast-front surge generator. Whether it collapses or rises it’s still a difference of potential here and you fail the motor winding because you have put this fast surge generator (gap) right next to it. This explains many many device failures that would not have occurred with gapless surge protectors.
So you wouldn't recommend spark gaps be used in this type of installation?
Whether it’s a transformer, motor or other devices, a well-intentioned voltage clamping by a gap can have adverse effects on stress distribution. And that’s one more reason why getting rid of gapped arrestors was a good move. Every time the gap fires you are stressing the first turns of the motor windings.
Perhaps the electronics giant, ABB, says it best. ABB, global leader in power and automation technologies is based in Zurich, Switzerland employing 145,000 people and operating in over 100 countries. ABB says: When it comes to surge protection of MV installations, "The state of technological development today demands the use of metal oxide surge arrestors without spark gaps..."
Read about damage to LV systems