"Surge Protection Devices
No! You did not read this incorrectly. These were the words of a highly influential electrical engineer at a conference during the question time of a speaker's presentation on lightning protection. His argument is the lightning wave-front is effectively instantaneous. This means, no matter how quickly the Surge Protection Device (SPD) responds to the threat, the high voltage gets through - so why install the SPD in the first place.
Using a timebase of one 50Hz cycle (20ms) as a reference, the waveform of a typical lightning strike can not be represented as anything more than an extremely fine perpendicular line on the graph i.e. a square front. This represents an infinite frequency meaning the offending voltage will instantly attain full potential.
Everything has a reaction time (delay till operation) so, if one adopts this argument it would mean that as the SPD has a delay in responding to the high-voltage threat, it will not be able to stop the high voltage getting through. It sounds like a reasonable argument, but it is horribly flawed.
If one "zooms in" on the lightning 'spike', the voltage build-up and decay is very traceable and is, by electronic standards, actually quite slow. The build-up to peak and half-peak decay timings from various industry accepted specifications are 5-20, 8-35, and 10-350 (all in microseconds).
Taking the shortest industry accepted time lightning takes to reach the peak current and therefore the peak offending voltage at 5µs, means the lightning waveform "front" is nowhere near "square". Actually, it has a significant delay of its own! This comes as no surprise as other research has shown the electromagnetic spectrum of a lightning strike centres around 10kHz.
The response times of good SPDs are around the 25ns mark. Weighing this against the 5µs attack time referred to above makes the response at least 200 times faster than the threat. Extrapolating this; If the lightning strike could induce 100kV, dividing this by 200 means the SPD will respond to the threat at a mere 500V above its trigger voltage.
I have seen some documentation claiming the voltage can peak in about 1.2µs based on the voltage being capacitively coupled (and, as such, leads the current by 90°), but in all my measurments with lightning (yes, I've actually had oscilloscopes connected to kit to record lightning and its effects) I have never seen slopes this fast. Just under 5µs, sure, but never as fast as 1.2µs!
One needs to remember that a conductor is an inductor. Inductors are "frequency dependent resistors". This means that the higher the frequency, the higher the 'resistance'. Taking this one step further; The response of the SPD will effectively generate at least a 20MHz waveform. With such a high frequency component, it will not take much inductance (in the form of conductor length) to diminish this to a level whereby the device it is protecting to not even be concerned about it, never mind dodging any damage.
This is said within reason as SPDs are not perfect and while attempting to clamp the voltage to an acceptable level they themselves "stretch" (in my lightning protection courses I usually say MOVs have "rubber ceilings"). A MOV as primary protection on a power circuit may, at best, clamp to 4kV under the full whack of a lightning strike, but that does not mean we should not install any at all.
This voltage that occurs as a result of response time even has a name! It's called the "Let Through Voltage" and it alone has been the basis of war amongst the various protection manufacturers. Some claim exceedingly low values, but then do look at the test currents and curves. Compare apples with apples, as the old saying goes.
This is very specifically said as I continually come across engineers educated (incorrectly, I might add) in lightning protection, rather than experienced in it - and they have been educated that the let through voltage is exceedingly high as opposed to what experience has shown me through measurement and experience.
"If you can't stand the heat..."
On the flip-side it is, sadly, a very misquoted fact that by simply installing impulse (transient) protection, one will save what is installed beyond it. As we have just shown, the primary SPD will at best, when under full impulse current conduction, limit the offending voltage to as high as 4kV on 230VAC circuits. There is little that will withstand such an onslaught for 350µs!
However, as we have also shown, letting this statement deter one from installing any form of transient protection is just as daft! It is knowing what to install and, more importantly, where to install it that is crucial. It therefore remains, unchallenged, that in less than a millisecond an unprotected circuit can suffer irrecoverable damage. It is very necessary to give it a fighting chance.
We have defined a spike and transient to be a short-duration electrical disturbance with high levels of voltage and an ability to deliver current - the duration being the main feature distinguishing the two. Transients will undoubtedly cause immediate, catastrophic damage to equipment. Spikes will leave things alone. However, there is a grey area in the middle that will not do immediate damage, but be the source of a continuous degradation of electronic systems and components... until they fail.
Transient protection, a means to clamp voltages when they exceed a predefined limit, is the obvious solution. This device's function is to offer the extraneous voltage a path away from a protected load - done by means of causing current to flow, usually to ground.
Such transient protection is really the only way to save equipment from damage, but low quality devices, or badly installed, will provide heartache. To ensure correct operation, the protection must be coupled to a superb earth and ground system as the current that has been diverted away from protected device needs to dissipate somewhere!
Furthermore, the secret is to use a 'staged' or cascaded approach. This involves starting at the service entrance with relatively high voltage high current devices, then using lower voltage lower current with each distribution point in the chain as one gets closer to the equipment.
It must not be forgotten that transients and (especially) spikes border on the realms of radio frequencies and therefore the principles. Cascading eliminates the high possibility of installing a single protection at a 'node' and having many thousands of volts still available at the equipment during a spike or transient - the multi-stage protection is likely to avoid this happening. By the way, you cannot have the stages next to one another. They must be spread over the premesis.
This method can be adopted from simple single-phase domestic properties, right through to major multi-storey buildings (and sites too). Even in our home I have a three-stage approach. A 3-way 1000V protection at the incomer, 3-way 600V protection at the distribution board, and then 3-way 375V protection at each plug point to each piece of electronic kit (from washing machine to hi-fi). Having had my radio mast struck three times with no damage suffered proves the effectiveness of this approach.
Before I get on to promoting some solutions, there is one thing I would like to cover that "shivers me timbers". It appears to be common practice to put a fuse in line with the protection. This is a disaster waiting for a time to happen. Don't do it. If the protection fails, taking out the fuse, then it may take a catastrophic failure to realize this has happened.
A far safer approach, if it is felt there must be some means of isolating the protection, would be to use a breaker such that should the protection become faulty then there is a physical indication of this, rather than relying on inspecting the health of a set of fuses during maintenance.