Installation Mistakes I Have Seen
Ten field situations where the devices were fully compliant and the system was still destroyed by lightning. And the pattern running through all of them.
The most frustrating failures in technical support all share one feature: the SPDs themselves were completely compliant, correctly specified, with full test reports — and then they were installed incorrectly.
The equipment was still destroyed. The customer's first reaction is that "the SPD quality was poor". But when you open up the site, the problem almost never lies in the device.
1. The Type 3 was installed in the distribution cabinet
A data centre project, equipment destroyed by a strike. Inspection found: a Type 2 in the rack, a Type 1 at the room entry — and nothing at the equipment inlet.
The engineer's explanation: "Type 3 has too high a voltage, connecting it might affect the equipment." A completely standard misunderstanding, and precisely backwards. Type 3 is the class with the lowest voltage protection level. It exists precisely to clamp down what the upstream stages leave behind.
A Type 3 typically achieves a residual of a few hundred volts or less, while a Type 2 is often above 400 V. Fitting the Type 3 substantially reduces the voltage reaching the equipment.
2. The PE lead ran half way across the factory
Equipment in Building A, distribution panel in Building B, forty metres apart. The SPD was correct and correctly specified.
But the PE lead from the SPD ran over thirty metres with several bends. During a strike, the induced voltage on that thirty-metre lead is added to the earth potential of the protected equipment.
Those thirty metres can contribute more voltage than the SPD's Up. The device did nothing.
3. Supply and load routed separately
Photographs from the site show: the incomer entering from the left of the cabinet, feeding the SPD, with the load cable leaving from the right via a separate tray, joined by a short flexible jumper.
The surge current therefore never followed the SPD's path — it travelled straight down the load cable into the equipment.
4. Type 1 and Type 2 mounted in the same cabinet, 20 cm apart
A pair of SPDs stacked vertically, under 20 cm apart. The supplier's reasoning: "better coordination that way."
In practice, the upstream device never has time to establish its discharge before the downstream one breaks down. The result: the Type 2 fails repeatedly while the Type 1 rarely operates.
This is very likely the answer to most "why does our Type 2 keep failing" questions.
5. Power-side protection only, across the whole system
A manufacturer whose server room, PLC control system, and ERP servers were all destroyed. Inspection found: Type 2 on the distribution board and not one piece of protection on a single communication line.
RS485 interface boards destroyed, Ethernet cards burned, PLC digital I/O modules all damaged.
6. The SPD was removed after the strike
The most baffling of all. After the first strike destroyed equipment, the site response was: the SPD was removed, on the grounds that "after a strike it fails and causes a short circuit that disrupts power".
The project then ran for two more years without SPDs, and lost equipment again.
The risk of an SPD failing short is controllable — through fused types, or Type 3 devices with replaceable modules. But that has to be thought about during selection, not after the incident.
7. Everything connected to the neutral
A residential installation by the owner: an SPD at the incomer, with the neutral connected to one of the SPD's ports, so the L–N surge was handled by the device.
The correct connection is phase-to-earth (an N-PE type). Lightning surges are common-mode — relative to earth — not differential. A device connected across L–N does not address the actual threat.
8. PV frames with no equipotential bonding
A 2 MW rooftop PV plant. A direct strike destroyed more than thirty modules. Investigation showed no equipotential bonding between the module frames, so the current wandered through the array and concentrated on the lowest-impedance path, burning that section.
There was interception and down-conduction, but no equipotential bonding — the current was brought in, but had no defined path out.
9. Uc selected too low
A Type 2 with Uc = 230 V, installed on a 230 V system. The sales rationale: "it gives a better-looking Up".
A Uc of 230 V means that as soon as the system voltage rises to 242 V through normal tolerance, the device exceeds its continuous working voltage. Persistent leakage, heating, degradation, failure.
The project ran for less than a year and every SPD failed. The owner's conclusion was that "SPDs just don't last".
10. Acceptance checked that they were fitted, not that they were right
The last category, and by far the most common.
Tick boxes: ☑ SPD installed. Model, quantity, position, lead length, earthing, stage separation, indicator visibility — none of it checked.
This checklist is the highest-leverage item in the entire lightning protection system. It costs essentially nothing, and it would have caught nine of the ten items above.
The pattern
Put the ten side by side and one pattern is unmistakable:
Devices remove energy. Equipotential bonding removes potential difference. Installation connects the path.
Three separate jobs, and missing any one of them renders compliant devices useless.
Nearly every field problem lives in the second and third. Device parameters can be verified against a datasheet; installation has to be looked at.
So a suggestion that is not very fashionable but very true: at acceptance, go and look at lead lengths and terminations. That is worth more than reading the datasheet again.
A digression
None of this is written to blame anyone. The overwhelming majority of site installers followed the drawings, and following drawings is not a failure. The problem is that the drawings did not say it clearly enough.
So my other suggestion is: when you produce the drawings, write "PE lead not to exceed X metres", "supply and load must share the same route", "minimum stage separation X metres" directly onto the drawing. Clauses buried in a specification get ignored; notes on the drawing get built.
A designer who omits one sentence may cost a site one piece of equipment.
If you have a specific field situation to discuss — damage symptoms, installation configuration, or a rating you are unsure about — get in touch.
