How to choose the right surge protective device?
How to choose the right surge protective device?
I still remember the day a single lightning strike cost my client 80,000 USD in PLC boards.
I pick SPDs by first mapping the network’s fault current and voltage withstand, then match a device whose Uc ≥ 1.2 × system voltage, In ≥ 5 kA, and Up ≤ 0.8 × equipment withstand.
Keep reading and I will show you the exact check-list I use so you never over-pay or under-protect again.
How to choose surge protection devices?
I narrow the field by asking three questions: What is the system voltage and earthing type? How much lightning exposure do the cables get? What is the weakest link inside the machine I want to save?
Once those numbers are on one page, the right SPD becomes obvious in less than five minutes.
I start every quote with a one-page “risk map”.
First, I write the system voltage at the top.
Then I add the cable length from the main breaker to the machine.
Long outdoor runs score high on my lightning scale.
After that, I open the machine manual and look for the impulse voltage rating.
Most PLCs list 1 kV, VFDs list 2 kV, and servo drives list 1.5 kV.
I circle the lowest number; that is my target protection level (Up).
Next, I pick the SPD class.
For a main panel I use Class I if the building has an external lightning rod; Class I can handle 25 kA 10/350 µs.
If there is no rod, Class II 8/20 µs is enough.
For sub-panels I stay with Class II, and at the socket level I add a Class III with Up < 0.8 kV.
I never mix brands in the same path; different clamping curves can shift the stress to the weaker unit.
Price comes last.
I ask the factory for two prices: standard 40 kA and heavy 80 kA.
If the gap is less than 4 % of the panel cost, I take the bigger unit.
The extra copper pays for itself the first time a storm rolls in.
Quick selection table
|
Location |
Class |
In (8/20) |
Uc (V) |
Up (kV) |
|
Main panel TN-S |
I/II |
60 kA |
275 |
1.5 |
|
Sub-panel |
II |
40 kA |
275 |
1.2 |
|
PLC rack |
III |
20 kA |
275 |
0.8 |
What surge protective devices should be chosen and where should they be installed?
I place a Class I or II unit at the main switchboard, a Class II at every sub-panel, and a Class III inside any cabinet that holds electronics worth more than 2 000 USD.
Cable length between stages must be longer than ten metres or I add a decoupling coil.
The main board is the easy part.
I bolt the SPD to the same brass bar as the utility meter.
I keep the lead length shorter than 30 cm total; every extra 10 cm adds 100 V to the clamping voltage.
If the bar is crowded, I bend the cable once and clamp it, but I never cut corners on wire gauge.
I use 16 mm² stranded copper because it fits the lug and keeps the temperature rise below 30 K at 100 kA.
Sub-panels need more thought.
I walk the plant and count the metal paths: cable trays, steel beams, and copper pipes.
If two panels share the same steel, I still give each its own SPD.
Shared paths do not share surges; they share potential, and that is what kills chips.
Inside the machine, I mount the smallest SPD on DIN rail right next to the 24 V supply.
I use a fused version so the maintenance guy can pull it with gloved hands.
I label the fuse “SPD ISOLATOR” in English and Spanish because our crews rotate.
Installation distance rule
|
Distance from main SPD |
Next SPD needed? |
Class |
|
< 10 m |
No |
– |
|
10–30 m |
Yes |
II |
|
> 30 m |
Yes |
II |
Coordination between the surge protective device and its disconnect circuit breaker
I size the breaker so its magnetic trip is above the SPD’s maximum follow current, but below the panel’s short-circuit rating.
That window is usually 2 kA to 6 kA, and a C-curve 32 A MCB lands right in the middle.
Follow current is the ugly twin of surge current.
After the SPD clamps, the mains voltage keeps feeding current through the device.
If that current is higher than the breaker will trip, the SPD catches fire.
I check the data sheet line called “Ifi” (follow current interrupt).
A good SPD lists 100 A.
I compare that to the breaker curve: a B-curve trips at 3 × In, a C-curve at 5–10 × In.
For a 32 A breaker, C-curve trips between 160 A and 320 A, so the SPD is safe.
I also watch the short-circuit rating.
In Germany we see 50 kA panels, in Vietnam 10 kA.
I pick a breaker whose Icu exceeds the local utility number.
If the SPD is rated 25 kA but the breaker is only 10 kA, the breaker becomes the fuse.
That is backwards logic, so I upgrade the breaker first.
Breaker–SPD match table
|
Breaker type |
Rating |
Magnetic trip |
SPD Ifi |
Match? |
|
B32 |
32 A |
96 A |
100 A |
No |
|
C32 |
32 A |
160 A |
100 A |
Yes |
|
D20 |
20 A |
200 A |
100 A |
Yes |
Coordination between the surge protective device and its disconnect circuit breaker in the event of a short circuit
I run a 50 kA bolted-fault test in ETAP and confirm the breaker will clear before the SPD housing reaches 200 °C.
If the software flags red, I upsize to a 50 A C-curve or add a 50 A gG fuse in series.
The worst case is not lightning; it is a dead short across the SPD terminals.
Copper vapour rises in less than 3 ms.
I need the magnetic trip to open in under 1 ms.
That means the fault current must be at least 5 × the magnetic setting.
In a 50 kA panel, a C32 sees 1 560 A, so the ratio is 1 560 / 160 = 9.7.
The breaker will trip in 0.01 s, well before thermal runaway.
If the utility impedance is low, the fault can hit 35 kA.
I check the breaker let-through energy (I²t).
The SPD must survive that energy.
I compare the breaker I²t at 50 kA with the SPD pre-arcing I²t.
When the breaker is larger, I am safe.
When the SPD is larger, I add a fuse.
Coordination between the surge protective device and its disconnect fuse in the event of a short circuit
I pick a gG fuse one size above the MCB rating so the fuse only steps in when the breaker fails.
A 40 A gG fuse lets a C32 breaker clear first, but still opens in 2 ms if the fault current tops 20 kA.
Fuses are my backup parachute.
They add cost, so I only use them when the fault current is above the breaker’s Icu.
In Italy we see 40 kA networks.
A 32 A breaker with 10 kA Icu is a toy there.
I add a 50 A gG fuse in series; its 50 kA breaking capacity covers the gap.
The fuse also protects the SPD if the breaker welds.
I mount the fuse in the same DIN housing so the tech needs only one spare.
Heat is the hidden enemy.
A fuse runs 1.6 W at rated current.
I derate 20 % in panels above 40 °C.
That means a 50 A fuse is good for 40 A continuous.
I note that on the panel door so no one swaps in a 63 A fuse during a midnight panic.
Conclusion
Map the risk, match the numbers, and let the breaker lead the fuse.
Do that once, and every storm will pass without a single fried board.