Key takeaways
- Clearance and creepage are two different numbers and both must be satisfied. Clearance is the shortest path through air, set by the impulse voltage. Creepage is the shortest path along a surface, set by the working voltage, the pollution degree and the CTI material group.
- IEC 60664-1 is a horizontal base standard. Your product standard (IEC 61439, IEC 62368-1, IEC 61010-1, UL 508A) applies it and sets the final number — start from the product standard, not from the base standard.
- Paint, powder coat and anodising do not count as insulation. Measure the path as if the metal were bare, or add a rated barrier.
- Altitude correction starts at 2 000 m. A BESS container at 4 000 m needs roughly 29% more clearance than the table value.
- Sheet metal has six free or cheap levers to buy spacing — standoff height, slot, rib, insulating barrier, sleeve, and re-segregation. All six cost pennies at the design stage and a re-tool at the tooling stage.
- Put the distance on the drawing as a dimensioned, toleranced feature. "Verify creepage at assembly" is not a specification.
Clearance and creepage are the two distances that decide whether a sheet metal enclosure passes its dielectric test or fails it in front of the certifier. In March, a European battery-storage integrator sent us a 1,500 V DC combiner cabinet that had cleared its own design review and then failed factory hipot on 11 of 40 units — at 3.2 kV against a required 4.0 kV. The electronics were fine. The busbar was fine. The failure was a 6 mm gap between a busbar support standoff and a formed rib on the side wall, across a powder-coated subplate. The drawing had never dimensioned that gap, so nobody owned it.
What follows is the guide we wish that customer had read before the first article was cut. It covers what clearance and creepage actually are, the four inputs that set every number in the standard, how to read the two tables you will actually use, why altitude matters more than most OEMs think, six ways to buy spacing in sheet metal without making the box bigger, and a 9-point DFM checklist we run on every power-rated enclosure before it goes to laser. All values are given as a practical subset — always verify against the tables in your own edition of IEC 60664-1 (current edition 3.0, IEC 60664-1:2020, with Amendment 1 published in 2025) and against the product standard that governs your assembly.
Why Spacing Is a Sheet Metal Problem, Not Just a PCB Problem
Almost every article written about clearance and creepage is written for printed-circuit-board designers. That is understandable — the standard's tables are most often applied to a board layout. But in a power enclosure, the longest and most dangerous creepage paths are rarely on the board. They run from a busbar to a painted wall, from a terminal block to a DIN rail, from a cable lug to a hinge, from a subplate standoff to a formed rib.
Those paths are created by metal geometry: how far a standoff lifts the busbar off the subplate, how deep a slot is cut, how tall a rib is formed, where a barrier is riveted, whether a coating is masked off. That means they are fixed at the moment the flat pattern is released — and they are essentially free to change at that moment. Once the tooling is cut and 40 cabinets are built, moving a standoff by 8 mm costs a re-tool, a re-test and a scrap pile.
Our 12-year quote log shows the same pattern every quarter: the spacing decision is pushed to the panel builder, the panel builder assumes the enclosure was designed for it, and the finding lands in the factory acceptance test. The three fixes that show up most often in our logs are raising a standoff, adding a slot, and adding a GPO-3 barrier — in that order of cost.
Clearance vs Creepage: Two Distances, Two Failure Modes
The two distances are measured between the same two conductive parts, and they fail in completely different ways. Confusing them is the single most common reason a design passes review and fails test.
| Attribute | Clearance | Creepage |
|---|---|---|
| Path measured | Shortest straight line through air | Shortest path along the surface of the insulating material |
| Sized by | Rated impulse withstand voltage (peak) | RMS or DC working voltage |
| Other inputs | Field shape (homogeneous vs inhomogeneous), altitude | Pollution degree, CTI material group |
| Failure mode | Instantaneous air breakdown — a flashover in microseconds | Slow tracking — a conductive carbon path builds over months or years |
| Typical test | Impulse / hipot with the rated impulse voltage | Tracking test (IEC 60112 for CTI) plus dimensional verification |
| Metal levers | Standoff height, barrier, re-routing, removing sharp edges | Slot, rib, groove, insulating barrier, insulating sleeve |
| 230/400 V, OVC III, PD 2, mat. grp II | ≈ 3.0 mm line-to-line and line-to-earth | Looked up from working voltage, pollution degree and CTI material group — no line-to-line / line-to-earth split; see the current edition of IEC 60664-1 |
Two consequences follow. First, one is often larger than the other, and you must satisfy the larger — you cannot average them or pick the convenient one. Second, a feature that helps one may not help the other. A rib across a creepage path lengthens the surface route but does nothing for the air gap; a taller standoff increases the air gap but, once the standoff is metal, may create a new creepage route over its own surface.
The Four Inputs That Set Every Number
Every value in every table of IEC 60664-1 is derived from four inputs. Get these four right on the drawing and the rest is arithmetic.
- Working voltage — the highest RMS (AC) or DC voltage across the insulation in normal operation, including recurring peak voltages. For a 400 V three-phase system, that is 400 V line-to-line, not 230 V.
- Overvoltage category (OVC I–IV) — where the equipment sits in the installation, which determines the transient it must survive. This sets the clearance.
- Pollution degree (PD 1–4) — how much conductive contamination and condensation the insulation surface sees. This sets the creepage, and it does not affect clearance at all.
- Material group (I, II, IIIa, IIIb) — the comparative tracking index (CTI) of the insulating surface the path crosses. This also sets the creepage.
Altitude is a fifth input that modifies clearance above 2 000 m, and frequency above 30 kHz takes you outside the scope of the standard entirely — relevant for the SiC and GaN stages inside modern EV chargers and storage inverters, where the internal converter is usually separately certified and the enclosure only carries the AC and DC distribution.
Overvoltage Category → Rated Impulse Voltage
The overvoltage category ranks a point in the installation by how much transient energy is available there. Category I is protected electronics; Category II is household appliances and portable tools; Category III is fixed industrial equipment — switchgear, motors, distribution boards, and most industrial control panels; Category IV is the origin of the installation, the service entrance and utility metering.
Cross-reference the category with the nominal line-to-neutral voltage of the supply and you get the rated impulse voltage. The subset below covers the cases we see on OEM drawings (values in volts peak, from IEC 60664-1 Table F.1):
| Nominal voltage line-to-neutral (V rms) | OVC I | OVC II | OVC III | OVC IV |
|---|---|---|---|---|
| ≤ 50 V | 330 | 500 | 800 | 1,500 |
| ≤ 150 V (e.g. 120 V US) | 800 | 1,500 | 2,500 | 4,000 |
| ≤ 300 V (e.g. 230 V EU/CN) | 1,500 | 2,500 | 4,000 | 6,000 |
| ≤ 600 V (industrial motors, ship power) | 2,500 | 4,000 | 6,000 | 8,000 |
So the standard European industrial panel — 230/400 V, fixed installation, OVC III — carries a rated impulse voltage of 4,000 V. That single number is what sizes every clearance inside the box. A US 120 V panel at OVC II carries 1,500 V. A 480 V motor circuit at OVC III carries 6,000 V, which is why motor-terminal compartments need visibly more room than the control compartment next door.
One trap: if you need reinforced insulation rather than basic, most standards outside IEC 62368-1 require you to step up one level in this table to find the impulse voltage you must design to — 4 kV becomes 6 kV, not 8 kV. Doubling applies to creepage, not to impulse voltage.
Pollution Degree and the Micro-Environment
Pollution degree classifies the contamination and condensation that the insulating surface is exposed to. It affects creepage only.
| PD | Definition | Typical location | Effect on creepage |
|---|---|---|---|
| 1 | No pollution, or only dry non-conductive pollution; sealed or hermetic | Sealed components, conformally coated assemblies | Shortest path; material group does not matter |
| 2 | Only non-conductive pollution; occasional temporary conductivity from condensation | Offices, labs, most enclosed industrial panels | The default assumption for a well-sealed indoor cabinet |
| 3 | Conductive pollution, or dry non-conductive pollution that becomes conductive when wet | Industrial plants with metal dust, unsealed outdoor cabinets, coastal sites with salt | Materially longer; must be looked up, not scaled |
| 4 | Persistent conductivity from rain, snow or conductive dust | Exposed outdoor equipment, washdown zones | Longest; usually demands a barrier or a sealed sub-compartment |
The useful engineering idea here is the micro-environment. The pollution degree that matters is the one immediately around the insulation — not the one outside the factory wall. A properly sealed IP65 enclosure standing in a steel plant (macro-environment PD 3) can legitimately be treated as PD 2 inside, provided the sealing, the gasket and the cable entries actually hold. That is exactly where our gasket sealing guide and our condensation control guide stop being housekeeping and start being compliance: a failed gasket does not just let water in, it silently reclassifies your creepage requirement.
The reverse is also true. An outdoor cabinet with an unsealed bottom gland plate, a field-drilled vent and a weep hole is PD 3 or PD 4 on the inside regardless of what the IP plate says — and the creepage table does not care what the plate says either.
Material Group and CTI — Including the One Everybody Gets Wrong
The comparative tracking index measures how well an insulating surface resists forming a conductive track under contamination and voltage. It is measured by IEC 60112 and sorted into four groups.
| Group | CTI range | Tracking resistance | Typical materials |
|---|---|---|---|
| I | ≥ 600 | Highest — shortest permitted creepage | Ceramics, GPO-3, some PBT and PA grades, mica |
| II | 400 – 599 | High | PA66-GF, many PBT grades, high-grade phenolics |
| IIIa | 175 – 399 | Moderate | Standard FR-4 laminate, many PC and ABS grades |
| IIIb | 100 – 174 | Lowest — longest required creepage | Low-cost phenolics, some recycled compounds |
Here is the mistake we see on roughly one drawing in four: bare or coated metal has no material group. Steel, aluminium and their coatings are conductors for this purpose — powder coat, anodising and paint are not credited as solid insulation by IEC 60664-1 because the film is thin, porous and easily scratched by a busbar during assembly. So a creepage path that runs over a painted subplate must be measured as though the plate were bare, which in practice means one of two things: the path is measured all the way around the obstacle, or you insert a real barrier with a declared material group and thickness.
That is precisely what happened in the 1,500 V DC cabinet. The designer assumed the powder-coated subplate gave him an insulating surface with a decent CTI. It gave him nothing. The fix was a 1.5 mm GPO-3 barrier (material group I) plus a standoff increase — $3.20 per unit, and a two-week schedule hit that could have been avoided with one dimension on a drawing.
The Creepage Table You Will Actually Use
With the working voltage, the pollution degree and the material group in hand, the creepage distance follows directly. The subset below is the one our customers ask for most (millimetres, basic insulation, from IEC 60664-1 Table F.4):
| Working voltage (V rms or DC) | PD 1, all groups | PD 2, group I | PD 2, group II | PD 2, group III |
|---|---|---|---|---|
| 63 | 0.20 | 0.63 | 0.90 | 1.25 |
| 400 | 1.00 | 2.00 | 2.80 | 4.00 |
| 800 | 2.40 | 4.00 | 5.60 | 8.00 |
| 1,000 | 3.20 | 5.00 | 7.10 | 10.00 |
Three working notes. Reinforced insulation doubles these figures — 400 V at PD 2 / group II becomes 5.6 mm, not 2.8 mm. Pollution degree 3 is tabulated separately in the same standard; do not scale the PD 2 column by eye, look it up. And DC is not automatically easier: a 1,000 V DC battery string uses the 1,000 V row, and because DC never crosses zero, any arc that starts does not self-extinguish — which is why our BESS customers generally design one material group better than the minimum.
Reading the table for our opening story: a 1,500 V DC busbar in a sealed cabinet at PD 2 with a GPO-3 barrier (group I) needs more than the 10 mm shown for 1,000 V — the row above 1,000 V, and the barrier has to extend past the busbar on both sides by at least the same distance, otherwise the path simply goes around the end of it.
Clearance: Impulse Voltage, Field Shape and Altitude
Clearance comes from the rated impulse voltage, with two modifiers. The first is field shape: a smooth, parallel geometry (homogeneous field) tolerates a much smaller gap than a pointed or edge-on geometry (inhomogeneous field), where the field concentrates at the tip. At 2.5 kV the difference is 0.6 mm versus 1.5 mm — a factor of 2.5 from nothing more than rounding a corner. The second is altitude.
| Rated impulse voltage (kV peak) | Clearance, inhomogeneous field (mm) | Clearance, homogeneous field (mm) |
|---|---|---|
| 0.33 | 0.20 | 0.20 |
| 0.80 | 0.45 | 0.25 |
| 1.50 | 0.80 | 0.50 |
| 2.50 | 1.50 | 0.60 |
| 4.00 | 3.00 | 1.20 |
| 6.00 | 5.50 | 2.00 |
| 8.00 | 8.00 | 3.00 |
The practical lesson for sheet metal: deburr and radius the edges near high-voltage conductors. How to write that on a drawing so it can actually be inspected is covered in our edge break and deburring guide. A busbar end with a sharp punched corner, a cut edge on a barrier bracket, or an un-radiused slot end creates an inhomogeneous field and quietly doubles the clearance you need. On our shop floor, the note "break all edges R0.5 within 20 mm of busbar" costs nothing and is one of the highest-value lines you can put on a power-enclosure drawing.
Altitude Correction
The clearance tables are valid to 2 000 m. Above that, air density falls and the breakdown strength of a given gap falls with it, so the standard requires a multiplication factor:
| Altitude (m) | 2,000 | 3,000 | 4,000 | 5,000 | 6,000 |
|---|---|---|---|---|---|
| Correction factor for clearance | 1.00 | 1.14 | 1.29 | 1.48 | 1.70 |
Creepage is not corrected for altitude. This matters more than it sounds: battery storage projects in the Andes, Colorado, Tibet and Xinjiang routinely sit above 2 000 m, and a 4 kV clearance of 3.0 mm becomes 3.9 mm at 4 000 m. If you are quoting a containerised BESS for a 4 300 m site in Chile and your drawing says 3.0 mm, you have an argument waiting to happen at the FAT.
Six Ways to Buy Spacing in Sheet Metal
When the number does not fit, the instinct is to make the box bigger. It is almost never necessary. Ranked by cost, here is what actually works:
| Lever | What it does | Typical gain | Cost | Watch-out |
|---|---|---|---|---|
| 1. Raise the standoff | Lifts busbar or terminal off the subplate / wall | 1 mm per mm of height, to both clearance and creepage | ≈ $0 at design; $0.30–1.50 per part for a longer machined standoff | Eats enclosure depth; check the door still closes and the bend radius still fits |
| 2. Cut a slot | A ≥ 1 mm slot across the creepage path forces the path down and around | Roughly 2× the slot depth added to the path | $0 at laser — it is one more contour line | Burr control is essential; a burr bridges the slot. Deburr both sides |
| 3. Form a rib or groove | A formed feature across the surface lengthens the route | 2× the rib height | $0 at design; tooling for a new form if added later | Rib must be continuous and taller than the required increment |
| 4. Add an insulating barrier | GPO-3, FR-4 or mica sheet between conductor and metal | Largest gain per mm; also introduces a declared material group (GPO-3 = group I) | $1.50–6.00 per part | Must be mechanically captive — riveted or clipped, never held by adhesive alone. Extend past the conductor on all sides |
| 5. Sleeve or shroud the conductor | Heat-shrink, extruded shroud or busbar boot | Local fix at the tight spot | $0.40–2.00 per point | Insulation must be rated for the working voltage and temperature; check the shrink ratio at the lug |
| 6. Re-segregate | Move the power section into its own compartment behind a barrier | Eliminates the path entirely | $0 at concept stage; a full redesign later | Adds a wall, a barrier and usually a second door or access panel — plan it from the start |
And the anti-lever, stated once so it is never repeated: do not count the coating. We have yet to see a paint film survive a busbar being dropped on it during assembly, and IEC 60664-1 will not let you claim it anyway.
Six Product Lines, Six Spacing Profiles
The same standard produces very different geometry depending on what you are building. This is the profile we use when we review a first-article drawing:
| Product line | Typical working voltage | Typical OVC | Typical micro PD | What actually causes the finding |
|---|---|---|---|---|
| New energy charging (DC fast charger) | 1,000 V DC | II – III | 2 | DC busbar to the painted side wall; DC does not self-extinguish an arc |
| Energy storage & power | 1,500 V DC | III | 2 | Rack-to-wall distance plus altitude above 2 000 m; barrier ends not extended far enough |
| Industrial control | 230/400 V AC, up to 690 V | III | 2 (3 in a dirty plant) | VFD output terminals: dv/dt from the drive is harsher than the supply category suggests |
| Industrial HMI & monitor | SELV ≤ 60 V DC | I – II | 2 | Spacing is easy here — the real issue is the bezel bond, not the gap |
| Telecom & network | –48 V DC | II | 2 | Spacing is easy; bonding and EMC shielding take over as the binding constraint |
| Outdoor security & municipal | 230 V AC + PoE | III | 3 | PD 3 plus condensation: the longest creepage of the six, and the most likely to be under-specified |
Note the pattern: the two highest-voltage lines and the one outdoor line are where the money is lost. For EV charging and storage specifically, we recommend designing to material group I (GPO-3 barriers) and OVC III as a default even when the minimum arithmetic would allow less — the barrier costs a few dollars and removes the argument entirely.
9-Point DFM Checklist for Clearance & Creepage
- Dimension and tolerance every critical gap. Not a note, a dimension. "6 ± 0.5 mm from busbar support to rib" is a specification; "maintain adequate clearance" is not.
- State the four inputs on the drawing title block: working voltage, overvoltage category, pollution degree (and whether it is macro or micro), material group. Without these, nobody downstream can check the numbers.
- Mask the coating wherever a barrier or a bonded joint must meet bare metal — and conversely, never rely on the coating as insulation anywhere else.
- Break all edges R0.5 within 20 mm of any high-voltage conductor. Sharp edges create inhomogeneous fields and can double the required clearance.
- Extend insulating barriers past the conductor on every side by at least the same distance as the barrier-to-conductor gap, or the path simply routes around the end.
- Make barriers mechanically captive. Rivets, PEM studs or formed clips. Adhesive alone is not a retained-insulation method and will not survive a hipot re-test after transport.
- Apply the altitude factor if the site is above 2 000 m, and put the corrected number, not the sea-level number, on the drawing.
- Keep the power compartment physically separated from the control compartment where the layout allows it — a barrier wall is cheaper than forty individual barrier plates.
- Specify a first-article dimensional check plus a dielectric test in the QC plan, with the test voltage and duration written out (for IEC 60204-1 assemblies, typically 2 × Un + 1,000 V or a minimum of 1,500 V AC rms for one minute).
Closing — What to Put on the Drawing
Four lines. If your enclosure drawing carries these, most spacing findings disappear before the first article is cut:
- "Insulation coordination to IEC 60664-1:2020 — working voltage ___, OVC ___, pollution degree ___ (micro-environment), material group ___."
- "Minimum clearance ___ mm, minimum creepage ___ mm, dimensioned at all critical points; altitude factor ___ applied for ___ m site."
- "Coating is not credited as insulation at any location. Mask bare metal at all barrier and bonding interfaces."
- "Break all edges R0.5 within 20 mm of busbar; deburr all slots both sides; all barriers mechanically captive."
Send us the single-line diagram, the voltage and the installation altitude, and we will return a one-page Insulation Coordination Pack within 24 hours — the four inputs, the required distances at every critical point, the barrier and standoff geometry we recommend, and a fixed-price line item. Use the contact form or email gongtianbao@xhlmarketing.com. We build to the drawing and to the standard; we do not sign the certificate for your assembly — see our UL 508A and IEC 61439 certification guide for that boundary.
Frequently Asked Questions
What is the difference between clearance and creepage distance?
Clearance is the shortest distance through air between two conductive parts; it is sized by the rated impulse voltage and guards against instantaneous air breakdown. Creepage is the shortest path along the surface of the insulation between the same two parts; it is sized by the RMS or DC working voltage together with the pollution degree and the CTI material group, and it guards against slow tracking over the life of the product. They are calculated separately and both must be satisfied.
How much clearance and creepage does a 400 V industrial control panel need?
A typical 230/400 V cabinet is overvoltage category III, pollution degree 2, material group II. That gives a rated impulse voltage of 4 kV and a clearance of about 3.0 mm line-to-line and line-to-earth. Creepage is not read off the same axis: it is looked up from working voltage, pollution degree and the CTI material group together, so a line-to-line versus line-to-earth split does not apply to it — that distinction belongs to clearance. Take the creepage value from the relevant table in the current edition of IEC 60664-1 (the table number has changed between editions, so do not cite a fixed one) and verify it against your own edition. Reinforced insulation doubles the basic-insulation creepage figure.
Does powder coating or paint count as insulation for creepage?
No. IEC 60664-1 does not credit paint, powder coat or anodising as solid insulation, because the film is thin, porous and easily scratched during assembly. Measure the path as if the metal were bare, or insert a rated insulating barrier with its own material group and thickness.
When do I need to apply the altitude correction factor?
Above 2 000 m. The factor is about 1.14 at 3 000 m, 1.29 at 4 000 m, 1.48 at 5 000 m and 1.70 at 6 000 m, and it applies to clearance only — never to creepage. Storage projects in the Andes, Colorado, Tibet and Xinjiang routinely exceed the threshold.
Who is responsible — the enclosure maker or the panel builder?
Both, at different points. The enclosure maker controls the manufactured geometry: standoff heights, subplate recess, rib and slot features, barrier mounting points, coating masking. The panel builder controls what goes inside: busbar routing, terminal spacing, conductor insulation. The practical fix is to put the required distances on the enclosure drawing as dimensioned, toleranced features, and to state the required insulation level in the RFQ.
The same interface appears wherever a conductor is held by an insulator: our busbar design guide treats support pitch and creepage path as one decision.