On a 2 m cabinet, a 200 mm enclosure viewing window is roughly one percent of the painted surface — and in our experience it is the first one percent to fail. The metal around it survives the impact, the sun and the washdown. The window yellows, hazes, cracks, fogs or leaks, and the whole enclosure gets blamed for a failure that happened in a small square. What follows is the sequence we work through on an OEM enclosure window: material, thickness, sealing, aging, and the awkward case where the window also has to conduct electricity. It is about the pane and the joint. Cutout dimensioning is its own subject, covered in the HMI enclosure guide and the bezel guides.
What is an enclosure viewing window?
An enclosure viewing window — also called a viewport or inspection window — is a transparent glazed panel fitted into an enclosure door or cover so operators can read displays, indicators or meters without opening it. It is a joint, not a part: the glazing, the seal and the sheet metal aperture must be designed together, because they expand, flex and age at different rates.
Why the window fails first
Every other surface on an enclosure has one job: separate inside from outside. A viewing window has four jobs at once, and three of them fight the fourth.
- Let light and sight through. That rules out almost everything that is good at the next three.
- Keep water and dust out to the same IP rating as the rest of the door. Both codes are described in our IP and NEMA ratings comparison.
- Take mechanical impact — a kick, a tool drop, a pressure washer, a stone from a mower. See also the vandal-resistant design guide.
- Survive the environment — UV, temperature cycling, cleaning chemicals, salt.
Metal does jobs 2, 3 and 4 and simply gives up on job 1. Glazing materials do job 1 beautifully and then negotiate the other three. The window failures we have been asked to fix share one pattern: job 1 was optimised in isolation — usually by picking the clearest material on the distributor's page — and the other three were discovered in the field.
The practical consequence is that a window is not a purchased part you drop into a hole. It is a joint between three materials with different stiffness, different expansion rates and different lifespans: the glazing, the seal, and the sheet metal aperture. Design the joint and the window survives. Design only the pane and you get a warranty claim.
Six glazing materials compared
The shortlist for industrial enclosure glazing is shorter than the plastics catalogue suggests. In practice it comes down to six options.
| Material | Surface hardness | Impact behaviour | Relative weight | Continuous service temp. | Relative cost | Light transmission | Best used for |
|---|---|---|---|---|---|---|---|
| Polycarbonate (PC) | Soft (scratches easily); hard-coat available | Does not shatter; the default for vandal and impact risk | ~0.5× soda-lime glass | Moderate — above PMMA, below every glass here | Low | High (~88–90%) | Outdoor kiosks, public-facing HMIs, vandal risk |
| PMMA / acrylic | Soft-moderate; hard-coat available | Brittle compared with PC | ~0.5× soda-lime glass | Low — the lowest of the six | Lowest | Highest (~92%) | Indoor display covers, presentation panels |
| Soda-lime float glass | Hard | Brittle | Baseline | High | Baseline | High (~90%) | Indoor inspection windows, budget builds |
| Chemically tempered aluminosilicate | Very hard | High, when correctly mounted | About the same as soda-lime | High | High | High | Outdoor touchscreens, high-abrasion sites |
| Borosilicate glass | Hard | Moderate | ~0.9× soda-lime glass | Highest of the six | Medium | High | High-temperature windows, wide-temperature HMI |
| Laminated safety glass | Hard | High; interlayer retains fragments | ~2× (two plies) | Moderate — limited by the interlayer, not the glass | High | Slightly reduced | Street furniture, overhead, safety-critical |
Two notes on that table, because the numbers get misused.
Cost and temperature figures are indicative, not specification values. Glass grades, hard-coat systems and interlayer constructions vary enough between suppliers that a relative ordering is only useful for shortlisting. Ask the glazing supplier for the actual data sheet before you commit. Relative weight and transmission are material properties and are more reliable, but still check the grade.
The temper has a temperature limit even though the glass does not. Soda-lime glass as a material is comfortable far above anything an enclosure reaches. The compressive surface layer that carries the temper is not — but the temperatures at which that layer measurably relaxes are well above normal service, so it is not the mechanism you should be designing an enclosure window against. Thermal shock and expansion mismatch are. Choose borosilicate for its low expansion rate and its thermal-shock behaviour. Where a window genuinely sits hot for years, we name borosilicate on the drawing rather than leaving the door open for a tempered soda-lime substitution at value-engineering time.
Impact ratings belong to the assembly, not the pane
This is the specification error we see most often. A team selects "IK10 polycarbonate", machines the aperture, bolts the pane in, and then fails the impact test on the door.
Impact ratings such as those defined in IEC 62262 are awarded to a complete assembly — pane, frame, gasket, fasteners, and the sheet metal behind them. IK07 is 2 J, IK08 is 5 J, IK09 is 10 J and IK10 is 20 J. A pane that survives 20 J in a rigid test fixture can still fail at 5 J in a door that flexes, because the flex loads the pane edge in a way the coupon never saw. If the rating matters to your customer, test the door, not the sample square.
The energy steps and the weak points around windows and vents are set out in full in our IK rating guide for enclosure surfaces.
Enclosure viewing window thickness: set it by span, not by material
Ask "what thickness do I need?" and the honest answer is "how wide is the unsupported span?"
A pane is a plate. Under impact, wind or water-jet load it deflects, and deflection scales with the fourth power of the unsupported span while falling with the cube of thickness. For the same material and support condition, doubling the span multiplies deflection by roughly sixteen; doubling the thickness divides it by about eight. Span wins, but only by a factor of two per doubling — which is why both levers belong on the table, and why "just make it thicker" is usually the expensive one.
In practice this means:
- A 3 mm polycarbonate pane in a 100 mm inspection window is comfortable.
- The same 3 mm pane in a 400 mm-wide HMI window will visibly bow under a water jet and can pop out of a compression seal. The bezel detail matters here too — see the touchscreen bezel guide.
- The fix is usually a divider or a frame that shortens the span, not a thicker pane. A 400 mm window split into two 180 mm panes with a central mullion is lighter, cheaper and stiffer than a single thick pane.
Window thickness for IK rating: starting bands
We do not publish a thickness-to-IK conversion, and we treat any published one with suspicion. The outcome depends on span, frame stiffness, gasket hardness, fastener spacing and the exact strike point — the pane is maybe half the answer. What follows is therefore not a table of what you will get. It is the band we price from, and every door we build to an IK target gets tested as a door before we claim it.
| Impact target | Quotation starting band (confirm by test on the complete door) |
|---|---|
| IK07 (2 J) | 3 mm PC or 4 mm tempered glass, short span, continuous support |
| IK08 (5 J) | 4–5 mm PC, or 4 mm chemically tempered glass with a compliant mount |
| IK09 (10 J) | 6 mm PC, or laminated construction; frame stiffness becomes the limiting factor |
| IK10 (20 J) | 8–12 mm PC, or laminated glass; treat the door as the test article |
The same pane configuration moves a full IK step with frame stiffness alone, which is why we will not put a tick against any of these rows. Put "verified on the complete door" on your drawing instead.
Note what changes as the target rises: past roughly IK08, adding glass thickness stops helping. The failure moves to the frame and the aperture. That is the point at which the sheet metal design — return flanges, stiffeners, fastener pitch — starts to dominate the result, and it is also the point at which thickening the pane stops being the answer. Our own stiffness work is written up in the stiffener rib design guide.
Keeping the IP rating across an enclosure viewing window
An enclosure is rated as a whole. If the door is IP66 and the window leaks, the assembly is not IP66. In our experience the leak is almost never through the glazing material — it is around it. The same principle governs outdoor IP-rated enclosures generally.
Four ways to seal an enclosure viewing window
| Method | How it works | IP potential | Serviceability | Notes |
|---|---|---|---|---|
| Compression gasket in a clamped frame | Pane captured between an outer bezel and the door, squeezing a gasket | High | Good — bezel unbolts | The workhorse. Needs controlled compression, not "tighten until it stops". |
| Structural adhesive bonding (tape / silicone / PU) | Pane bonded directly to the metal reveal | High, when the reveal is flat and clean | Poor — destructive removal | Excellent for flush, tamper-resistant fronts. Surface prep and cure discipline decide everything. |
| Overmoulded / co-moulded window | Glazing moulded into a plastic bezel, bezel then fitted | Very high | Moderate | Best at volume; tooling cost needs amortising. |
| Gasket plus adhesive (two independent barriers) | Bond provides the seal, mechanical bezel provides retention | Highest | Poor | Used on washdown and outdoor equipment where a single failure mode is unacceptable. |
Two rules cut across all four.
Control the compression. An elastomer seal needs to be compressed by a defined fraction of its free height — the height of the section as supplied, before you squash it. The range you will see quoted is roughly 10–25%; it moves with profile and hardness, so we take the number off the gasket supplier's data sheet rather than out of a handbook. Design a metal-to-metal stop or a shouldered spacer so the compression is set by geometry, not by how hard the assembler turns the driver. The numbers are worked through in the gasket sealing guide.
The reveal must be flat. The reveal is the flat land around the aperture that the gasket or the adhesive sits on. Bonding and gasket sealing both fail on a wavy one. Laser-cut sheet is flat, but a large aperture in a thin door is not — and welding a frame around the aperture puts heat into exactly the area that has to stay flat, which is a problem our weld distortion guide covers from the process side. Specify flatness on the drawing.
5 ways an enclosure viewing window leaks
- Corner radii too tight for the gasket. A gasket that has to turn a sharp corner will bridge it. Give the aperture corners a radius the gasket can actually follow.
- Fastener pitch too wide. Compression decays between fasteners. Double the pitch and the mid-span is doing nothing — see the fastener selection guide.
- Seal sitting on a burr. A cut edge with a burr under the seal is a leak path and a cut gasket. Burr direction on the aperture matters and should be specified.
- Pane walking under thermal cycling. See the expansion section below. Repeated movement is also a vibration and shock qualification concern.
- Condensation read as a leak. Fog on the inside of the pane is not a seal failure; it is a dew point problem. It is addressed in the condensation control guide and no amount of extra sealant will fix it.
Sun, heat and chemistry: what actually ages a window
Four degradation modes come up again and again in the window replacements we quote. They are material-specific enough to drive selection.
| Mode | What it looks like | Which materials | Mitigation |
|---|---|---|---|
| UV yellowing / haze | Pane goes amber or milky, display unreadable | PC without a UV-stabilised grade or hard coat is the classic case; PMMA is inherently far more UV-tolerant | Specify UV-stabilised PC with a hard coat, or switch to PMMA indoors |
| Abrasive scratching | Fine swirl marks; glare increases | All plastics; PC is the softest | Hard-coat; specify a cleanable surface and a cleaning procedure |
| Solvent / chemical crazing | Fine crack network, often near fasteners | PC is susceptible under stress with the wrong cleaner; PMMA also attacks easily | Check cleaner compatibility; avoid stress concentrations at clamping points |
| Thermal expansion mismatch | Seal extrusion, pane walk, corner cracking | All glazing against metal, worst with plastics | Design the joint for differential movement |
The expansion number nobody checks
This is the calculation we most often end up running for a customer after the fact. It is not exotic. Coefficients of thermal expansion, as typical values: soda-lime glass about 9 µm/m·K, borosilicate around 3, carbon steel roughly 12, aluminium roughly 23, polycarbonate about 65–70, and PMMA roughly 70–80. Those figures come from a measurement method rather than a catalogue, and your supplier's data sheet will give you the number for the specific grade.
Take a 300 mm polycarbonate window in a steel door, on equipment that swings 60 K between a cold night and full sun. The pane wants to grow by roughly 300 × 70 × 10⁻⁶ × 60 ≈ 1.26 mm. The steel around it grows by roughly 300 × 12 × 10⁻⁶ × 60 ≈ 0.22 mm. The joint has to absorb about a millimetre of relative movement, in a seal that is also trying to keep water out, thousands of times over the life of the product.
A rigidly clamped pane will either bow, extrude the seal, or crack at the fasteners. The fix is to let it move: an oversized aperture clearance, a compliant gasket with room to shear, slotted or oversized fixing holes in the bezel, and a hard stop that positions the pane without pinning it.
Rule of thumb: a 300 mm polycarbonate pane in a steel door needs about 1 mm of joint movement absorbed per 60 K of swing. None of that needs a special part. It needs the clearance drawn in. Material choice interacts with this too — see the material selection guide.
When the window has to conduct: EMC and touch
A display window in a shielded enclosure is a hole in a Faraday cage. If the equipment has to meet an EMC requirement and it also has to show a screen, the glazing has to be electrically conductive while still being optically clear. Those two properties trade against each other.
| Technology | Typical shielding effectiveness | Typical transmission | What it costs you |
|---|---|---|---|
| ITO (transparent conductive coating) | ~20–40 dB | ~80–90% | Limited attenuation; the coating is on the surface and can be scratched or cleaned off |
| Laminated metal micro-mesh | ~30–80 dB depending on mesh density and frequency | ~40–85% | Visible mesh texture, and moiré against a fine display pixel pitch |
| Mesh + ITO hybrid | ~40–70 dB | Mid-range | Highest cost; best combined low- and high-frequency behaviour |
All of those figures are typical ranges from supplier literature, not guaranteed values at a given frequency. Shielding effectiveness varies with frequency, with aperture dimensions and with how well the perimeter is bonded, so any EMI shielding window number quoted without test conditions is marketing.
Making the shield continuous
A conductive pane that is not bonded to the enclosure around its full perimeter is a slot antenna with a decorative window in it. Three things matter:
- 360° perimeter contact. A conductive gasket between the pane frame and the panel, all the way round: conductive elastomer, fabric-over-foam, or a conductive tape bond, depending on how much compression the pane edge can take. Finger stock belongs on doors and removable panels — it is a wiping contact, and against a glass or PC edge it is both too stiff and too abrasive.
- Defined compression. Same discipline as the environmental seal: compression set by geometry, not by torque. Conductive elastomers are usually specified a little tighter than environmental ones — take the figure from the gasket supplier, because it is a function of that gasket's compression load. Our grounding and bonding guide covers the same principle on metal-to-metal joints.
- Close fastener pitch. Uniform compression along the perimeter, and tighter through the corners, which is where contact resistance goes bad first. Pitch is a function of the gasket's compression load per unit length, not a constant — get it from the supplier, then check it is buildable.
If the window also has to be a touchscreen, the conductive layer becomes part of the sensing stack and the two requirements have to be designed together. That is the point at which the mechanical design, the display module and the EMC requirement stop being separable, and it is much cheaper to resolve before the display module is ordered. The enclosure side of the problem is set out in the EMI shielding guide.
AG, AR and hard coat: what each actually fixes
These three get bundled as "window treatment" and specified as a package, which is how you end up paying for an anti-reflective coating on a pane that only needed a hard coat.
| Treatment | What it actually fixes | What it costs you |
|---|---|---|
| Hard coat | Abrasion and scratch resistance | Essential on plastics; without it PC hazes within months of cleaning |
| Anti-glare (AG) | Diffuses reflected light by micro-roughening the surface | Costs some sharpness and adds slight haze |
| Anti-reflective (AR) | Reduces reflection by thin-film interference, keeping clarity | More expensive and more delicate than AG |
| Anti-fingerprint (AF) | Cleanability and water shedding | Cheap; useful on outdoor touchscreens |
AG and AR do opposite things to the surface. They are not substitutes. If a drawing says "AG/AR" because nobody wanted to choose, the supplier will pick one and the display engineers will find out later.
Six product lines, six different answers
The same window question resolves differently across the product families we build, which is why we do not keep a house standard.
| Product line | What the window has to do | Typical starting spec |
|---|---|---|
| EV charging | Readable in direct sun; vandal and impact exposure; payment screen | Chemically tempered glass or hard-coated PC, AG, IK08–IK10, UV-stable seal |
| Energy storage | Status indication; long unattended life; often outdoor | Small inspection window, PC or glass, UV-stable, conservative sealing over easy service |
| Industrial HMI | Operator looks at it all shift; touch likely; EMC often in scope | Chemically tempered or ITO/laminated per EMC need; AG indoors, AR for colour accuracy |
| Telecom and network | Indicator and label viewing; outdoor; UV and temperature cycling | Small PC or PMMA windows; UV-stabilised; seal designed for expansion |
| Industrial control | Inspection of breakers, meters, PLC LEDs; chemical and washdown exposure possible | Glass or PC sized to span; chemical compatibility with site cleaners checked |
| Outdoor municipal | Street furniture, impact and weather; cleaning regime unknown | Hard-coated PC or laminated glass; IK08+; conservative sealing |
On unattended outdoor equipment we favour two independent barriers over a seal that is easier to service, because there is no maintenance visit scheduled to catch the first one failing.
12 DFM rules for enclosure window apertures
These are the aperture-side rules we apply before a window drawing goes out, and they are where most of the avoidable cost sits.
- Put generous radii on the aperture corners. Sharp corners concentrate impact stress in the pane and bridge the gasket.
- Size the aperture to the pane plus clearance, not to the pane. Clearance is a design input, not a tolerance accident.
- Specify burr direction and a deburr requirement on the aperture edge. The seal sits on it — see the deburring guide.
- Hold flatness on the sealing reveal with an explicit callout, and check it after any welding.
- Add a return flange or stiffener around large apertures. The aperture is where the door lost its stiffness. This is the same reasoning as the door and access design guide, applied to a cut-out.
- Set fastener pitch from the gasket requirement, then check it is buildable, not the other way round.
- Design a compression stop. Metal-to-metal or shouldered spacer, so seal compression is geometric.
- Allow for differential thermal expansion in the aperture clearance and in the fixing holes.
- Keep the weld away from the reveal if the reveal is bonded. Heat distorts; adhesive does not forgive.
- Recess the pane below the door face where impact is a risk. Setting the glazing back puts the strike load onto metal rather than onto the pane edge.
- Specify the coating before the window goes in, or protect the reveal. Overspray on a bonding surface is a field failure waiting to happen — see the surface finishing guide.
- Call out the treatment explicitly — hard coat, AG or AR, UV-stabilised grade. "Clear polycarbonate" is not a specification.
Three windows we were called in to fix
A coastal EV charger. The original design used 4 mm polycarbonate in a 380 mm-wide aperture, clamped rigidly. Units came back with hazed panes and, in a handful of cases, cracked corners. Two changes fixed it: a central mullion that halved the span, and slotted bezel holes that let the pane move. The hazing was a separate problem — the grade had no hard coat. The mullion went on at drawing stage; retrofitting it meant a new door, a new bezel and another round of impact testing.
A food-plant control panel. Windows fogged within weeks and the customer assumed the seal had failed. It had not — the panels were washed down hot and then cooled, and the dew point was inside the enclosure. Our first suggestion was a breather vent with a membrane; it was too small for the washdown cycle and the fogging came back on the second batch. We sized up to a larger vent with a drain path and it cleared. The window spec never changed, and the customer's original instinct — more sealant — would have made it worse. Chemical compatibility with the site's cleaning agents is the wider subject of the washdown design guide.
An outdoor telecom cabinet with an EMC requirement. The display window had been specified as plain laminated glass because it looked right. The assembly failed radiated emissions, and the leakage was traced to the window. A mesh-laminated pane with a conductive perimeter gasket passed. The mesh cost more per unit; the alternative was re-working the shielding across the whole cabinet, which came back higher and would have pushed the programme out by weeks.
Paste-ready specification block
Useful starting language for a drawing or RFQ. Delete what does not apply — do not send it as-is.
WINDOW APERTURE:
Clear sight opening ......... ___ x ___ mm
Corner radius (min) ......... ___ mm
Aperture tolerance .......... +/- ___ mm
Reveal flatness ............. ___ mm over the sealing land
Edge break / burr ........... deburr, burrs away from seal face
GLAZING:
Material .................... PC (UV-stabilised) / PMMA / tempered
glass / chemically tempered / borosilicate
Thickness ................... ___ mm
Coating ..................... hard coat / AG / AR / AF
Transmission (min) .......... ___ %
Impact target ............... IK___, verified on the complete door
EMC ......................... none / ITO / mesh-laminated, SE ___ dB
at ___ MHz–___ GHz
SEALING:
Method ...................... compression gasket / bonded / both
Gasket material ............. ___ (check cleaner compatibility)
Compression ................. ___ % of free height, set by metal stop
Target IP rating ............ IP___, tested on the complete assembly
ENVIRONMENT:
Service temperature ......... ___ to ___ °C
Cleaning agents ............. ___
UV exposure ................. indoor / outdoor, ___ orientationFrequently asked questions
What is the best material for an enclosure viewing window?
There is no best material, only a best fit. Polycarbonate is the default where impact or vandal risk dominates and a hard coat is specified. Chemically tempered glass is the default where abrasion, clarity or a touchscreen dominates. PMMA is a strong indoor choice where UV stability and clarity matter and impact does not.
How thick should an enclosure window be?
Thickness follows the unsupported span and the impact target, not the material data sheet. Deflection scales with the fourth power of span and falls with the cube of thickness, so a short-span 3 mm polycarbonate inspection window can outperform a wide 6 mm pane. Splitting a wide window with a mullion is usually cheaper than thickening the pane.
Does a window affect the IP rating of the enclosure?
Yes. The assembly is rated as a whole, and in our experience the window perimeter is one of the most common leak paths. The seal, the flatness of the reveal, the corner radii and the fastener pitch decide the outcome far more than the glazing material does.
Why does my polycarbonate window go yellow or hazy?
Two different causes. Yellowing is UV degradation, addressed by a UV-stabilised grade with a hard coat. Hazing is usually surface abrasion from cleaning, addressed by a hard coat and by specifying how the surface may be cleaned. Check which one you have before changing material.
Can a viewing window meet an EMC requirement?
Yes, with a transparent conductive glazing — ITO coating, laminated metal micro-mesh, or a hybrid — bonded to the enclosure with a conductive gasket around the full perimeter. In our experience the perimeter bond is usually the limiting factor, not the pane.
What is the difference between anti-glare and anti-reflective?
Anti-glare deliberately micro-roughens the surface to scatter reflections, sacrificing a little sharpness. Anti-reflective uses thin-film interference to reduce reflection while keeping clarity. They are not interchangeable, and are specified for different display requirements.
Do impact ratings apply to the window material or the assembly?
The assembly. A pane that survives an impact in a rigid fixture can fail at a lower level in a flexible door, because the door flexes and loads the pane edge. If the rating matters, test the complete door with the window fitted.
Why does my window fog on the inside?
That is condensation, not a leak. Moisture inside the enclosure is condensing on the coldest surface, which is usually the window. Adding sealant will not fix it and can trap more moisture. Address the internal dew point through ventilation, drainage or a heater.
The short version
If you take one thing from this: the pane is the easy part. Size the span before you size the thickness. Hold flatness on the reveal and make the seal compression geometric rather than torque-dependent. Give the pane room to grow — a polycarbonate window in a steel door moves about a millimetre over a 60 K swing, and something has to absorb that. And if the window also has to conduct, design the perimeter bond at the same time as the pane, not after it.
Where we fit
We build the metal side of this problem: the aperture, the reveal, the frame, the compression stop, the fastener pitch, and the glazing assembly that goes into it. We will tell you when a pane choice will not survive the span you have drawn, and we will build the door and put it through impact testing with you rather than assert a rating nobody has measured.
What we do not do is certify the assembly for you. Impact ratings, IP ratings and EMC performance belong to the complete product and its test report. We supply the parts and the evidence from our own checks; the declaration is yours. Qualification planning is covered in the enclosure test plan guide.
If you have a window aperture on a drawing and you are not sure the spec will hold, send us the door drawing. DFM review is free, and a comment on the aperture before tooling costs nothing — a revised bezel after tooling does not.