Why Dock Access Points Die in 14 Months

At 6:15 a.m. on a Tuesday in late April, the dockmaster at a 180-slip marina walked C Dock with a tablet, checking which boats had gone out overnight. Somewhere between slip C-14 and slip C-28 the tablet fell from three bars to nothing and stayed there. The access point on the C Dock pedestal — installed fourteen months earlier — still showed a green LED. Its switch port still showed link at 1 Gbps. It answered every ping sent to it. It simply was not moving client traffic above roughly 3 Mbps, and only to devices within about forty feet.

When the unit came off the pedestal the following week, the radio was fine. The board powered up on a bench and passed traffic. What had failed was a U-bolt. Rust had swollen inside the clamp, split the powder coat on the bracket, and let the assembly rotate about 30 degrees — downward, and slightly seaward. The antenna was aimed at the water.

The purchase order for that access point said IP67.

An IP rating and a salt-fog rating answer different questions

The phrase "IP67, rated for outdoor use" is doing a lot of quiet work in a marina quote. It answers a narrow question, and it answers it precisely. Under IEC 60529, the first digit of 6 means dust-tight — no ingress of dust at all, verified under vacuum. The second digit of 7 means the enclosure survives immersion, tested with its lowest point 1,000 mm below the surface. That is a real, repeatable, well-defined test, and a device that cannot pass it has no business on a dock.

It is also silent on the thing that killed your U-bolt. The IP code is a water and solids ingress standard. Standard IP testing is performed with fresh water, and protection against other fluids — salt water, oils, solvents — is not guaranteed by the rating. Chloride-driven corrosion is not in scope. A brand-new IP67 enclosure and a corroded IP67 enclosure both pass.

The tests that do speak to chlorides are separate standards, and they are worth knowing by name:

• ASTM B117 is the baseline salt spray practice. It specifies a continuous atomized fog of 5 percent sodium chloride solution at 35°C, pH between 6.5 and 7.2, collected at 1 to 2 mL per hour in a 10 cm funnel. Critically, B117 does not set a duration or a pass/fail threshold — those are agreed between customer and manufacturer. "Passed B117" without hours attached means nothing.
• IEC 60068-2-52 Test Kb is the cyclic version, and it is more honest. The 2017 edition defines eight severity methods that alternate salt mist at 35°C (50 g/L solution) with a humidity storage phase at 93 percent RH and 40°C, and in some methods a dry phase. Wet-dry cycling produces corrosion mechanisms that differ from constant wet exposure, which is what a pier actually does twice a day.
• MIL-STD-810 Method 509 is the defense equivalent. If a vendor cites it, ask which revision and how many 24-hour cycles.
• NEMA 250 quietly carries the corrosion test people think IP gives them. Outdoor NEMA types face 600 hours of salt spray benchmarked against a G90 galvanized specimen; Type 4X adds 200 hours benchmarked against passivated 304 stainless. NEMA states the distinction plainly: an IP degree rating "only considers protection against ingress of solid foreign objects and ingress of water," while the NEMA types also consider icing, corrosion and lubricant resistance, and construction details.

Look at a real datasheet with that in mind. Cisco's Aironet 1570 outdoor access point lists "IP67 NEMA Type 4X" and an operating range of −40 to 65°C with no solar loading, dropping to 55°C under 743 W/m² of solar loading. That is a genuinely well-specified outdoor product. The corrosion assurance in that line comes entirely from the four letters most buyers skip. And 200 hours of salt spray is a screening test, not a service-life prediction — the salt spray test has limited value in predicting real-world corrosion because it does not replicate real-world conditions.

None of this is a knock on the IP code, which does its job well. It is a knock on using it as the only environmental number in a specification. If you want a shorthand for what a slip actually is, ISO 9223 classifies atmospheric corrosivity from C1 through CX, where C5 covers coastal zones with strong salinity and CX covers offshore structures and areas with frequent salt spray. A finger pier in St. Andrew Bay is not an office parking lot with weather.

The radio usually outlives the bracket

Most dock access point failures are mechanical, and they happen at the mount. Three mechanisms account for nearly all of it.

304 stainless is not marine hardware. The difference between 304 and 316 is roughly 2 to 3 percent molybdenum, which strengthens the passive oxide film against chloride attack. Without it, 304 is susceptible to chloride pitting. The bracket in the vendor's box is frequently 304, sometimes plated carbon steel, and almost never marked. If it did not come with a material callout, assume the worst.

Even 316 loses in a crevice. Under a washer, inside a thread, beneath a gasket, oxygen gets depleted while the surrounding metal stays oxygen-rich. That differential creates an anode inside the gap, the passive film cannot regenerate without oxygen, and chloride ions migrate in to balance charge — forming concentrated, acidic metal chlorides that accelerate the attack. Molybdenum buys you resistance, not immunity. Design the crevice out where you can, and inspect where you cannot.

Dissimilar metals at the mount finish the job. Galvanic corrosion needs three things: a conductive electrolyte wetting the joint, metal-to-metal electrical contact, and a potential difference. A salt-spray-soaked pedestal supplies the first for free. In the seawater galvanic series, stainless sits noble to aluminum, so the aluminum housing or bracket corrodes preferentially to protect the stainless bolt. Area ratio decides how fast — a small anode against a large cathode concentrates the attack, which is why aluminum fasteners into stainless are unacceptable, and why the general guidance is that the wetted area of the corroding metal should be roughly ten times that of the noble metal. Isolate the joint with a non-conductive bushing or barrier compound, or make the whole assembly one metal.

What "sealed" does to a box that heats up and cools down

A fully sealed enclosure on a piling is a small pressure vessel. Sun heats it, air expands, some escapes past the gasket. Overnight it cools, a partial vacuum forms, and the enclosure pulls in whatever is outside — including water vapor, which condenses inside and has no route out. Do that every day for a year and the water inside is not the water the IP67 test kept out.

The fix is a vented enclosure, not a tighter one. ePTFE protective vents equalize pressure while blocking liquid — typical airflow around 2,000 mL per minute, water entry pressure above 0.3 bar for 30 seconds, service range −40 to 125°C, IP67 maintained. When you evaluate a dock enclosure, look for the vent. Its absence is the tell.

Conformal coating on the board is the second layer. IPC-CC-830 qualifies coatings — acrylic, epoxy, urethane, silicone, parylene among them — against moisture and insulation resistance, thermal shock from −65 to 125°C, dielectric withstand at 1500 VDC, flexibility, and fungus. A coated board buys you time in a humid, salt-laden box. It is not a substitute for keeping the salt out, and connectors and headers are usually masked anyway.

The jackets fail before anything electrical does

Cable and radome degradation is slow, visible, and easy to ignore. UV exposure fades color first and reduces mechanical properties second, ending in cracking that lets water into a jacket that was never meant to hold it. Black jackets with high carbon black content offer the best inherent UV resistance. There is a catch worth knowing: cable manufacturing standards generally define no test requirements for UV resistance at all, which means "UV resistant" on a spec sheet is a marketing term unless the vendor names a test. Buy black, buy sunlight-resistant listed cable, and put the jumper inside conduit or a drip loop where you can.

Four hundred feet of pier, no conduit, no power

The structured cabling model is not negotiable. TIA-568 gives you a 100-meter channel — 328 feet — made up of a 90-meter permanent link and up to 10 meters of patch cords. A 400-foot pier is about 122 meters. You are over budget before you leave the building.

Power over Ethernet does not extend that. Every PoE tier tops out at the same 100 meters: 802.3af delivers 15.4 W at the source and 12.95 W at the device, 802.3at gives 30 W and 25.5 W, 802.3bt Type 3 gives 60 W and 51 W, and Type 4 gives 100 W and 71.3 W. Higher wattage buys you a hungrier radio, not a longer run.

Fiber solves the distance and creates a power problem you then solve locally. Over multimode, 1000BASE-SX reaches 550 meters on OM2 and 220 meters on OM1; over single-mode, 1000BASE-LX reaches 10 kilometers. Run single-mode down the pier, terminate in a vented outdoor enclosure at the far end, and put a PoE injector or media converter there fed by a local circuit — the pedestal power is usually already there for the slips. Two additional benefits follow: fiber is dielectric, so you have removed the long copper conductor that would otherwise carry surge energy down a lightning-prone structure, and you have created one mid-span enclosure to inspect rather than 400 feet of guesswork.

Why a directional antenna beats an omni over open water

An antenna adds no power. It is a passive device that redirects the energy the transmitter gives it, which is why raising gain lengthens reach and narrows the coverage angle at the same time. A dipole reference is 2.14 dBi. An omni on a pedestal spends most of its pattern illuminating open water and sky where no client exists.

Water also punishes you specifically. Reflection off the surface causes multipath interference and fading, and the reflection geometry shifts with every tide and every wake. You want at least 60 percent of the first Fresnel zone clear, with 70 to 80 percent preferred on long links, and the surface moves. Measured work on near-shore maritime links found that at 2.4 GHz with antennas 2 meters up, moving from over-land to over-seawater cost 2 to 3 dBm of additional path loss — a 25 to 40 percent range reduction — while 5 GHz with antennas at 5 meters showed no meaningful loss under the same conditions.

The practical translation: sector antennas aimed down the fairway rather than omnis aimed everywhere, mounted as high as the structure allows, with enough downtilt that the main lobe lands on decks and cockpits instead of skipping off the water. Favor 5 GHz for coverage and reserve 2.4 GHz for the handful of older chartplotters and marine electronics that still need it.

The fuel dock is a cardholder data environment

Every marina network eventually contains one segment nobody drew on the diagram: the fuel dock terminal, sitting on the same flat VLAN as the guest SSID serving 200 transient boaters who have never been vetted and will be gone by Thursday.

The PCI Security Standards Council is direct about this. Network segmentation is not a PCI DSS requirement — it is strongly recommended, because it reduces assessment scope, cost, and risk. Without it, every system with connectivity to the cardholder data environment is in scope, and that includes anything that can reach it through another system, anything that can affect its configuration or security, and anything providing it security services. A flat marina network puts the guest Wi-Fi in scope for a fuel transaction.

The bar for adequate segmentation is worth memorizing, because it is stricter than most people assume: a segmented, out-of-scope system component could not impact the security of the cardholder data environment even if an attacker obtained administrative access on that out-of-scope system. Not "is unlikely to." Could not. And segmentation is validated by penetration testing at least annually under Requirement 11.3.4, which means it needs to be a documented boundary, not a VLAN somebody remembers creating.

While you are in there, note that the EMV liability shift for automated fuel dispensers took effect on April 16, 2021 for Discover, American Express, and Mastercard, and April 17, 2021 for Visa. If the dispenser is older than that shift, the network conversation is only half the exposure.

What turns fourteen months into six years

The failure at C Dock was not a product defect. It was a specification that measured one thing and assumed another.

Ask the vendor for the salt-fog standard, the method number, and the hours — not the IP rating. Specify 316 for every fastener, bracket, band, and washer, and put the grade in the purchase order rather than trusting the box. Isolate dissimilar metals at every joint. Buy vented enclosures and treat a sealed one as a condensation trap. Run single-mode fiber down the pier with a local injector at the end. Point sectors at the slips. Put the fuel dock behind a boundary you could defend to an assessor. Then walk the docks twice a year with a wrench and a camera, because the thing that fails is cheap, visible, and always the same part.