Skip to main content
InterCar

Resources

Why EV chargers fail, and what the field data actually shows

The short answer

EV chargers fail in four places rather than one: the physical connector, cable and screen a driver touches; the network link between the station and its management system; the payment path; and the back end itself. Reported uptime measures the station, so a charger can be counted as up while no driver can start a session on it.

Last updated

There are two ways to measure an EV charger. One asks whether the equipment was online and able to dispense electricity. The other asks whether a driver who pulled up actually left with a charge. Those two numbers are not close to each other, and almost every argument about charger reliability is really an argument about which one is being quoted.

Uptime measures the station, not the session

Publicly funded charging carries a numeric reliability obligation, and it is worth reading precisely, because the definition does most of the work.

Read the exclusions again. A utility outage, a vehicle-side fault, planned maintenance, vandalism, and hours the site says it is closed all come out of the denominator. Each exclusion is defensible on its own — an operator cannot be held to account for a substation — and the cumulative effect is a metric that can stay high through a period in which drivers repeatedly could not charge.

There are also failures the definition does not see at all. A dead card reader on a station that is otherwise online. A stall blocked by a parked combustion car. A screen that responds but never authorizes. None of those makes the port "down", and all of them send a driver away.

What the field data says

The gap between the two measurements has been tested directly, by people driving to stations and trying to charge.

That is the finding that reframed the whole conversation: not that operators were lying, but that they were reporting a different quantity than the one drivers experience.

The second half of that is the part a site owner should sit with. Reliability does not hold flat and then fall off a cliff. It degrades on a curve as hardware ages in the weather, which means a site commissioned with excellent numbers is not a site that stays that way without anyone touching it.

Failure one: the parts a driver touches

The most common causes of a dead charging port are physical, and they are physical in an unglamorous way.

Cables and connectors take the worst of it. A DC fast charging cable is heavy, often liquid-cooled, dropped on concrete several times a day, driven over, and sometimes cut for the copper. Latch mechanisms wear. Pins corrode where a connector has been left face-up in the rain. Screens fail from ultraviolet exposure, moisture ingress and touch layers wearing through.

None of that is a surprise to the manufacturers, which is why their own schedules treat these items as consumables rather than as failures.

A site running on the assumption that a charger is fixed infrastructure — that it is installed, commissioned and then only touched when something breaks — will find those components failing on roughly the schedule the manual predicted, but unplanned, on a Saturday, with a driver standing next to them.

Failure two: the connection to the back end

The second family of failures is the network link between the station and the management system that authorizes sessions, meters them and bills them.

Connectivity dominating that list is consistent with where chargers live: underground garages, the back of parking lots, rural interchanges. Cellular signal is marginal in exactly the places where installing a charger is cheap.

The protocol matters here too. Open Charge Point Protocol (OCPP) is the standard messaging layer between a charge point and its central system, and equipment that speaks it can, in principle, be moved between networks. Equipment that speaks a proprietary dialect is tied to one vendor's back end, which is a commercial fact until it becomes a reliability fact.

Failure three: payment

Drivers and telemetry disagree about payment, and both are right.

Telemetry sees payment as a small share of failures because most sessions on a given network are started in that network's app by drivers who already have an account. Drivers report it as a large share because the failure happens to the people least able to work around it — someone at an unfamiliar network, without that app, on a site with poor cell service, in front of a reader that does not respond.

Which means a payment failure is a site-design problem as much as a hardware one. A working card reader, legible instructions and enough signal to complete a transaction are what stop a functioning charger reading as a broken one.

Level 2 and DC fast chargers do not fail the same way

It is worth separating the two, because a site with both will see two different maintenance profiles and be tempted to manage them as one thing.

An AC Level 2 unit is comparatively simple: a contactor, a control board, a cable and an enclosure. Most of its failures are the connection to the network, the cable and connector, and the electrical supply behind it — a tripped breaker, a loose lug, a GFCI that keeps opening. They are cheap to fix and easy to miss, because a Level 2 charger that has quietly stopped reporting looks exactly like one nobody used that week.

A DC fast charger is a small industrial power plant. It has rectifiers, contactors, cooling, filtration, communication modules and a liquid-cooled cable on the high-power units, and each of those is a separate thing that can fail while the rest of the unit reports itself healthy. It also fails more expensively and takes longer to fix, because parts are model-specific and often on a lead time.

The practical consequence is that the two need different service arrangements. Level 2 wants monitoring that notices silence and a technician who can attend a few units in one visit. DC fast wants scheduled preventive work, a spares position for the parts with long lead times, and a diagnosis path that does not start with a truck roll.

Failure four: the back end goes away

The fourth failure mode is the one nobody plans for, and it takes out working hardware.

This is the strongest practical argument for buying equipment that speaks an open protocol and for keeping the network contract separable from the hardware contract. A charger whose management system can be replaced is a charger that survives its vendor. One that cannot is a stranded asset the day the vendor stops answering, and reconnecting it costs a truck roll per unit even when the firmware exists.

Why the same fault gets three different names

A site with mixed hardware and more than one network sees the same underlying problem described differently by each vendor, which makes it very difficult to tell whether a station has one recurring fault or four unrelated ones.

For an operator, adopting that vocabulary costs almost nothing and pays back immediately: it turns a pile of vendor-specific alerts into something countable, and something countable is something you can hold a maintenance contract to.

What this means if you own the site

Three things follow from all of the above.

First, measure what the driver experiences, not only what the station reports. Whether a session succeeded on the first attempt is the number that correlates with the complaints you receive, and it is the metric the national laboratories have been pushing toward for exactly that reason.

Second, treat the consumables as consumables. Filters, cables, coolant, connectors and screens have intervals, and planned replacement during business hours costs a fraction of an emergency call-out at the moment a port fails.

Third, keep the paths open — an open protocol, a network contract you can exit, and a service arrangement that includes parts and labor rather than diagnosis alone. The difference between a two-hour fault and a two-week one is usually whether somebody local is authorized to open the cabinet.

We are an intermediary: we connect you with vetted, licensed local technicians who carry out the diagnostics and the repair, and the technician confirms the scope and price before work starts. What we can tell you before that call is which of the four failure modes the symptoms point at, because a connectivity fault, a worn cable and a card reader are three different visits. Start at ev charger repair for the service itself, or at fleet roadside assistance program if the chargers are one line in a larger service account.

Where these statements come from

Anything numeric or regulatory on this page carries the document it was read from and the date it was checked against it. A claim with no source is left off the page.

  • Federally funded charging carries a numeric reliability floor: 23 CFR 680.116(b) requires that each charging port have an average annual uptime of greater than 97%, calculated monthly over the previous twelve months.

    23 CFR 680.116 — Data submittal and uptime (eCFR)Checked August 30, 2026

  • The same rule excludes from downtime any outage minutes caused by reasons outside the operator’s control — electric utility interruptions, a failure to charge caused by the vehicle, scheduled maintenance, vandalism or natural disasters — plus hours outside the station’s identified hours of operation.

    23 CFR 680.116 — Data submittal and uptime (eCFR)Checked August 30, 2026

  • A field study of all 181 open public DC fast charging stations in the greater San Francisco Bay Area found that only 72.5% of the 657 CCS connectors were functional, with 22.7% failing because of unresponsive or unavailable screens, payment system failures, charge initiation failures, network failures or broken connectors.

    Rempel et al., Reliability of Open Public Electric Vehicle Direct Current Fast Chargers (Cogitatio / eScholarship)Checked August 30, 2026

  • Those field results conflicted with the 95% to 98% uptime reported by the charging network operators of the same stations, and the authors called for shared, precise definitions of reliability, uptime, downtime and excluded time, verified by third-party evaluation.

    Rempel et al., Reliability of Open Public Electric Vehicle Direct Current Fast Chargers (Cogitatio / eScholarship)Checked August 30, 2026

  • An operations vendor’s analysis of more than 100,000 charging sessions across 2,400 chargers reported that roughly one-third of charging attempts fail even where those chargers reported average uptime near 97%, and that first-time charge success averaged 85% at new stations but fell below 70% by year three.

    ChargerHelp — EV charging reliability analysisChecked August 30, 2026

  • The same field technicians found that damaged or failed physical components — interactive screens, cables and connectors — topped the list of symptoms for down charging ports, and that together with payment-system issues traceable to communications and software glitches these accounted for about two-thirds of all downtime symptoms.

    ChargerHelp — EV charging reliability analysisChecked August 30, 2026

  • Network telemetry from one charging operator attributed charging failures to station connectivity (55%), internal station faults and errors (38%), connector or cable issues (4%), credit card readers (1%) and display screens (1%).

    EV Connect — EV charging station maintenance and upkeepChecked August 30, 2026

  • A California Air Resources Board survey of 1,290 drivers, 1,122 of them EV drivers, found the top reasons for contacting charging customer service were the station not working (42%), payment issues such as no credit card reader or insufficient cell service for a required app (25%), and a broken vehicle connector on the station (17%).

    California Air Resources Board — EV charging customer service surveyChecked August 30, 2026

  • A charging back end can disappear underneath working hardware. When Enel X Way closed its North American electric mobility business, roughly 125,000 JuiceBox chargers were affected — about 100,000 residential units that reverted to basic charging, and about 25,000 commercial units rendered largely non-functional without the back-end software.

    Consumer Reports — What JuiceBox owners should do after Enel X Way’s shutdownChecked August 30, 2026

  • Those commercial units could not be repointed at another back end because they were a closed system that did not fully support OCPP; migration required pushing an OCPP-capable firmware build to each charger and rebooting it, often needing a physical site visit, before it could be reconnected elsewhere.

    Consumer Reports — What JuiceBox owners should do after Enel X Way’s shutdownChecked August 30, 2026

  • The ChargeX Consortium — Argonne, Idaho and NREL national laboratories, funded by the Joint Office of Energy and Transportation — published Recommendations for Minimum Required Error Codes, defining 26 common charging error codes in a CX### format so manufacturers, networks and operators describe the same failure the same way.

    ChargeX Consortium — Recommendations for minimum required error codes (NREL)Checked August 30, 2026

  • Manufacturer schedules treat several DC fast charger components as consumables on fixed intervals: one 350 kW charger manual calls for replacing air filters every two years, charging cables every three years and coolant every five years, alongside annual internal vacuuming, terminal torque checks and interlock tests.

    BTC Power — 350 kW DC fast charger operation and maintenance manualChecked August 30, 2026

  • The US Department of Energy advises EV charging station owners to budget average maintenance costs of up to $400 annually per charger, and notes that annual extended warranties for DC fast chargers can cost over $800 per charger per year.

    US DOE Alternative Fuels Data Center — Operation and maintenance for EV charging infrastructureChecked August 30, 2026

Tell us what you need

Describe the job once. We match it to a vetted, licensed local operator and come back with an estimate. Nothing is dispatched until you agree the price.

Estimates only. The local operator confirms the final price before dispatch.