LITF-PA-2026-239 · EV Charging / Connector Safety

System and Method for Detecting Electric Vehicle Charge Connector Degradation Using Vehicle-Side Control Pilot Signal Analysis, Inlet Temperature, and Fleet Connector Health Sharing

One in seven public charging visits ends without a charge, and broken connectors are part of the reason. Temperature sensing in the handle is already patented, and so is coolant-loop derating. What this disclosure would claim instead is a narrower combination that works from the vehicle side, where no station cooperation is needed. It would read the station connector's health from signals the vehicle already measures, the control-pilot waveform and the inlet temperature, score the specific connector the car is plugged into, and share that score with the fleet so the next driver knows before plugging in.

Electric vehicle plugged into a DC fast charging station at dusk, the charging handle mated to the car's inlet with a warm glow at the connector interface
⚖️ Prior Art Notice: This document is published as defensive prior art under 35 U.S.C. § 102(a)(1). It constitutes prior art against later patent claims covering the inventions described herein, as of the publication date above.

Abstract

A proposed system would detect degradation of an electric vehicle charge connector from the vehicle side, using fused control-pilot signal, proximity-pilot resistance, inlet temperature, and charge current signals processed by edge inference in the vehicle. A charge controller in the vehicle would sample the control-pilot waveform at its inlet, measure the proximity-pilot resistance, and read the inlet temperature. Each connector would be judged against its own learned history and against the same vehicle's history at other connectors, not against a fixed threshold. Two connectors at one station share the same grid, the same weather, and nearly the same duty, so a connector that runs hotter per amp squared than its neighbors is failing even when its absolute temperature looks normal. Three elements of that sensing are already patented or the subject of published patent applications, and this disclosure concedes them: temperature-monitored charging plugs that derate or interrupt on overtemperature, coolant-temperature-differential derating of liquid-cooled cables, and two-step current reduction as a charging device housing heats up. What this disclosure adds is the combination that none of them covers: inferring the station connector's contact condition from the vehicle's own measurements of the pilot signals, scoring the connector per session against a learned baseline, correlating scores across vehicles by station connector identity, and derating from the vehicle side by drawing less current than the station offers. All scoring would run in the vehicle, with no station modification and no cloud connection needed to warn the driver in the moment.

Field of the Invention

This invention relates to electric vehicle charging safety. It covers a vehicle-side combination of control-pilot and proximity-pilot signal analysis, inlet temperature modeling per amp squared, liquid-cooling failure detection, per-connector baseline learning, fleet connector health sharing, and vehicle-side derating, with driver warning and stall routing.

Background

Public charging fails drivers often enough that the failure rate is a published statistic. J.D. Power's 2025 U.S. Electric Vehicle Experience Public Charging Study, fielded January through June 2025 with 7,428 owners of battery electric and plug-in hybrid vehicles, found 14 percent of public charging visits ended without a charge, down from 19 percent in 2024. The most common reason for a failed visit was the charger being out of service or not working properly, which accounted for 60 percent of failed visits. Compounded, that means about 8.4 percent of all public charging visits failed because the charger was out of service. A 2022 field study of 657 DC fast charging connectors at 181 open public stations in the Greater Bay Area found 476 connectors functional (72.5 percent), 149 nonfunctional (22.7 percent), and 32 with a cable too short to reach the vehicle's inlet (4.9 percent). The 149 nonfunctional connectors failed for a short list of reasons that included unresponsive screens, payment failures, charge initiation failures, network failures, and broken connectors. Those two figures come from different years and methods and are not comparable, but they agree on the shape of the problem: a driver who arrives at a public charger has a real chance of leaving without a charge, and broken connectors are part of the reason.

Connectors degrade in a way that is invisible until it is not. Every mating cycle wears the contact plating a little, every thermal cycle relaxes the spring tension a little, and dust, salt mist, and carbon pitting add resistance at the contact interface. Contact resistance is the dominant hotspot in a connector: it grows with mating cycles, contamination, and corrosion, and it is invisible until it produces heat or failure. The heating follows the square of the current. At 40 amps on a Level 2 AC session, a contact that has degraded to 15 milliohms dissipates 24 watts of heat inside the handle, against 0.8 watts for a healthy 0.5 milliohm contact. At 350 amps on a DC fast charge, a 1 milliohm contact dissipates about 122 watts, and at 500 amps a 2 milliohm contact dissipates 500 watts. The standard that bounds this, IEC 62196-1, limits the temperature rise of electrical terminals in the charging connector to 50 kelvin. Past that limit, housings soften, plating degrades faster, and the heat makes more heat. A Tesla owner who came back to a public Level 2 charger in 2019 found the J1772 adapter had melted to the charging receptacle, and the discussion that followed settled on the likely cause being a dirty or worn-out socket that was borderline acceptable for other vehicles but could not carry a Model 3 drawing the full 32 amps for hours.

The signaling that governs every AC session already carries the information needed to see this coming, and the vehicle already measures it. The control pilot is a plus or minus 12 volt, 1 kilohertz pulse-width-modulated square wave generated by the charging station, which is the Electric Vehicle Supply Equipment (EVSE), through a 1 kiloohm source resistor. The vehicle signals its state by loading the line with a resistor: an open circuit reads plus 12 volts in State A, 2.74 kiloohms pulls the positive peak to approximately plus 9 volts in State B, 882 ohms pulls it to approximately plus 6 volts in State C, and 246 ohms pulls it to about 2.4 volts in State D, reported as plus 3 volts in the standard within its plus or minus 1 volt tolerance. The duty cycle advertises the available current: between 6 and 51 amps, the allowed current equals the duty cycle percentage multiplied by 0.6, so 50 percent means 30 amps. A 5 percent duty cycle signals that digital communication is required, which is how ISO 15118 sessions are announced over the same pilot wire. ISO 15118 is the international standard for digital communication between the vehicle and the charger, covering identity, authorization, power limits, and metering. The proximity pilot is a resistor in the handle itself: in the SAE J1772 scheme a fully seated connector reads 150 ohms and pressing the release button reads 68 ohms, while the IEC 62196 ladder uses 1500, 680, 220, and 100 ohms to encode cable ampacity. Both signals are read by the vehicle through simple analog-to-digital converters, and neither is used today for connector health.

Patented Connector-Monitoring Art

Three load-bearing ideas in this field are already patented or the subject of published patent applications, and this disclosure does not claim them on their own. US 11,117,479 covers a temperature-monitored charging system for transmitting electric charge currents, with temperature sensors in the plug-in connector that interrupt charging or reduce current when a limit is exceeded, and notes the IEC 62196-3 limit of a 50 kelvin temperature rise, a 200 amp continuous limit for largely standardized uncooled connector geometries, and up to 400 amps in cooled systems. Published application US 20230088001 covers a protected charging connector with temperature sensors at several positions on a liquid cooling circuit, controlling charge current from temperature differences between those sensors, and observes that temperature measurements are usually too slow to serve as a safety mechanism. Published application US 20150054462 covers a method of operating a charging connection device in which a temperature value inside the housing is continuously detected and the maximum permissible charging current is reduced in a first step when a lower threshold is crossed and cut off at a higher threshold. Any claim here to connector temperature sensing with derating alone would read on that art.

US 12,083,911 covers a liquid-cooled heavy-current charging cable with a cooling unit, including charging at 500 amps and 1000 volts for 500 kilowatts, with non-cooled cables used up to 250 amps. US 12,296,703 covers a liquid-cooled cable arrangement for high-power fast charging and notes that the contact resistance between the power contacts of the connector and the mating contacts in the car socket is typically a few percent of the ohmic resistance of the conductors, with a 20 to 30 degree Celsius gradient expected along the power contacts at 500 amps. The 50 kelvin terminal temperature rise limit appears in both IEC 62196-1, for general connector requirements, and IEC 62196-3, for DC charging, as cited in US 12,296,703 and US 11,117,479 respectively. Any claim here to liquid cooling of a charge cable alone would read on that art.

What exists today in products measures at the station, not at the vehicle. Tesla's charging handles carry NTC thermistors, which are negative temperature coefficient resistors whose resistance falls as they heat, used to sense high temperatures and shut off, and teardown work shows a 10 kiloohm at 25 degrees Celsius thermistor wired to ground in the handle. OpenEVSE's support forum notes that Tesla is the only maker its contributors were aware of implementing temperature cutoff in the J1772 handle, and that there have been a few cases of a handle or connector melting. Connector makers such as Workersbee now fit temperature sensors into CCS1, CCS2, GBT, and NACS connectors, with a two-level response: reduce current past a warning threshold, stop charging past a critical limit. High-power CCS1 plugs carry two PT1000 sensors, platinum resistance thermometers, at the power contacts, a specified contact resistance of 0.5 milliohms maximum, a terminal temperature rise under 50 kelvin, and a design life of more than 10,000 mating cycles. Crowdsourced availability exists too: PlugShare check-ins tell drivers whether a station worked for the last driver. None of these measures the connector from the vehicle side, none of them scores a specific connector against its own history, and none of them shares an electrical health score with the next driver. A station's own sensor can only protect the stations that have one, and most J1772 handles do not.

PlugShare check-ins do not make this combination obvious. A check-in records whether a station worked, not why, and it records nothing electrical. Combining a crowdsourced availability log with vehicle-side electrical measurements is the step no product has taken, and the gap between "the last driver could not charge" and "this connector's contact resistance has doubled" is the whole invention. The search behind this disclosure covered US and European patent art through the Justia and Google Patents databases and commercial charging products through manufacturer documentation, as cited in the references.

A search of US and European patent art and of commercial charging products found no system combining these elements. The combination has seven elements:

  • vehicle-side measurement of the control-pilot waveform and proximity-pilot resistance as probes of the station connector's contact condition;
  • inlet temperature modeled per amp squared, with the connector judged against its own learned baseline and against the same vehicle's history at other connectors;
  • a liquid-cooling failure signature for DC fast charging, detected from temperature rise decoupled from current squared;
  • per-connector baseline learning keyed by station connector identity from the ISO 15118 handshake;
  • fleet correlation across vehicles using the same station connector;
  • vehicle-side derating by drawing less current than the station offers, with graded driver warnings; and
  • scoring rejection rules for bad data.

One concession is needed on scope. Connector temperature sensing with derating is patented, and the vehicle-side temperature channel in this disclosure reads on that art when taken alone. The distinction here rests on the pilot-signal channels and the fleet correlation, not on temperature sensing alone, and on the system requiring no station modification and no sensor in the handle. Two distinctions are worth stating plainly. First, the temperature derating in US 20150054462 uses fixed absolute thresholds in two steps, while the derating here uses a per-connector learned baseline fused with the pilot-signal channels, so a connector is judged against its own history rather than a single number. Second, the cooling-failure signature here is inferred from the vehicle side, from inlet temperature rising while current is flat or falling, with no access to coolant data at all, while US 20230088001 measures coolant temperatures at the station and derates from their difference. That combination, packaged as vehicle software plus a fleet service, is the gap this disclosure addresses.

Detailed Description

This section describes a proposed system. No prototype exists. Every threshold, rate, and cost figure in it is a design target, not a measured result. Conditional language is used deliberately throughout.

1. System Overview

Four parts would make up the system. A charge controller in the vehicle would sample the control-pilot waveform and the proximity-pilot resistance at the inlet, read an inlet temperature sensor, and record the charge current it is drawing. A scoring unit in the vehicle would combine those signals into a connector health index for the specific station connector the vehicle is plugged into. A fleet service would collect per-connector scores from participating vehicles, correlate them by station connector identity, and return a health rating for each stall to vehicles routing to that station. A driver display would show the connector's state and issue the warning. No part of the charging station would be modified, no sensor would be added to the handle, and no cloud connection would be needed to score a connector or warn the driver in the moment. First-visit baselines and pre-arrival warnings would need the fleet service, since a vehicle that has never used a connector has no baseline for it. The fleet service would add the cross-vehicle view, and only with explicit opt-in.

2. Control Pilot Channel: The Pilot Line as a Probe

A degrading contact changes the divider the control-pilot line forms. The station generates plus 12 volts through a 1 kiloohm source resistor, and the vehicle loads the line with 2.74 kiloohms in State B or 882 ohms in State C. In State C the nominal positive peak is 12 volts times 882 divided by the sum of 1000 and 882, which is 5.62 volts, reported by the standard as approximately plus 6 volts with a plus or minus 1 volt tolerance. Added series resistance in the control-pilot pin contact, from oxidation or fretting corrosion on the same handle that carries the power pins, raises the total series resistance and pulls the measured peak down. With 300 ohms of added contact resistance, the peak becomes 12 volts times 882 divided by the sum of 1300 and 882, which is 4.85 volts, a shift of 0.77 volts from nominal. That shift sits inside the plus or minus 1 volt state tolerance, so the state machine would still read State C and charging would proceed. But a 12-bit analog-to-digital converter over a 15 volt range resolves about 3.7 millivolts per count, so a 0.77 volt shift is about 200 counts, easily measurable. The control-pilot channel is therefore a trending signal, not a fault signal: a single reading inside tolerance proves nothing, while a peak that has drifted 0.77 volts from the same connector's own baseline over months is evidence of contact degradation on that connector. The 0.5 volt flag in the claims corresponds to about 185 ohms of added series resistance: 12 volts times 882 divided by the sum of 1000, 882, and 185 gives 5.12 volts, which is 0.5 volts below the 5.62 volt nominal. A 0.5 volt flag is about 137 converter counts, far above a noise floor of a few counts, and the flag would stay inert until a baseline exists for that connector.

A second, weaker signal would ride on the rising edge of the pilot waveform. The pilot cable and the inlet present a few nanofarads of capacitance, and the 1 kiloohm source resistor with that capacitance sets the edge time constant. Added series resistance in the contact slows the rising edge further, by roughly 30 percent for 300 ohms of added resistance against a 1 kiloohm source. Measuring that change needs microsecond-resolution edge timing, from a comparator with timestamping or a fast sampling converter, and the disclosure does not pretend a slow polling converter can see it. The edge channel would be used only where the hardware supports it, and the scoring would say which channels are active. Duty cycle would be checked as a third control-pilot item: jitter in the measured duty cycle beyond the learned baseline would indicate an unstable pilot contact, and a duty cycle that reads 5 percent would mean the session is using digital communication under ISO 15118, in which case the pilot waveform channel would be suspended and the DC current-demand channel in section 4 would take over. On DC fast charging the analog pilot channel would be suspended entirely, since the session runs on ISO 15118 digital communication, so DC protection would rest on the temperature and cooling-signature channels alone.

Proximity-pilot readings would work the same way, as a resistance measurement rather than a state check. In the SAE J1772 scheme a fully seated connector reads 150 ohms and pressing the release button reads 68 ohms, with an open circuit above 1 kiloohm. A seated connector reading 250 ohms instead of 150 ohms would indicate about 100 ohms of added resistance in the proximity path, which shares the handle's environment with the power pins. In the IEC 62196 scheme the ladder is 1500, 680, 220, and 100 ohms encoding cable ampacity, so a 32 amp cable carries a 220 ohm coding resistor and 100 ohms of added contact resistance would read as 320 ohms total, which matches no standard coding value. The vehicle would flag the reading rather than trust it, and would treat the mismatch itself as evidence about the handle. Neither proximity reading can convict a connector alone, because the proximity resistor lives in the handle and its own tolerance and temperature drift must be separated from contact resistance. The vehicle would compare seated readings with seated readings and pressed readings with pressed readings. A degraded pressed reading of 168 ohms is indistinguishable from a healthy seated reading of 150 ohms, so the state must be known before the reading means anything, and the flag would trigger only outside the tolerance band for the known state. The proximity channel is a corroborating channel, and the disclosure says so.

Seating instability during the first 30 seconds after plug-in would be a check of its own. As the latch seats, the control-pilot peak and the proximity resistance fluctuate. A connector that never settles, with peak jitter above 0.2 volts or proximity readings swinging between the seated and open-circuit bands for more than 30 seconds, would be flagged as incompletely mated, and the display would prompt the driver to reseat the connector before charging starts. Those numbers are design targets.

3. Temperature Channel: Heat Per Amp Squared

Temperature is the channel that carries the most weight, because heating is what contact resistance does. Unit conventions for this section: temperature rises are stated in kelvin, absolute temperatures in degrees Celsius, and resistances in ohms unless a kiloohm value is stated. The scoring unit would model the expected inlet temperature rise above ambient as a function of current squared times contact resistance, plus cable heating, and compare the measured rise against that model. The arithmetic is the whole argument: at 40 amps on a Level 2 session, a healthy 0.5 milliohm contact dissipates 0.8 watts while the same current through a degraded 15 milliohm contact dissipates 24 watts, thirty times as much. A healthy 0.5 milliohm contact at 500 amps still dissipates 125 watts, which is why high-power cables are liquid cooled, and why the model must separate the contact's heat from the cable's heat rather than treating all temperature rise as a connector fault.

Watts dissipated per contact, by current and contact resistance (values rounded):

Current0.5 mΩ1 mΩ2 mΩ15 mΩ
40 A0.8 W1.6 W3.2 W24 W
48 A1.15 W2.3 W4.6 W34.6 W
350 A61 W122 W245 W1,838 W
500 A125 W250 W500 W3,750 W

Each connector would get its own learned model, not a single setting for all connectors. On the first sessions at a new connector, the vehicle would record the temperature rise per amp squared and store it as that connector's baseline. On later sessions the vehicle would compare the measured rise against the baseline for the same connector and against the vehicle's own history at other connectors. A connector whose rise per amp squared has doubled since its baseline, while the same vehicle at other connectors has not changed, would be scored as degrading. Where the vehicle has history at two connectors at the same station, the comparison would also run between neighbors: a connector running hotter per amp squared than its stall-mate under the same weather and duty would be scored as degrading even before its own baseline shows a trend. First-warning thresholds, as design targets, would be a rise per amp squared 50 percent above the connector's baseline, or an absolute inlet temperature rise above 35 kelvin on an AC session. Pull-over thresholds would be a rise per amp squared double the baseline, or an absolute rise above 50 kelvin, the IEC 62196 limit. Those numbers are design targets, pending measurement, and the absolute thresholds would be adjusted for ambient temperature, which the vehicle already measures.

Sun and weather would complicate the comparison. A handle left in direct sun arrives at the session already warm, so the system would score the rise from the inlet temperature at plug-in, not the absolute temperature, and suspend scoring when the starting temperature is so high the rise cannot be separated from the soak. Rain would add leakage paths that shift both pilot channels, so wet sessions would suspend the pilot channels. Section 7 lists the full rejection rules.

4. DC Fast Charging and Liquid-Cooling Failure

DC fast charging changes the physics, because the currents are ten times higher and the cables are liquid cooled. At 350 to 500 amps, the cable itself generates enough heat that the cooling loop is part of the rating: non-cooled cables are used up to 250 amps, and above that the coolant carries the heat away. If the coolant pump fails or the loop loses flow, the cable and connector heat up even at moderate current, and the vehicle's own battery thermal management can mask the signature, since the vehicle may already be limiting current for the battery's sake. Cooling failure would show up as a temperature rise decoupled from current squared: inlet temperature rising while current is flat or falling, or rising faster than the current-squared model predicts at a current the model says should be safe. That decoupling is the signal, because a degrading contact makes heat in proportion to current squared, while a failed cooling loop makes heat accumulate regardless of current. On DC sessions the analog pilot channel is suspended, since the session runs on ISO 15118 digital communication, so DC protection rests on the temperature and cooling-signature channels alone. That is a real gap: the highest-current case is the one with the fewest channels. The disclosure states it rather than hiding it.

Response would differ by charging mode. On AC, the vehicle would reduce its onboard charger draw below the advertised limit, which is a derating the vehicle controls directly, since the advertised duty cycle is a maximum, not a command. On DC, the vehicle would cap the current it requests in the ISO 15118 charge loop, lowering its current demand step by step as the connector score worsens. If the score goes red, the vehicle would stop the session by returning the control pilot to State B, which opens the station contactor within 100 milliseconds, and tell the driver to move to another stall. No part of this design would wait for the station's own temperature sensor to act, because most handles do not have one, and the stations that do have one are the stations that need this system least.

5. Per-Connector Baseline and Fleet Correlation

A score is only useful if it attaches to the right connector. On DC sessions the station sends its EVSE identifier in the ISO 15118 session setup exchange, and the vehicle would key its per-connector record by that identifier. On AC sessions, which usually run on the analog pilot signal with no ISO 15118 handshake at all, the identity would come from the station and stall as known to the vehicle's navigation and charging apps. Either way the record would store the baseline rise per amp squared, the control-pilot peak baseline, the proximity resistance baseline, the mating count, and the score history for that connector. Each session at that connector would update the record, and the record would travel with the vehicle, not with the station.

The fleet service would do what one vehicle cannot: separate a bad connector from a bad vehicle. If one vehicle reports a connector degrading, the cause could be that vehicle's own inlet. If three or more vehicles report the same connector degrading within 30 days, with no change in their own inlet baselines at other connectors, the cause is the connector. Those numbers are design targets. The fleet service would then warn the next driver before plug-in, through the navigation display, and route the driver to a healthy stall at the same station where one exists. It would also count matings per connector. High-power CCS1 plugs are designed for more than 10,000 mating cycles, and other connector types carry their own rated counts, so the fleet service would flag a connector approaching the rated count for its type. Cycle count is the one wear measure that needs no sensor at all, and the count is a lower bound: sessions by vehicles that do not participate are never counted, so the flag would fire on the counted matings, not on a complete history. No GPS coordinates would leave the vehicle, and only feature vectors and outcome labels would sync, only with explicit opt-in. Direct identifiers would be removed before sync: the vehicle identifier would be replaced with a rotating pseudonym, and records would be retained for 12 months. That claim needs a qualifier: the fleet service would still receive timestamps and station identities, which can reveal travel patterns, so "no GPS coordinates" means no position fixes, not no information about movement.

6. Vehicle-Side Derating and Driver Warning

The index would combine the three channels by subtraction from 100. Temperature rise per amp squared above baseline would subtract up to 40 points: none at baseline, 40 at double the baseline, scaled linearly between. Control-pilot peak shift would subtract up to 30 points: none at no shift, 30 at a 1.0 volt shift, scaled linearly. Proximity drift would subtract up to 20 points: none at no drift, 20 at a coding mismatch. A cooling-failure signature would subtract 10 points. A worked example: a connector at 50 percent above its temperature baseline loses 20 points, with a 0.77 volt pilot shift losing 23 points and no proximity drift, for an index of 57, which is amber. When the baseline for a connector is missing or scoring is suspended, the display would show the connector as unrated rather than green.

IndexStateVehicle responseDriver message
80–100GreenNo actionConnector within baseline
50–79AmberCap current drawUse another stall next time
Below 50RedStop session or refuse to startMove to another stall
No baselineUnratedScore after this sessionFirst visit, no history yet

A red score would not refuse a charge when the battery is below 10 percent and no alternative stall or station is within reach. In that case the vehicle would cap current to the lowest level that still charges and warn continuously, since a stranded driver is worse than a slow charge. The driver would always be able to override a red refusal from the display, with the override logged. The fleet rating would appear in navigation before the driver arrives, so a red stall is one the driver never plugs into, not one the driver discovers by smell. The vehicle would also carry the warning into the charging app's session summary, so a driver who ignored an amber warning at a station they use every day sees the trend, not just the single session.

Thresholds, their basis, and their status:

ThresholdValueBasisStatus
Control-pilot peak flag0.5 V shiftDivider arithmetic, about 185 Ω addedDesign target
Temperature first warning50% above baseline, or 35 K risePer-connector model; IEC 50 K limitDesign target
Temperature pull-overDouble baseline, or 50 K riseIEC 62196 50 K limitDesign target
Fleet correlation3 vehicles in 30 daysCross-vehicle agreementDesign target
Warm-up suspensionFirst 10 minutesThermal transientDesign target
Minimum session15 minutesHeating signalDesign target
Mating-cycle flagRated count for the type (10,000 for HPC CCS1)Manufacturer specificationDesign target

Derating from the vehicle side is the element that makes the system useful on stations the driver does not own. Many public handles carry no temperature sensor at all, and a driver cannot add one to a station they do not own. But the vehicle already controls how much current it draws, on AC through its onboard charger and on DC through its ISO 15118 current request, so the vehicle can protect itself and its own inlet from a bad connector without the station's permission. That is the whole point of doing this from the vehicle side.

7. Scoring Rejection Rules

Scoring would be suspended when the data cannot support it. The scoring unit would reject the first 10 minutes of a session, while inlet temperatures are still rising toward equilibrium. It would reject sessions shorter than 15 minutes, which do not generate enough heating to score. It would reject scoring when ambient temperature is above 40 degrees Celsius, when the inlet is sun-soaked at plug-in, and when the vehicle is current-limited by its own battery state of charge or temperature rather than by the connector, since a battery-limited session cannot separate connector heating from battery heating. It would reject the pilot channels on wet sessions and the temperature channel when the inlet temperature sensor is out of its calibrated range. Each rejection would be logged with its reason, and the display would show scoring as paused rather than green. These rejection rules are part of the claimed system, because a score produced from bad data is worse than no score.

8. Figures Description

  • Figure 1: System architecture showing the vehicle charge controller sampling the control-pilot waveform and proximity-pilot resistance at the inlet, the inlet temperature sensor, the scoring unit, the driver display, and the opt-in link to the fleet service.
  • Figure 2: Control-pilot voltage divider diagram for State C, showing the station's 1 kiloohm source resistor, the vehicle's 882 ohm load, the nominal 5.62 volt positive peak, and the shift to 4.85 volts with 300 ohms of added contact resistance, inside the plus or minus 1 volt state tolerance.
  • Figure 3: Heating table showing watts dissipated per contact at 40, 48, 350, and 500 amps for contact resistances of 0.5, 1, 2, and 15 milliohms, with the IEC 62196-1 50 kelvin terminal temperature rise limit marked.
  • Figure 4: Temperature rise per amp squared for a healthy connector and a degrading connector across sessions, showing the learned baseline, the 50 percent first-warning threshold, and the doubling pull-over threshold.
  • Figure 5: Liquid-cooling failure signature for a DC fast charge session, showing inlet temperature rising while current is flat or falling, against the current-squared model prediction.
  • Figure 6: Fleet correlation diagram showing three vehicles reporting the same station connector degrading, the stall-level alert, and the routing of a fourth vehicle to a healthy stall at the same station.

In patent language, that system is:

Claims

  1. A system for detecting degradation of an electric vehicle charge connector, comprising: a charge controller in a vehicle configured to sample a control-pilot waveform at a charge inlet of the vehicle, measure a proximity-pilot resistance at the inlet, read an inlet temperature sensor at the inlet, and record a charge current drawn by the vehicle; a scoring unit in the vehicle; and a driver display; wherein the scoring unit computes, for a station connector mated to the inlet, a control-pilot positive peak shift relative to a learned baseline for that station connector, a proximity-pilot resistance drift relative to a learned baseline for that station connector, and an inlet temperature rise per amp squared relative to a learned baseline for that station connector; combines the shift, the drift, and the rise into a connector health index from 0 to 100 by subtracting from 100 up to 40 points for temperature rise per amp squared above the learned baseline, up to 30 points for control-pilot positive peak shift, up to 20 points for proximity-pilot resistance drift, and 10 points for a cooling-failure signature, the index being 80 and above for green, 50 to 79 for amber, and below 50 for red; reports the connector as unrated rather than green when the learned baseline for that station connector is missing or scoring is suspended; and issues a graded warning through the driver display when the index crosses a threshold.
  2. The system of claim 1, wherein the scoring unit computes the control-pilot positive peak shift in State C, the nominal positive peak being 12 volts times 882 divided by the sum of 1000 and 882, and flags a shift of more than 0.5 volts from the learned baseline while the state remains C, the shift being inside the plus or minus 1 volt state tolerance.
  3. The system of claim 1, wherein the scoring unit measures a rising-edge time of the control-pilot waveform and flags an edge that has slowed by more than 25 percent relative to the learned baseline, the slowing indicating added series resistance in the control-pilot pin contact.
  4. The system of claim 1, wherein the scoring unit measures the proximity-pilot resistance and flags a reading that matches no standard coding value, the standard coding values being 150 ohms seated and 68 ohms release-button-pressed under SAE J1772 and 1500, 680, 220, and 100 ohms under IEC 62196, the mismatch indicating added resistance in the proximity path of the station connector handle.
  5. The system of claim 1, wherein the scoring unit, in combination with the control-pilot positive peak shift of claim 2, models an expected inlet temperature rise as current squared times contact resistance plus cable heating, and flags a measured rise per amp squared more than 50 percent above the learned baseline for that station connector, or an absolute inlet temperature rise above 35 kelvin, as a first warning, and more than double the baseline, or an absolute inlet temperature rise above 50 kelvin, as a pull-over warning, the measured rise being compared against the learned baseline for that station connector and against the same vehicle's history at other connectors at the same station.
  6. The system of claim 1, further comprising a liquid-cooling failure detector that, during a DC fast charging session in which the analog control-pilot channel is suspended, flags a cooling failure when inlet temperature rises while charge current is flat or falling, or rises faster than the current-squared model predicts, the flag being corroborated by the control-pilot positive peak shift of claim 2 recorded for that station connector, and caps a requested current in an ISO 15118 charge loop in response.
  7. The system of claim 1, further comprising a vehicle-side derating unit that reduces a current drawn by the vehicle below a current advertised by the station, on AC by reducing an onboard charger draw and on DC by lowering a current request in an ISO 15118 charge loop, when the connector health index falls below 80, and returns the control pilot to State B to open a station contactor when the index falls below 50, except that the derating unit does not refuse a charge when the battery is below 10 percent and no alternative stall or station is within reach, in which case it caps current to the lowest level that still charges and warns continuously, and a driver override from the display is always available with the override logged.
  8. The system of claim 1, further comprising a per-connector record keyed by a station connector identity received in an ISO 15118 session setup exchange for DC sessions, or by station and stall identity from the vehicle's navigation for AC sessions, the record storing a baseline control-pilot positive peak, a baseline proximity-pilot resistance, a baseline temperature rise per amp squared, a mating count, and a score history for that station connector.
  9. The system of claim 8, further comprising a fleet service that receives per-connector scores from a plurality of vehicles, only with explicit opt-in from each vehicle, correlates scores by station connector identity, issues a stall-level alert when three or more vehicles report the same station connector degrading within 30 days with no change in their own inlet baselines at other connectors, and returns a health rating for each stall to a vehicle routing to that station.
  10. The system of claim 9, further comprising a mating-cycle counter maintained by the fleet service that increments a mating count for a station connector on each charging session reported by a participating vehicle and flags the station connector for inspection when the mating count approaches the rated cycle count for that connector type, the rated cycle count being 10,000 matings for high-power CCS1 connectors.
  11. The system of claim 1, further comprising scoring rejection rules that suspend scoring during the first 10 minutes of a session, for sessions shorter than 15 minutes, when ambient temperature is above 40 degrees Celsius, when the inlet is sun-soaked at plug-in, when the vehicle is current-limited by battery state of charge or temperature, and on wet sessions for the control-pilot and proximity-pilot channels, and that log each suspension with its reason and report scoring as paused in the driver display.
  12. The system of claim 1, wherein the driver display shows a health rating for each stall at a station in a navigation display before the vehicle arrives, and shows a trend of the connector health index across sessions at a station the vehicle uses repeatedly.
  13. The system of claim 1, further comprising a handle-seating check that measures instability in the control-pilot waveform and proximity-pilot resistance during the first 30 seconds after plug-in and prompts the driver to reseat the connector when the instability exceeds a threshold of 0.2 volts of control-pilot peak jitter or proximity-pilot readings swinging between the seated and open-circuit bands for more than 30 seconds.
  14. A method for detecting degradation of an electric vehicle charge connector, comprising: sampling a control-pilot waveform at a charge inlet of a vehicle; measuring a proximity-pilot resistance at the inlet; reading an inlet temperature sensor at the inlet; recording a charge current drawn by the vehicle; computing, for a station connector mated to the inlet, a control-pilot positive peak shift relative to a learned baseline for that station connector, a proximity-pilot resistance drift relative to a learned baseline for that station connector, and an inlet temperature rise per amp squared relative to a learned baseline for that station connector; combining the shift, the drift, and the rise into a connector health index from 0 to 100 by subtracting from 100 up to 40 points for temperature rise per amp squared above the learned baseline, up to 30 points for control-pilot positive peak shift, up to 20 points for proximity-pilot resistance drift, and 10 points for a cooling-failure signature, the index being 80 and above for green, 50 to 79 for amber, and below 50 for red; reporting the connector as unrated rather than green when the learned baseline for that station connector is missing or scoring is suspended; and issuing, through a driver display, a graded warning when the index crosses a threshold, and reducing the charge current drawn by the vehicle below a current advertised by the station when the index falls below 80, and returning the control pilot to State B when the index falls below 50, except when the battery is below 10 percent and no alternative stall or station is within reach.

Implementation Notes

Bill of materials for the vehicle-side base version targets the cost of software alone on vehicles that already measure inlet temperature and sample the pilot signals, since the control-pilot and proximity-pilot channels use analog-to-digital converters the charge controller already has. For vehicles without an inlet temperature sensor, a single NTC thermistor at the inlet adds about 2 dollars in volume. The scoring unit fits in under 50 kilobytes and runs in under 5 milliseconds per scoring cycle on an automotive microcontroller, both targets. Scores would update every 30 seconds during a session, also a target. The fleet service stores per-connector records keyed by station connector identity, with direct identifiers removed, and syncs only on opt-in. The driver display shows each stall as a single number with a color state, the fault class in plain words, and a one-line instruction. The vehicle stores per-connector history locally and syncs only on opt-in.

Calibration discipline matters more than sensor quality. The control-pilot baseline for each connector must be learned from that connector's own early sessions, because pilot peak voltages differ between station hardware, and the temperature baseline must be learned per connector for the same reason. The scoring unit enforces this by refusing to score a connector whose baseline is missing and by asking the fleet service for a prior baseline for that connector identity where one exists from other vehicles. A connector that starts degraded and stays degraded looks stable to a trend detector, and only the cross-vehicle comparison catches it. The scoring unit must also enforce the rejection rules in section 7. These rejection rules are what keep the score honest.

Limitations

This disclosure describes an unbuilt system. No prototype has been constructed, no accuracy figure has been measured, and every threshold in it is a proposed value. The 0.5 volt control-pilot shift threshold, the 50 percent and doubling temperature-rise thresholds, the 35 and 50 kelvin absolute rise thresholds, the 30 second scoring interval, the 50 kilobyte model size, the 5 millisecond inference time, the three-vehicle fleet correlation rule, and the 10,000 mating-cycle flag are design targets, not measured results.

No false-positive or false-negative rate is known, because no system has been built and no connectors have been driven to failure under observation. The control-pilot channel is the weakest channel in the system: a 0.77 volt shift from 300 ohms of added contact resistance sits inside the plus or minus 1 volt state tolerance, so the channel can only trend, never convict, and the edge-timing channel needs microsecond-resolution hardware the disclosure does not specify. Temperature is confounded by sun, ambient, cable heating, and the vehicle's own battery thermal state, and the per-connector baseline needs several sessions at the same connector before it means anything, which a driver who never visits the same station twice will never build. The fleet correlation needs adoption: three vehicles reporting the same connector requires three participating vehicles to have used that connector, and a rural station may never see three. False red warnings carry their own hazard, since they refuse a charge a driver may need to get home, and repeated false warnings would train drivers to ignore the system. The labeled training dataset the classifier needs does not exist yet at scale. Collecting it requires instrumenting vehicles, charging them at degrading connectors, and labeling the failures by teardown. Connector failures are rare events, so the dataset would suffer severe class imbalance. That collection effort is the largest barrier to building this system.

Sensor error is unbudgeted. An inlet temperature sensor measures temperature at the inlet, not at the contact interface inside the handle, and it lags behind contact temperature during rapid heating. At low currents the rise-per-amp-squared ratio becomes noisy, since a small temperature error is divided by a small current squared, so the scoring unit would suspend the temperature channel below 6 amps. No accuracy, drift, or dropout figures exist for the sensors in this design. Station identity is a failure mode of its own: an EVSE identifier that changes after a hardware swap, a stall renumbering that is never entered, or two connectors sharing one identifier all corrupt the per-connector baselines. Privacy is a second failure mode: station identities plus timestamps can reveal travel patterns even with GPS coordinates removed, and the opt-in must say so plainly. Finally, the system predicts connector failure, not charging failure. Payment faults, network outages, and unresponsive screens cause more failed charging visits than connectors do, and no connector gauge fixes a charger whose screen is dead.

Strongest Case Against

The honest objection is that a temperature sensor in the handle already solves this problem, and vehicle-side inference is an elaborate workaround for the many handles that were built without one and will not be retrofitted. Tesla puts NTC thermistors in its handles, Workersbee fits them into every CCS1, CCS2, GBT, and NACS connector it makes, and high-power plugs carry two PT1000 sensors at the power contacts. A sensor at the contact measures the thing itself, while this system measures a 0.77 volt shift inside a 1 volt tolerance and a temperature at the inlet that lags the contact by an unknown amount. A system that needs fleet correlation across three vehicles to convict a connector is a system that cannot protect the first driver, and the first driver is the one who melts the adapter. No test of this system exists, so that objection stands as a reason not to build it. The argument for building it is a hypothesis: that most handles will never get a sensor, because the stations that need monitoring most are the ones whose operators spend least on maintenance. No data in this disclosure tests that.

A second objection is the wall receptacle. The most common cause of Level 2 derating is not the vehicle connector at all: one repair guide's tallies put NEMA 14-50 receptacle thermal derating at 55 percent of Level 2 derating cases, against 25 percent for J1772 or NACS pin oxidation. This system measures the inlet, and it cannot see the wall receptacle at all. A driver who trusts the connector score while the receptacle behind the wall is melting is worse off than a driver who checks the receptacle. A third objection is behavioral: a green connector score could give false confidence, and false confidence at 500 amps is worse than no gauge at all, because the driver who trusts the gauge stops inspecting the handle. That objection also stands. If a gauge caught some melting connectors early, it would beat a gauge that only works at stations with new hardware. But that is conditional on the gauge working, which is unproven, and on nobody mistaking the gauge for a safety device, which nothing in this disclosure can guarantee.

What a Driver Can Do Now

None of this needs to be built before it helps. Look at the pins on every public handle before plugging in: dull gray or black oxidation rings, grit, or bent pins mean a connector that has been abused, and a handle that looks burned, melted, or smoking has already failed and must not be touched or used. After a session, touch the handle and the inlet with the back of your hand only if the handle looks undamaged. Warm is normal. Too hot to hold is not, and a handle that is too hot to hold at 32 amps is a connector that should be reported, not used again. If a charger throttles your session to half speed or flashes red or amber, try another stall before assuming the car is at fault, and dial the charge current down to 20 amps: if the session then completes without error, heat at the wall receptacle or the handle is the likely cause, and that receptacle should not be used again until an electrician has inspected it. Check recent PlugShare check-ins before driving to a station you have never used, and leave a check-in of your own when a stall fails, since the next driver's only warning today is the last driver's report. Report a broken or melted connector to the station operator immediately, and do not use a stall whose handle is visibly damaged, no matter how empty the station is. At home, hardwire a Level 2 charger instead of plugging it into a NEMA 14-50 receptacle, since hardwiring eliminates four high-current friction blade connections and removes the most common failure point in residential charging. If a plug-in install is unavoidable, use a commercial-grade receptacle, torque the connections to spec, and check the faceplate for brown heat discoloration every few months. Carry the adapter the car came with, keep it clean, and inspect its pins the same way.

Know someone who charges at public stations and has never looked at the pins on the handle? Send them this.

Prior Art References

  1. Control Pilot Signal Explained: States and PWM (Klitv): control pilot is a plus or minus 12 volt, 1 kilohertz PWM signal through a 1 kiloohm source resistor; States A through F with 2.74 kiloohm, 882 ohm, and 246 ohm loads; duty cycle 6 to 51 amps equals duty percentage times 0.6, 50 percent means 30 amps, 96 percent means 80 amps, 5 percent signals digital communication
  2. CP and PP Signals in EV Charging (Feyree): control-pilot handshake sequence with 2.74 kiloohm State B and 882 ohm State C loads; proximity pilot reads 150 ohms seated and 68 ohms with the release button pressed under SAE J1772, and the IEC 62196 ladder uses 1500, 680, 220, and 100 ohms; returning to State B opens the station contactor within 100 milliseconds; the 1 kiloohm series resistor is a mandatory safety element
  3. J.D. Power 2025 U.S. Electric Vehicle Experience Public Charging Study (August 13, 2025): 14 percent of public charging visits ended without a charge, down from 19 percent in 2024; the charger being out of service or not working properly accounted for 60 percent of failed visits; 7,428 owners surveyed January through June 2025
  4. EV Charger Uptime: A Maintenance Plan That Keeps Stations Working (evfc): 2022 field study of 657 DC fast charging connectors at 181 stations in the Greater Bay Area found 476 functional (72.5 percent), 149 nonfunctional (22.7 percent), and 32 with a cable too short to reach the inlet (4.9 percent); federally funded chargers must average above 97 percent uptime per port under 23 CFR 680
  5. CCS1 and NACS Charging Cable Safety: Managing Temperature Rise, Derating, and Connector Risk (MIDA): connectors standardized through SAE J3400, up to 1000 volts DC, evaluated to UL 2251; contact resistance is the dominant hotspot, grows with mating cycles, contamination, and corrosion, and is invisible until it produces heat or failure; thermal sensing converts connector temperature into derating, alarms, and shutdown
  6. 500A CCS1 Charging Cable, Liquid Cooled HPC Plug (MIDA): 500 amp, 500 kilowatt liquid-cooled CCS1 plug with two PT1000 temperature sensors at the power contacts, 0.5 milliohm maximum contact resistance, terminal temperature rise under 50 kelvin, designed for more than 10,000 charging cycles
  7. Why Workersbee's EV Charging Connectors Use Temperature Sensors for Safer Charging: NTC thermistors in CCS2, CCS1, GBT AC and DC, and NACS AC and DC connectors; two-level response of current reduction past a warning threshold and charging stop past a critical limit
  8. J1772 Handle and Connector Temperature Protection (OpenEVSE): Tesla charging handles carry NTC thermistors used to sense high temperatures and shut off; Tesla is the only maker the contributors were aware of implementing temperature cutoff in the J1772 handle; a few cases of a handle or connector melting
  9. J1772 Adapter Melted (Tesla Motors Club, 2019): J1772 adapter melted to a public Level 2 charging receptacle; discussion attributes the cause to a dirty or worn-out socket that was borderline acceptable for other vehicles but could not carry a Model 3 drawing the full 32 amps
  10. EV Level 2 Charger Flashing Amber or Red and Cutting Charge Speed in Half (WhyIt'sBroken): 15 milliohms of contact resistance dissipates 24 watts of heat inside the handle at 40 amps; the guide's tallies put NEMA 14-50 receptacle thermal derating at 55 percent of Level 2 derating cases and J1772 and NACS pin oxidation at 25 percent; hardwiring eliminates four high-current friction blade connections
  11. US 11,117,479: Temperature-monitored charging system for transmitting electric charge currents; temperature sensors in the plug-in connector interrupt charging or reduce current past a limit; IEC 62196-3 limits temperature rise to 50 kelvin; 200 amps continuous uncooled, up to 400 amps in cooled systems
  12. Published application US 20230088001: Protected charging connector; temperature sensors at several positions on a liquid cooling circuit, charge current controlled from temperature differences between sensors; temperature measurements usually too slow to serve as a safety mechanism
  13. Published application US 20150054462: Method for operating a charging connection device for electric vehicles; temperature inside the housing continuously detected, maximum permissible charging current reduced at a lower threshold and cut off at a higher threshold
  14. US 12,083,911: Heavy-current charging cable for charging an electric vehicle; liquid-cooled cable for charging above 250 amps, 500 amps at 1000 volts for 500 kilowatts; non-cooled cables used up to 250 amps
  15. US 12,296,703: Electrical vehicle charging system for charging an electrical vehicle; liquid-cooled cable arrangement; contact resistance between power contacts typically a few percent of conductor resistance; 20 to 30 degree Celsius gradient along power contacts at 500 amps; IEC 62196-1 limits terminal temperature rise to 50 kelvin
  16. LITF-PA-2026-091: EV charger grid impedance tomography; related charging infrastructure sensing, distinct target and no connector contact measurement
  17. LITF-PA-2026-161: EVSE voltage droop and transformer loading estimation; related charging electrical sensing, grid-side rather than connector-side
  18. LITF-PA-2026-182: EV 12 volt auxiliary battery failure prediction; related vehicle-side electrical prognostics, distinct system and failure mode
  19. ISO 15118 SECC Module (EVerest): DIN SPEC 70121 sends the EVSEID from the EVSE manager in the session setup exchange; ISO 15118-2 and ISO 15118-20 define the session setup, authorization, and charge loop exchanges between the vehicle and the charging station