The Solar EV Diversion Audit: Micro-Generation Thresholds vs. EV On-Board Charger Efficiency

The Conflict Between Variable Generation and a Fixed Protocol Floor

Residential solar EV diversion is routinely marketed as the most elegant closed-loop energy strategy available to a UK household: photovoltaic generation that would otherwise be exported at a marginal Smart Export Guarantee rate is instead routed into a traction battery, displacing grid imports at the full retail unit price. On paper the arithmetic is compelling. In practice, the strategy is constrained by a protocol-level characteristic that most specifiers, domestic energy consultants and system integrators encounter only after commissioning, when an installation that modelled well in a spreadsheet begins to stutter on partly cloudy afternoons. That characteristic is the minimum charging current mandated by the IEC 61851 conductive charging standard, and its consequences ripple through every layer of the system: the control pilot signalling, the vehicle's On-Board Charger (OBC), the contactor hardware inside the wallbox, the diversion algorithm, and ultimately the net energy ledger of the household.

This audit characterises the physics and the protocol logic behind that floor, quantifies the efficiency penalty paid when an OBC is throttled to its lower operating limit, models the dynamic friction introduced by cloud-edge transients, and compares single-phase and three-phase architectures. The objective is not to dismiss solar diversion, which remains a legitimate and valuable asset-optimisation measure for correctly specified systems, but to define the engineering envelope within which it actually delivers the savings and carbon benefit that are claimed for it.

Deconstructing the IEC 61851 Control Pilot

In a Mode 3 charging arrangement, the Electric Vehicle Supply Equipment (EVSE) does not simply energise the cable and allow the vehicle to draw whatever it wishes. Instead, the wallbox and the vehicle conduct a continuous low-voltage dialogue over a dedicated conductor known as the Control Pilot (CP). This single wire, referenced to protective earth, carries both the state of the connection and the maximum current the EVSE is prepared to supply. It is the mechanism through which a wallbox with a 32 A circuit rating can safely serve a vehicle that wishes to draw 16 A, and equally the mechanism through which a diversion controller can modulate charging power in response to solar surplus.

The signalling operates on two layers. The first is the voltage level on the pilot line, which identifies the connection state. With no vehicle connected (State A) the EVSE presents a steady positive 12 V. When a vehicle is plugged in, a diode and resistor network inside the vehicle inlet pulls the positive peak down to approximately 9 V (State B, vehicle connected but not requesting charge). When the vehicle closes its own switch and places a resistance of roughly 2.7 kilohms in parallel (State C), the positive peak falls to approximately 6 V, signalling that it is ready to charge and does not require ventilation. State D, at approximately 3 V, historically signalled a vehicle requiring ventilation and is not permitted by most modern residential EVSE designs. The EVSE monitors these levels and will only close its main power contactor once State C or D has been verified.

The second layer is the pulse-width modulated (PWM) waveform that the EVSE superimposes on the pilot signal. Once a vehicle is detected, the EVSE oscillates the pilot line between +12 V and -12 V at a nominal frequency of 1 kHz. The proportion of each 1 ms cycle spent in the positive state, the duty cycle, encodes the maximum continuous current the EVSE declares available. The standard defines a precise linear relationship for the common operating range: between 10% and 85% duty cycle, the available current in amperes equals the duty cycle percentage multiplied by 0.6. A 16.7% duty cycle therefore advertises 10 A, a 26.7% duty cycle advertises 16 A, and a 53.3% duty cycle advertises 32 A. Above 85%, a different expression applies, permitting high-power DC-adjacent and heavy-duty AC currents up to 80 A, but this region is irrelevant to residential single-phase equipment.

Why the Floor Sits at a 10% Duty Cycle

The question that matters for solar diversion is why the PWM scheme cannot simply continue downward to advertise 2 A, 3 A or 4 A, thereby tracking a small surplus. The answer lies in the way the standard partitions the duty cycle range. A duty cycle of 5% is reserved as a special signal that indicates the EVSE wishes to establish high-level digital communication with the vehicle, as used in ISO 15118 and DIN SPEC 70121 sessions, rather than transmitting a current limit. The region between 5% and 10% is a protected guard band: it is explicitly not defined as a valid analogue current limit, and any vehicle receiving a duty cycle in that band is expected to treat the pilot as invalid and refuse to charge. The first valid analogue current command therefore begins at 10% duty cycle, which maps through the 0.6 A per percentage-point relationship to exactly 6 A.

This floor is further reinforced by the electrical design assumptions behind the standard. Contactors, cable thermal models, residual current detection thresholds and the vehicle's own protective logic were all characterised around a minimum meaningful continuous load. Furthermore, the duty cycle measurement tolerance, nominally within plus or minus one percentage point, is proportionally far more significant at low duty cycles than at high ones. A one percentage point error at 10% represents a 10% error in the declared current, whereas the same absolute error at 50% represents only 2%. The 6 A baseline is therefore a deliberate compromise between signalling resolution, hardware certification and interoperability across an installed vehicle base that includes thousands of models from dozens of manufacturers.

A diversion controller cannot circumvent this floor in software. A wallbox that attempts to advertise less than 6 A will either be ignored by the vehicle, which continues to draw the amount it prefers within the valid range, or will cause the vehicle to refuse the session entirely. The only legitimate action available when surplus generation falls below the threshold is to suspend charging, by holding the pilot at a steady positive 12 V or by opening the contactor, and to wait for surplus to recover.

The Resulting UK Activation Threshold

On a nominal 230 V UK single-phase supply, 6 A corresponds to a minimum apparent power draw of 1.38 kW at the charging point, calculated as 230 V multiplied by 6 A. Because the charging session must be sustained by surplus generation rather than grid import, the photovoltaic array must be producing at least 1.38 kW more than the rest of the household is consuming at that moment before charging can begin, and must continue to do so for the duration of the session. In reality the threshold is higher, because the OBC itself imposes conversion losses and because a vehicle at the 6 A limit also supports auxiliary loads such as battery thermal management, so the effective wall-side requirement frequently exceeds the nominal figure.

Quantifying On-Board Charger Efficiency Degradation

The OBC is the specialised power conversion unit inside the vehicle that rectifies alternating current from the wallbox into the regulated direct current required by the high-voltage traction battery. A modern OBC is typically a two-stage architecture. The first stage is an active power factor correction (PFC) front end, usually a bridgeless or totem-pole boost topology, which shapes the input current to follow the grid voltage waveform and holds an intermediate DC link at several hundred volts. The second stage is an isolated DC-DC converter, commonly a resonant LLC or dual-active-bridge design, in which high-frequency switching drives a galvanic isolation transformer before the output is rectified and filtered to match the battery's instantaneous voltage and charge profile.

Each of these stages incurs losses that fall into two broad categories. Load-dependent losses, principally conduction losses in semiconductor switches and winding resistance, scale with the square of the current. Fixed or quasi-fixed overhead losses, in contrast, are largely independent of the delivered power: gate-drive consumption, control and communications electronics, cooling fans or coolant pumps, magnetising current in the isolation transformer, core losses, and switching losses that persist as long as the converters are energised. The efficiency of any converter is the ratio of useful output power to total input power, and it follows directly that a fixed overhead consumes a disproportionately large share of the input when the delivered power is small.

OBCs are engineered and optimised around their rated operating points, which for the majority of European vehicles correspond to charging currents in the range of 16 A to 32 A on a single phase, or the equivalent per-phase currents on a three-phase connection. At these loads the conduction losses remain moderate while the fixed overhead is amortised across several kilowatts, and peak conversion efficiencies of 90% to 95% are achievable. At 6 A single-phase, the OBC is operating at roughly a fifth of a 7.4 kW unit's rated capacity. The resonant converter is pushed away from its optimal switching frequency, the PFC stage may enter a discontinuous or burst-mode regime, and the fixed overheads of the cooling and control systems remain almost entirely unchanged. Measured and published characterisations of production vehicles commonly show conversion efficiencies falling to between 70% and 80% in this region, with some vehicles performing better and some materially worse.

The practical consequence is that a household delivering 1.38 kW from its solar array to the charging point may store only 1.0 kW to 1.1 kW in the battery. The remaining 0.28 kW to 0.41 kW is dissipated as heat, much of which the vehicle then has to manage with its own thermal systems, adding a further parasitic draw. The effective round-trip value of that surplus is therefore substantially lower than the headline export-displacement calculation implies. For a vehicle with a 16 A to 32 A efficiency of roughly 90%, charging at the higher rate from the grid during a cheap overnight tariff window can in some cases deliver more stored energy per pound of net cost than diverting marginal solar through the 6 A floor, a counter-intuitive result that should be explicitly modelled before specification.

The matrix above should be read as indicative rather than universal. Individual OBC designs vary considerably, and a small number of vehicles using more modern wide-bandgap silicon carbide or gallium nitride switching devices exhibit flatter efficiency curves. Nevertheless, the qualitative shape is consistent across the market: efficiency is poorest exactly at the 6 A floor, which is the operating point a solar diversion system is most likely to occupy for the greater part of its active charging time on a modest array.

Solar Clipping and Cloud Macro-Cycling

The term clipping, in this context, describes the loss of otherwise harvestable solar energy caused by the mismatch between a continuous generation curve and a stepped load that can only exist in an on or off state above a minimum power. Consider an array of 3.5 kWp on a well-oriented south-facing roof. In clear sky conditions around midday in summer it may produce 2.8 kW to 3.1 kW, comfortably above the 1.38 kW floor even after allowing for a baseline household demand of 0.3 kW to 0.5 kW. Charging begins, the controller holds the vehicle at an appropriate current, and the system performs as modelled.

The difficulty emerges under cumulus cloud, which is the dominant summer condition across most of the UK. Passing cloud banks can reduce irradiance from 900 W per square metre to 200 W per square metre within seconds and restore it equally quickly. Generation therefore oscillates between, for example, 3.0 kW and 0.7 kW with a period of anything from half a minute to several minutes. Every time net surplus falls below 1.38 kW the diversion controller must decide whether to stop. Every minute that the vehicle is not charging is surplus energy that is either exported at a low rate or, in a system without export, simply curtailed. Conversely, if the controller is configured to tolerate short dips in order to avoid stopping, the household imports grid power to hold the 6 A floor, partially defeating the purpose of diversion.

The Mechanical and Digital Cost of Chasing the Cloud

If eco-throttling algorithms are misconfigured with insufficient hysteresis, delay timers or minimum run times, the system enters a pattern of macro-cycling, in which the charging session is repeatedly started and stopped. Each cycle carries a real physical and digital cost. The wallbox main contactor, a mechanical electromagnetic relay rated for a finite number of operations, closes and opens under load, with each operation causing contact erosion; mechanical endurance ratings typically span hundreds of thousands of cycles, but a system cycling every few minutes for six hours a day across a summer can accumulate tens of thousands of operations in a single season. The audible clicking that homeowners often report is not merely a nuisance but a direct indicator of this mechanical wear.

Each restart also requires a full handshake sequence. The pilot must transition from State B back to State C, the vehicle must wake the OBC, perform its pre-charge and insulation checks, close its own high-voltage contactors, and ramp its output. Depending on the vehicle, this process can take 5 to 30 seconds during which no useful energy is transferred, and a further period during which the OBC operates in its least efficient start-up regime. Battery management system (BMS) activity also increases: the BMS must reconnect the pack, re-establish cell balancing states and, in some vehicles, wake the thermal management loop. Repeated cycles of this kind are generally not catastrophic for lithium-ion chemistry, but they contribute to contactor wear inside the vehicle, as well as to unnecessary wake events that draw energy from the 12 V auxiliary battery.

A well-configured diversion algorithm mitigates the effect through several parameters: a minimum surplus threshold with explicit hysteresis, a start delay that requires surplus to be sustained for several minutes before starting, a stop delay that allows the session to ride through brief dips by importing a limited amount of grid energy, and a minimum run time that prevents rapid cycling. The trade-off is inescapable. Every second of tolerated deficit imports grid energy at the retail rate, and every second of enforced pause forfeits surplus. The optimal calibration depends on the array size, the base load profile, the local irradiance variability, the import and export tariffs, and the vehicle's own wake-up characteristics, and should be treated as a site-specific commissioning task rather than a default setting.

The Single-Phase versus Three-Phase Disconnect

A frequent assumption among installers is that a three-phase property is inherently better suited to EV charging, and for fast grid-based charging this is correct: a 22 kW supply replenishes a battery substantially faster. For solar diversion, however, the three-phase architecture introduces an unfavourable multiplier. The IEC 61851 control pilot is a single signal that applies to the vehicle as a whole, and the 6 A minimum applies to every energised phase. A vehicle with a three-phase OBC connected to a three-phase wallbox is therefore required to draw at least 6 A on each of the three phases simultaneously, giving a minimum power of 3 multiplied by 230 V multiplied by 6 A, or 4.14 kW.

This threshold is three times the single-phase equivalent and is beyond the reach of most UK domestic solar arrays during most of the year. A typical 4 kWp installation seldom exceeds 3.2 kW in practice, and after the household base load is subtracted, the available surplus will almost never reach 4.14 kW. An unmanaged three-phase diversion system would therefore spend most of its time idle, delivering negligible solar charging and relying on the grid for the remainder.

To solve this, specialised wallboxes designed for solar integration incorporate phase-switching contactors that can disconnect two of the three phases and present only a single-phase load to the vehicle. In this mode the 6 A floor reverts to 1.38 kW. Provided the vehicle's OBC supports single-phase operation, which virtually all three-phase-capable OBCs do, the controller can down-shift to single-phase operation when surplus is scarce and up-shift to three-phase operation when surplus exceeds the 4.14 kW floor, where the three-phase OBC operates more efficiently across its three parallel channels. This phase-switching introduces its own complexity. Not every vehicle tolerates a phase change within a session and some require the session to be terminated and re-established, resulting in the handshake overhead described earlier. Frequent switching must therefore be restrained with the same hysteresis discipline applied to start-stop logic.

Furthermore, a single-phase charging mode on a three-phase supply raises a separate load-balancing consideration. Drawing 6 A to 16 A continuously on one phase may unbalance the property's supply, particularly where other single-phase loads such as heat pumps, induction hobs or immersion heaters share that phase. System integrators should confirm phase allocation during commissioning and, where supported, configure the controller to rotate the phase assigned to the vehicle across sessions.

Specifier Recommendations and Engineering Implications

The analysis supports a set of practical conclusions for those specifying solar-linked EV charging infrastructure. First, the array and base load must be assessed against the 1.38 kW floor, or 4.14 kW for a three-phase system without phase switching, using measured half-hourly or one-minute generation and consumption data rather than annual yield estimates. A system that clears the threshold only briefly each day offers little diversion benefit regardless of its nameplate capacity.

Second, eco-mode configuration must be treated as a commissioning step. Hysteresis, delay timers and minimum run times should be tuned to local conditions and documented. The default settings supplied by manufacturers are frequently optimised for demonstrations in stable summer irradiance rather than for the variable conditions typical of UK shoulder seasons.

Third, the efficiency penalty at the 6 A floor should be incorporated into financial modelling. Where a household also has access to a low-cost off-peak tariff, the model should compare the effective cost per stored kilowatt-hour delivered through diverted solar, accounting for 70% to 80% conversion efficiency and any grid import required to hold the floor, against off-peak grid charging at a higher and more efficient current. For many properties a hybrid approach is optimal: solar diversion when surplus is robust and sustained, and scheduled grid charging at an efficient current to cover the remaining requirement.

Fourth, a home battery storage system changes the equation materially. A battery can smooth cloud transients, absorb short generation dips and supply the shortfall required to hold the 6 A floor, converting a stuttering diversion profile into a sustained one. The cost and round-trip efficiency of that additional storage must, however, be weighed against the simpler option of increasing the array size or adopting a smart tariff.

Finally, integrators should select hardware that exposes the control pilot behaviour, supports phase switching where the property is three-phase, and provides logging of contactor operations and session starts. These data allow macro-cycling to be detected and corrected before it produces contactor wear or customer complaints. Solar diversion is a sound engineering proposition when it is specified against the real constraints of the IEC 61851 floor, the OBC efficiency curve and local irradiance variability. It becomes a thermodynamic and mechanical liability when those constraints are ignored.