Worked example: discharge above the target in ISO 15118-20
The bpt-limit demo read step by step: an ISO 15118-20 bidirectional DC session in which the car asks for 20 A of discharge and the charger draws 32 A for seven seconds, beyond the 30 A both sides announced, until the car ends the transfer.
What you see
- Load the demo Discharge above target (bpt-limit): 15.9 s of an ISO 15118-20 DC session with bidirectional power transfer, over plain TCP as test benches record it with TLS switched off.
- Energy mode DC-BPT. Every stage from SLAC to Stop passed.
- Two errors, IS20-03 and IS20-04, and two warnings, HS-03, for ISO 15118-20 offered and selected without TLS. The warnings are the bench setup, not the fault.
The verdict
Completed, Completed normally, confidence high: the charger answered SessionStop at 15.8 s after 11.1 s of energy transfer. The verdict is about how the session ended, and it ended in order. What went wrong during it is in the chain of evidence, which holds IS20-03 and IS20-04.
Reading the capture step by step
- EVSE: ServiceDiscoveryRes, 2.220 s: DC and DC_BPT offered
- EV: ServiceSelectionReq, 2.280 s: DC_BPT, parameter set 1
- EV: DC_ChargeParameterDiscoveryReq, 2.400 s: BPT DC, EVMaximumDischargeCurrent 30 A, EVMaximumDischargePower 11 kW
- EVSE: DC_ChargeParameterDiscoveryRes, 2.460 s: BPT DC, EVSEMaximumDischargeCurrent 30 A, EVSEMaximumDischargePower 11 kW
- EVSE: ScheduleExchangeRes, 2.600 s: Scheduled
- EV: PowerDeliveryReq, 3.800 s: ChargeProgress Start, BPT_ChannelSelection Discharge
- EV: DC_ChargeLoopReq, 4.100 s: BPT Scheduled, EVTargetCurrent -20 A, EVTargetVoltage 398 V
- EVSE: DC_ChargeLoopRes, 4.142 s to 7.142 s: EVSEPresentCurrent -20 A, four times
- EVSE: DC_ChargeLoopRes, 8.142 s to 14.142 s: EVSEPresentCurrent -32 A, seven times
- EV: PowerDeliveryReq, 15.200 s: ChargeProgress Stop
- EV: SessionStopReq, 15.700 s: ChargingSession Terminate
- EVSE: SessionStopRes, 15.760 s: OK
- A negative current flows out of the car. The car asks for -20 A in every one of the eleven loops, and for the first four the charger reports exactly that.
- From 8.142 s on the charger reports -32 A while the request stays at -20 A. That is 12 A more discharge than asked for, and 2 A more than the 30 A maximum both sides announced in DC_ChargeParameterDiscovery.
- IS20-04 compares the size of the current with the size of the target. It allows 5 % plus 0.5 A over the target, so 21.5 A here, and gives the charger 5 s to get back under it. The sixth response in a row above that, at 13.142 s, 5 s after the first, raises the finding: the car asked for -20 A and the charger reported -32 A for more than 5 s.
- IS20-03 compares a discharge with the discharge limits both sides announced: EVSEMaximumDischargeCurrent and EVMaximumDischargeCurrent (30 A each), EVSEMaximumDischargePower and EVMaximumDischargePower (11 kW each). With the same margin of 5 % plus 0.5 A, 32 A is exactly at the edge of the 30 A limit, so the current alone passes. The power does not: 32 A at 398 V is 12.7 kW, over the 11 kW both sides announced plus the margin, so IS20-03 reports each of the seven responses, collapsed into one finding.
- IS20-05 passes, because DC_BPT was selected, so power may flow back. IS20-02 finds the control mode Scheduled in all eleven loops, which is what makes IS20-04 apply: in dynamic mode the charger sets the operating point.
- The car ends the transfer at 15.200 s, after one more loop. SessionStopReq carries no termination code, so the capture does not say why.
pw.currentMarginPct (5 %) and pw.currentMarginA (0.5 A) are lab practice and flagged for verification in the limit profile; pw.settlingTimeS (5 s) is the time IS20-04 lets the charger follow a new target. Limit profiles: the numbers behind every check explains how to change them.
What the cause is
- The charger does not hold the discharge current at the target of the car. In scheduled control mode the car names the target, and the charger follows it within its limits.
- The overshoot is steady at 32 A from one loop to the next, which points at the control of the charger, for example a current limit that is set too high for discharge or a current loop that works for charging but not for the reversed direction.
- The likely party is the charger: it reports the current it is drawing, and that current is above the target.
What to fix
- Check the discharge control of the charger: which setpoint it uses in BPT scheduled mode, and whether it takes EVTargetCurrent with its sign or only its size.
- Clamp the discharge current to the smallest of the target of the car, EVMaximumDischargeCurrent of the car and EVSEMaximumDischargeCurrent of the charger.
- Measure the DC current at the output against EVSEPresentCurrent, to rule out a measurement that reports more than flows.
- Bidirectional power transfer in ISO 15118-20, message by message follows a bidirectional session that runs as it should.
How to confirm the fix
- EVSEPresentCurrent stays within 21.5 A of discharge for a -20 A target in every DC_ChargeLoopRes, so IS20-04 passes, and the discharge power stays under 11 kW, so IS20-03 passes.
- The Curve tool shows the discharge current flat at the target, never below -30 A.
- HS-03 keeps warning as long as the bench runs without TLS. In the field ISO 15118-20 runs over TLS 1.3, and HS-03 passes.
Sources
- ISO 15118-20:2022, DC_BPT service, BPT_DC_CPDResEnergyTransferMode, DC_ChargeLoop in scheduled control mode
- SPEC.md of this tool, FR-ENC-05, the failing demo sessions