Why a DC session stops at PreCharge or CableCheck
Two steps sit between a finished negotiation and the first ampere. Both are electrical tests in message form, and the values in them, not the response codes, tell you what went wrong.
Electrical tests in message form
A DC session can negotiate perfectly, agree on every limit and still never deliver a single ampere. Between the parameter discovery and PowerDelivery Start sit two steps that are really electrical tests. In CableCheck the charger checks the insulation of its output and the cable, up to the car's open contactors. In PreCharge it brings its output to the battery voltage, so the car can close its contactors without a surge.
The messages only report on what the hardware does. That is why the response codes usually read OK while the session is failing, and why the useful information sits in the values. A DC session, message by message covers the whole sequence, and Isolation monitoring and CableCheck the measurement behind CableCheck.
Stuck in CableCheck
CableCheck repeats while the charger answers EVSEProcessing Ongoing, and ends with Finished or with a refusal. When it never ends, three causes are worth checking first:
- The car is still in state B. The charger applies its test voltage only once the car has switched to state C, so a car that sends CableCheckReq from state B waits until a timer ends the session. A Control Pilot log settles this in seconds.
- The insulation monitor sees a fault or a warning: a wet connector, a damaged cable, high capacitance to earth on the car side. In DIN SPEC 70121 and ISO 15118-2 the responses carry EVSEIsolationStatus in DC_EVSEStatus. DC_CableCheckRes of ISO 15118-20 has no such field, so there the charger's log is the only witness.
- Two insulation monitors disturb each other, the car's and the charger's measuring at the same time. The tell is a warning with one car only, on a charger that passes with others.
The Timing tab shows how long the loop ran. The car's CableCheck timer is not in the limit profile, so compare the loop length with it by hand.
Stuck in PreCharge: read the shape of the voltage
PreCharge is a loop as well. The car sends its target voltage, the charger answers with EVSEPresentVoltage, and when the two are close enough the car closes its contactors and asks for PowerDelivery Start. The shape of EVSEPresentVoltage over the loop tells you most of what you need:
| What the voltage does | What it suggests |
|---|---|
| Flat, far below the target | An output capped below the target, or a reported value that is not a live reading |
| Climbing, but too slowly | Too little pre-charge current or slow regulation; the car's timer runs out first |
| Settles a few volts outside the tolerance | A calibration offset between the charger's measurement and the car's |
| Reaches the target, but no PowerDelivery follows | The car waits for something else; read the status fields in its requests |
DC-01 does this reading for you. It takes the last PreCharge response before the session ended, reports how far the output was from the target against dc.prechargeToleranceV, and says whether the loop outlasted dc.prechargeTimeoutS.
The precharge-fail demo shows the first row in its purest form. The car asks for 400 V; the charger answers 312 V forty times in a row, every 200 ms, always with OK, and the car gives up after about eight seconds. 400 V lies well inside the 150 to 1000 V the charger announced a few messages earlier, so the charger promised something it did not deliver, or reported something it did not measure. Worked example: pre-charge stalls at 312 V walks through the capture.
Measure where the messages cannot
Every number in these messages is one side's measurement, reported to the other. When the charger says 312 V, that is what the charger believes. The only way to settle a pre-charge dispute is an independent measurement: the charger's own log of its DC output, or a meter at the connector, with a timestamp you can line up with the capture.
- The output really stops short: look for what caps it, such as a power module range or a voltage limit lower than the range the charger announced.
- The output reaches the target but the message does not: the measurement path is broken, or the value is not refreshed.
- Both agree and the car still waits: compare with the car's own reading where its request carries one, EVPresentVoltage in ISO 15118-20. The pre-charge current is small, so a lasting difference between the two readings is calibration, not loss in the cable.
Who owns the fix
- A car in state B during CableCheck: the car.
- Insulation warnings and faults: whoever owns the leak, and only a measurement finds it. Start with the cable and the connector.
- An output that does not reach a target inside its own announced range: the charger. EVSEMinimumVoltage and EVSEMaximumVoltage are a promise, and a charger that cannot keep it should announce less.
- A reading a few volts off: whichever side is miscalibrated, which again only a meter tells you.
dc.prechargeTimeoutS (7 s) and dc.prechargeToleranceV (20 V) are flagged for verification in the limit profile, and the CableCheck timeout is not in it at all. Check them against the edition of the standard you test to.
DC session stalls in CableCheck or PreCharge lists the causes of both stalls with the fields that confirm each one.