Leakage Current Detection in EV Chargers Under EMC Stress
An electric vehicle charger contains a power stage that switches at high frequency, contactors that open and close, an auxiliary supply and a communication module, all inside one enclosure. In that environment a safety function has to detect a smooth direct residual current of a few milliamperes and trip reliably. The disturbance sources are orders of magnitude larger than the signal being detected, and they are physically close to the detection circuit.
This is the part of charger design that EMC requirements test most severely. When the immunity level applied to the product rises, the question is no longer whether the protection circuit can detect a fault in a quiet laboratory, but whether it still detects that fault while the equipment is being disturbed. A detection circuit that trips correctly on the bench and misses a fault during an immunity test is a safety problem, not a compliance detail.
A Milliampere Signal in a Noisy Enclosure
A type B residual current device has to identify a smooth direct residual current, and the trip threshold for that quantity is measured in milliamperes. A representative device might be specified with an actual trip current between 4.0 and 6.0 milliamperes and a typical value near 5.1 milliamperes, against a standard operating range from 3 to 6 milliamperes. That does not leave much room for error, and the signal has to be extracted from a measurement at the sensor while power circuits are switching nearby.
The difficulty is not the magnitude of the threshold but the ratio between the signal and the interference. A cable carrying a serial link and a power module switching at tens of kilohertz both produce fields inside the enclosure, and the sensing path is a low level analogue channel. Once interference reaches the front end, no amount of digital filtering recovers the original information, so the design effort belongs in the sensor, the wiring, the grounding and the supply.
Where the Interference Enters the Signal Chain
The coupling paths are the usual ones, with different weights in a charger. Radiated fields from the power stage couple into the loop formed by the sensing cable and its return. Parasitic capacitance between the power switching node and the sensing circuit injects displacement current. Common impedance in the ground connection converts power stage return current into a voltage that appears in series with the sensor output. And conducted noise on the auxiliary supply reaches the conditioning circuit through its own power pins.
The signal path determines which of these dominates. A current transformer output is a low level voltage that is sensitive to series pickup along its cable and to the loop area that pickup encloses. Lengthening that cable, routing it near the power stage or sharing a connector with switching signals all increase the exposure, and the resulting common mode voltage can saturate the front end even when the differential signal is intact.
Comparing Split and Integrated Architectures
In a split design the current transformer produces a low level analogue output that travels on a cable to a separate conditioning board, where it is amplified, filtered and converted before the controller makes the trip decision. The advantage is flexibility in mechanical arrangement; the disadvantage is that the most sensitive signal in the product is routed outside the sensor, where cable length, routing and shielding all affect the result.
An integrated architecture moves the conditioning and the threshold decision into the sensor module and delivers a digital output to the controller. The analogue signal path is short and enclosed, and the interface to the rest of the system is a logic level that is far less susceptible to coupling. This does not make the function immune to interference, but it reduces the length of the exposed path and therefore the number of ways interference can enter. The choice between the two still depends on the mechanical arrangement, the power circuit layout and the outcome of the EMC test.
Reading Trip and Response Specifications
A single trip current is not enough to characterise the function. The action value has a spread between units and a dependence on temperature, so a typical figure near the middle of the range does not indicate margin. A device with a typical value of about 5.1 milliamperes within a 4.0 to 6.0 milliampere spread, against a standard range of 3 to 6 milliamperes, has to hold its behaviour across an operating temperature range that may extend from minus 40 to plus 85 degrees Celsius, and a laboratory measurement at room temperature is only the starting point.
Response time is read the same way. A single figure such as 40 milliseconds is sometimes treated as a universal requirement, but the standard specifies different break times for different residual current magnitudes: a current at the threshold allows a longer time, while a current at fifty times the threshold must be interrupted far faster, within a few tens of milliseconds. The action value and the action time have to be evaluated together for each test condition rather than as two independent numbers.
Self Test and Long Term Reliability
A protection function that never operates in normal service has no opportunity to reveal that it has failed, which is why a self test is required at every power up. A typical implementation clears the detection circuit and then injects a simulated residual current internally to confirm that the chain responds, producing a status output for the duration of the check. The main circuit must not be closed while the clearing and self test are in progress, because residual current present in the line would corrupt the zero reference.
The self test verifies the detection chain, not the whole installation, and it does not replace an end of line functional test. Its value is in correcting zero drift and in confirming at each start-up that the sensing path and the output are working. Where the assembly is expected to survive a contaminated environment, the design should also account for the surface leakage paths that humidity and flux residue create, as described in PCB cleaning, and the layout discipline that keeps low level analogue wiring away from switching circuits follows the same reasoning as mixed signal design guidelines and ground current control.
Testing the Function Under Disturbance
The detection chain should be verified while the product is being disturbed, not only in a quiet state. That means running the immunity tests with the protection function active and monitoring both the trip output and the internal status, so that a false trip and a missed trip are both detected. A missed trip is the more dangerous result and is also the easier one to overlook, because the product appears to pass the rest of the test.
The measurement should be repeated at the extremes of the operating temperature range, since component tolerances and the behaviour of the sensor both change with temperature. Where the trip threshold is adjustable, the setting should be chosen to give margin at the temperature where the sensitivity is lowest rather than at room temperature, and the zero point should be re-established by the self test at each power up to remove drift accumulated between cycles.
FAQ
Why is leakage current detection harder than a simple overcurrent measurement? Because the quantity being measured is a few milliamperes of direct current in a path exposed to radiated and conducted interference, while the surrounding circuits switch amperes. The signal to interference ratio at the sensor, not the threshold value, is the design problem.
Does an integrated sensor remove the need for EMC design? No. It shortens the sensitive path and reduces the exposure, but the sensor still needs a clean supply, a controlled ground and separation from power wiring. Electromagnetic immunity is a property of the whole installation.
How should the auxiliary supply be treated? As a potential injection path. Filter it at the module, keep its return separate from the sensing return, and verify with the power stage operating under load rather than in a passive state.
