Residual / Leakage Detection for EV High-Voltage Systems
← Back to: High-Voltage Energy & Safety
This page is where I pull together everything I need to plan residual and leakage detection in an EV high-voltage system – where to place it, which sensing path to use, how to set thresholds and response times, and how to choose ICs and PCB details so the function is safe, diagnosable and easy to source.
What is Residual / Leakage Detection in an EV HV System?
In an EV high-voltage system, residual current is the difference between the current that leaves the source and the current that returns. If some of that current finds an unintended path through the vehicle body, cables or the environment, it becomes leakage current. Residual and leakage detection makes these unwanted currents visible so they can trigger a safe shutdown.
AC residual current is typically associated with mains-frequency and harmonic components flowing from line to protective earth or chassis, while DC leakage current is more related to steady or slowly varying currents from the HV DC rails to the vehicle body. Many real systems contain a mix of AC and DC components, so the sensing solution must match the expected waveform and magnitude.
On-board chargers and EV supply equipment are usually designed to comply with standards such as IEC 62752 and IEC 62955 on top of the more general IEC 61851 framework. These documents define trip thresholds, frequency bands and response times for residual and leakage currents, and they strongly influence how you select the sensing topology, AFEs and safety logic on this page.
This page focuses on current-based residual and leakage detection, where CTs, Hall sensors or shunt-based AFEs measure the unwanted current directly. If you need to analyse insulation resistance and impedance-based monitoring in detail, you can complement this topic with a separate Insulation Monitoring Device (IMD) page.
Where Residual / Leakage Detection Sits in the HV Energy Backbone
Residual and leakage detection is not a single part sitting in isolation. It is woven into your HV energy backbone wherever a dangerous current could flow from high-voltage conductors into the vehicle body, ground or user-accessible parts. Typical deployment points include the AC inlet and on-board charger, the DC link and pack shell, and sometimes the high-voltage junction box inside the vehicle.
On the AC charging side, a zero-sequence CT around the line and neutral conductors near the EV inlet or OBC input detects residual AC current leaving the intended path. In the DC path, leakage detectors monitor current from the HV DC rails to the pack enclosure or chassis, especially in DC fast-charge and internal HV distribution. In some architectures, additional sensors inside the BJB / HV junction box supervise branch circuits that are exposed to harsh environments or long cable runs.
This page covers the parts of the backbone where leakage is detected via current sensing: CTs, Hall sensors or shunt-based AFEs plus comparators and isolation links. An insulation monitoring device (IMD) that injects a signal and measures insulation impedance is usually present in the same HV network, but it belongs to a dedicated topic and page. In the system diagrams below, IMDs are drawn with dashed outlines and referenced as “see Insulation Monitoring Device page” so you can keep both roles visually connected without mixing the design rules.
Sensing Technologies for Residual / Leakage Current
Once the need for residual and leakage detection is clear, the next question is which sensing technology actually fits your EV project. The choice depends on whether your risk is dominated by AC mains and harmonics, slow DC leakage, mixed waveforms or cost and integration constraints. In practice, most designs end up using a small number of well-proven options: zero-sequence current transformers, Hall-based leakage sensors, shunt plus differential AFEs, or dedicated residual current monitor ICs.
Zero-sequence and differential CTs are the classic solution for AC residual current at the inlet of an OBC or EVSE. They have excellent bandwidth for mains and harmonic content and introduce no insertion loss, but they are inherently blind to pure DC. If the standard requires response to DC components, you need a suitable RDC-DD variant and must pay attention to CT saturation, burden selection, rated residual current range, bandwidth and insulation strength.
Hall-effect leakage sensors extend coverage into the DC region and are often used on DC links and pack shells where DC leakage is a major concern. They can sense a mix of AC and DC but bring their own design challenges: offset and drift over temperature, noise, finite bandwidth and sensitivity to common-mode disturbances. A clean layout, proper filtering and calibration strategy decide whether a Hall-based solution is robust enough for the intended safety function.
Shunt plus differential AFE approaches are attractive when integration, cost or diagnostics drive you toward a more flexible solution. They let you combine precise gain, filtering and digital conversion, but at low leakage levels the design is constrained by microvolt-level offset and noise, CMRR limits and high dv/dt stress from nearby power stages. Getting the RC network and AFE bandwidth right is key to avoiding both missed trips and nuisance trips.
When standards compliance and response behavior must be tightly controlled, dedicated residual current monitor ICs can consolidate the AFE, digital filtering, comparators and sometimes built-in self-test into a single device. These parts are usually designed with IEC residual and leakage requirements in mind and can simplify safety assessment, at the cost of less flexibility and a heavier reliance on device-specific capabilities. The overview below links each sensing technology to the use cases where it tends to work best.
Trip Thresholds, Frequency Bands and Response Time
Selecting a sensing technology is only half of the job. Residual and leakage detection has to trip at the right current, in the right frequency band and within a defined time window, otherwise the safety function is either ineffective or constantly generating nuisance alarms. Instead of starting from standard paragraphs, it is more practical to start from a few engineering questions: how many milliamps or amps must I detect, what waveforms do I expect and how fast must the system react?
The trip threshold defines the leakage level at which your system must enter a safe state. For AC residual protection near the inlet, thresholds are often in the tens of milliamps range, while DC leakage protection deeper in the HV network may use different limits. These thresholds translate directly into the required sensitivity, noise floor and usable dynamic range of your CT, Hall or shunt plus AFE chain.
The frequency content determines how wide your detection bandwidth must be. Mains-frequency and low-order harmonics require a clean low-frequency response, whereas inverter-related ripple and EMI must be suppressed enough to prevent false trips. DC leakage and slowly varying currents stress the offset, drift and low-frequency linearity of the AFE. In practice, you end up trading bandwidth against noise and dv/dt immunity.
Finally, response time ties the sensing chain to contactor dynamics and the overall functional safety concept. Fast paths built around comparators and well-defined filters can react in a few milliseconds, while MCU-based processing may add latency but enables more advanced discrimination. The faster you trip, the more carefully you must control noise and transients to avoid nuisance trips, especially in harsh EMC conditions.
Once the required leakage level, frequency band and response time are mapped out, they become concrete requirements for your AFEs, filters, comparators and isolation links. The cheat sheet below visualises how AC residual protection, DC leakage protection and nuisance trip regions occupy different parts of the leakage-versus-time plane.
Residual / Leakage Detection Signal Chain Architectures
Choosing a sensing technology and planning thresholds is only useful if the signal chain can support them. In an EV high-voltage system, residual and leakage detection typically follows one of three patterns. A fast analog path uses CT or Hall sensors, an analog AFE and a threshold comparator that drives safety logic directly. A more flexible digital path adds a sigma-delta or precision ADC and lets the MCU or DSP implement thresholds and filtering. In larger programs, a dedicated residual current monitor IC can integrate most of the front-end and hand a clean status signal to the safety MCU.
In the CT or Hall plus analog comparator architecture, the focus is on fast and predictable reaction. The sensor and AFE provide the required gain and filtering, while a hardware comparator enforces the trip threshold and latching. The safety MCU mainly records events and coordinates system actions. This approach gives you millisecond-level response times and a clear hardware safety path, but self-test and diagnostics for the sensor, AFE and comparator must be designed explicitly.
In the CT or shunt plus sigma-delta or ADC architecture, the leakage signal is digitised and processed by an MCU or DSP. This enables digital filtering, waveform capture and adaptive thresholds that can be tuned over the vehicle life. It is attractive when you need rich diagnostics or want to distinguish true leakage from noisy conditions. However, the overall response time and safety case now depend on software execution, scheduling, watchdog concepts and how well the safety-related code is verified.
A dedicated residual current monitor IC can wrap most of the analog front-end, filtering, comparators and even test injection into a single device, often with behavior tailored to IEC residual and leakage requirements. The safety MCU then focuses on reading status, logging events and coordinating contactor control. This simplifies system integration and safety assessment, at the cost of reduced flexibility and a stronger dependency on a specific IC family. The block diagram below highlights how each architecture partitions work between sensors, AFEs, the MCU and a possible dedicated RCM IC.
Built-In Self-Test, Diagnostics and Functional Safety Hooks
A leakage detection chain is only useful if you can trust it. Built-in self-test and diagnostics are needed to prove that the sensor, AFE, ADC and comparators are still working and that the trip path has not silently failed. In a safety-rated EV program, you typically combine a power-on self-test with periodic tests during operation and simple plausibility checks that run continuously in the background.
On the sensor side, a common approach is to inject a known test signal. For CT-based solutions, this can be a dedicated test coil that drives a small AC current, allowing the system to verify gain, polarity and saturation margins. Hall and shunt-based designs can use internal bias current sources or DAC-driven test voltages to check the AFE and conversion path. If the measured response falls outside a defined window, the safety MCU can flag a diagnostic fault instead of waiting for a real leakage event.
Around the AFE, ADC and comparator, self-test and monitoring focus on offset, saturation and stuck outputs. Periodically checking for unrealistic DC offsets, codes stuck at full scale or comparator outputs that never toggle helps to catch open circuits, short circuits and internal failures. Some dedicated residual current monitor ICs provide built-in test modes and status flags for these conditions, which can simplify the safety analysis compared with a fully discrete implementation.
From a functional safety perspective, each test mechanism contributes to diagnostic coverage and must be mapped to the faults it can detect. Power-on tests are good at finding manufacturing or latent faults before the HV system is armed, while periodic tests and runtime monitors help control residual and latent faults during driving or charging. The safety concept then defines which faults require an immediate safe state and which can be handled through controlled degradation or delayed service actions.
For leakage detection specifically, the chain usually ends at a request to open contactors or block new HV activations. The detailed weld-detection logic, contactor redundancy and HV energy backbone reconfiguration are handled on a dedicated BDU and contactor weld detection page, so this section stays focused on how the leakage sensing path proves its own integrity and delivers trustworthy trip and diagnostic signals to the rest of the safety architecture.
IC Selection Map: AFEs, Monitors and Isolation Links
Once the signal chain architecture is clear, the next step is to translate it into concrete IC types that I can put on a sourcing list. For residual and leakage detection in an EV HV system, I usually split the problem into a small number of functional blocks: low-noise differential or isolated amplifiers, sigma-delta modulators and precision ADCs, comparators with well-defined thresholds and latching behavior, dedicated residual current monitor ICs, and the isolation links and power supplies that connect everything to the rest of the vehicle electronics.
Low-noise differential and isolated amplifiers form the front-end for shunt and CT-based leakage sensing. They must handle small signals on top of noisy HV environments, so I look for high gain accuracy, low offset and drift, good CMRR and sufficient bandwidth across the expected leakage spectrum. When the sense element is on the high-voltage side, isolated amplifiers or isolated modulators with defined isolation ratings and creepage distances become the starting point for the architecture.
Sigma-delta modulators and precision ADCs are the backbone of digital leakage detection paths. They convert the analog leakage signal into a bitstream or code that a microcontroller or DSP can filter and evaluate. Here I care about effective number of bits, input bandwidth, noise, latency and how the digital interface lines up with my safety MCU. For fast, hard-wired reactions, I add precision comparators with latching outputs, focusing on threshold accuracy, propagation delay, hysteresis options and whether the device supports wired-OR or open-drain outputs into a safety monitor.
Dedicated residual current monitor ICs combine many of these functions and are designed specifically for residual and leakage detection, often with behavior aligned to IEC residual-current standards. When I use such a device, the safety MCU reads status bits, trip flags and diagnostic information instead of implementing every detail in discrete AFEs and firmware. Finally, digital isolators and isolated DC/DC converters close the loop between the high-voltage sensing front-end and the low-voltage domain, so I specify isolation rating, CMTI, propagation delay and power start-up behavior as explicit requirements.
Across the seven major vendors I usually work with — TI, ST, NXP, Renesas, onsemi, Microchip and Melexis — leakage detection can follow two main sourcing strategies. In a general AFE combination strategy, I pick differential or isolated amplifiers, ADCs, comparators and isolators from their analog and power catalogs and assemble a custom solution tuned to my thresholds, bandwidth and diagnostics. In a dedicated RCM or leakage monitor strategy, I select parts that explicitly target residual current protection, with built-in test features and well-documented safety behavior, and surround them with standard isolation and MCU components.
When I prepare a sourcing or RFQ sheet, I describe what I need in terms of IC types and key parameters rather than part numbers, so that different suppliers can propose compatible devices. Examples of the kind of lines I use include:
- AFE must support leakage currents from approximately 10 mA to 2 A with at least 70 dB CMRR and less than 5 µV/°C offset drift.
- Comparator or RCM output must provide a latched trip signal with propagation delay below a few milliseconds at the specified threshold.
- Isolation devices and isolated power supplies must meet the required insulation coordination and creepage distances for the vehicle voltage class.
- Preferred devices include built-in test injection support or diagnostic flags suitable for ISO 26262 safety analysis.
Specific part numbers and brand recommendations are managed in a separate Brand IC selection mapping so that this page focuses on IC types and selection criteria. The next section turns those criteria into concrete BOM fields and board-level notes that I can reuse every time I design or source a leakage detection function.
BOM & Board-Level Notes for Leakage Detection
To make residual and leakage detection repeatable across projects, I treat its requirements as concrete BOM fields and layout constraints, not just comments in a design review. A clear checklist helps the design team, purchasing, suppliers and safety engineers read the same intent. The goal is that any engineer who picks up the schematic and PCB stack-up can see why the selected sensors, AFEs and isolation parts look the way they do.
On the BOM side, I usually define at least the following items for the leakage detection function:
- Sensor type and range — CT turns ratio, Hall current range or shunt value and power rating, plus the expected leakage current window and waveform.
- Required leakage thresholds and bands — AC and DC trip levels in milliamps or amps, along with the frequency range that must be detected and the components that should be rejected.
- Response time and trip type — whether the function must react instantly or with a short delay, and whether the behavior should be non-latching, latching or controlled by the safety MCU.
- Isolation and insulation requirements — target isolation level, creepage and clearance distances and any coordination with pack voltage class and system safety goals.
- Self-test and diagnostic hooks — test coil or injection path, ADC or comparator monitoring needs and the minimum diagnostic coverage expected for the safety case.
- Bias and power constraints — supply voltages, isolated power requirements and start-up sequencing that ensure the leakage detection function is alive whenever HV can be present.
At the PCB and layout level, small details often decide whether a theoretically good architecture behaves well in real vehicles. I record board-level notes alongside the BOM for each design, for example:
- CT secondary and shunt sense traces must be short, routed as tight pairs or with shielding where possible, and kept away from high dv/dt nodes such as gate driver outputs and bootstrap loops.
- Hall sensor grounds and reference nodes should return cleanly to the AFE and avoid sharing noisy return paths with power stages or digital logic.
- Comparator and AFE input RC networks must have well-defined values and tolerances so that bandwidth and filtering are repeatable and do not drift into nuisance trip or missed trip regions.
- Isolated amplifier or modulator inputs and outputs need appropriate spacing, guard traces and decoupling to meet the isolation rating and to avoid coupling fast transients into leakage measurement nodes.
- Test injection lines and diagnostic sense points should be physically accessible for end-of-line testing while remaining protected against accidental mis-use in the field.
Locking these requirements into BOM fields and board-level notes makes leakage detection a repeatable design block rather than a one-off exercise. It also gives the safety team and purchasing a stable reference when reviewing alternative ICs or PCB changes, and connects this page cleanly to the broader HV energy and BDU architecture work elsewhere in the system.
