HV Disconnect & Pre-Charge Unit for ESS and EV Packs
← Back to: Energy & Energy Storage Systems
This page shows how to design an HV disconnect and pre-charge unit that safely tames inrush, manages contactor weld risk and coordinates surge/UV protection. It turns bus stresses and safety goals into concrete topologies, checklists and IC choices that can be discussed directly with suppliers.
What this page solves
High-voltage battery packs in ESS containers, EVs and UPS systems connect directly to large DC-link capacitors and power converters. If this HV bus is hard-switched with no dedicated disconnect and pre-charge stage, inrush currents can easily reach hundreds of amperes. For example, an 800 V system with a 2 mF DC-link can see instantaneous surge currents far beyond the mechanical rating of typical contactors and fuses.
At the same time, contactor weld or sticking faults are critical safety hazards. A contactor that is welded closed while the system believes it is open exposes maintenance staff to unexpected live conductors and prevents controlled isolation during faults. Robust weld detection and clear interlock feedback are therefore mandatory functions, not optional diagnostics.
This page focuses on the HV Disconnect & Pre-Charge Unit as a dedicated hardware subsystem. It maps inrush and weld risks into concrete design requirements for contactor and pre-charge drivers, surge/UV comparators, current and voltage sensing, insulation-monitor interfaces and isolated power. The goal is to provide a hardware/IC-level design map that can be turned directly into specifications, schematics and safety-relevant diagnostics.
Bus stresses and design limits
HV disconnect and pre-charge design is driven by a small set of bus-level parameters: nominal DC voltage, the total DC-link capacitance seen by the battery, expected maximum inrush current, and the duty cycle of connect / disconnect events. Typical ESS and EV platforms work around 400 V, 800 V or 1000 V+ DC with effective DC-link capacitances ranging from hundreds of microfarads per rack up to several millifarads when cabinets are paralleled.
For a given bus voltage and DC-link capacitance, the theoretical inrush current during a hard connection can exceed the mechanical make-and-break ratings of contactors, PCB traces and fuses. As bus voltage and capacitance scale up, single-stage pre-charge topologies become insufficient and multi-stage, timed or current-limited schemes are required. Standards for high-voltage isolation, dielectric strength and switching (such as EV and ESS safety regulations) indirectly enforce limits on allowable surge, creepage, clearance and insulation coordination.
These bus stresses translate directly into IC-level requirements. Voltage dividers and comparators must withstand the maximum DC and surge levels, isolated drivers and supplies must meet reinforced or basic insulation ratings, and UVLO and supervision thresholds must guarantee controlled behaviour during brown-outs and fast transients. The matrix below helps position a design and indicates when a dedicated, more complex pre-charge and protection strategy is required.
Functional chain overview
The HV disconnect and pre-charge unit sits between the battery pack and the DC-link of the PCS or inverter. Along this path, energy flows through main and negative contactors and a controlled pre-charge branch, while voltage and current sensing, weld detection, surge and UV supervision, and isolated power and communication interfaces observe and control the process. The functional chain connects high-voltage switching hardware with low-voltage logic and diagnostics.
Main and negative contactors provide the primary make and break capability for the HV bus. A pre-charge branch, built from a relay or MOSFETs and a resistor, limits inrush when first charging the DC-link capacitance. Dedicated sensing points on the pack side and DC-link side measure voltage and current, enabling inrush control, weld detection and verification that the bus is actually de-energized when commanded off.
Around this core energy path, surge and under-voltage comparators supervise the bus and enforce safe operating windows for closing or opening the contactors. Isolated flyback or push-pull power stages and bias LDOs supply the drivers, comparators, current-sense AFEs and local controllers. Digital isolators and interface transceivers carry interlock, fault and status information back to the BMS, PCS and site EMS so that higher-level control decisions can be coordinated with the hardware capabilities and limits of the disconnect unit.
From an IC perspective, this chain maps to high-side and low-side coil drivers for contactors, MOSFET gate drivers for any solid-state pre-charge branches, current-sense amplifiers or sigma-delta modulators, window comparators and precision references, isolated DC-DC converters and bias regulators, and digital isolation and communication devices. The block diagram below summarizes this signal and power flow from left to right.
Contactor and pre-charge driver topologies
The way contactor coils and pre-charge branches are driven determines inrush behaviour, thermal stress and diagnostic coverage. At the simplest end, a single high-side MOSFET can drive a coil from a 12 V or 24 V rail. At the more advanced end, dedicated coil drivers adjust current between pull-in and hold levels, supervise line faults and detect release during brown-out. Pre-charge branches follow a similar spectrum, from mechanical relays and a fixed resistor to MOSFET-based or hybrid stages with controlled current profiles.
Coil driver topology selection is driven by the number of contactors, coil power, ambient temperature and the required level of diagnostics. Simple MOSFET drivers are attractive for a single contactor with modest coil current, but multi-contact units with high pull-in current and tight release requirements benefit from dedicated automotive coil driver ICs that manage energy, freewheel paths and fault reporting. Freewheel diodes and TVS clamps must be placed carefully, especially when several coils share the same supply or ground.
Pre-charge driver topologies are tied to bus voltage and DC-link capacitance. Mechanical relay pre-charge stages with a single resistor work well in moderate stress regions. At higher voltage or capacitance, MOSFET-based or hybrid pre-charge makes it easier to shape the inrush current, split the process into multiple stages and coordinate timing with current and voltage sensing. These choices map into gate driver voltage, peak and average coil driver current, the need for current sense amplifiers, and the protection and supervision features expected from each IC.
Weld detection and state feedback
A welded contactor breaks the fundamental assumption that the commanded state matches the physical state of the HV bus. If a main contactor welds closed while the BMS, PCS or EMS believes the bus is open, maintenance staff and other subsystems may interact with live conductors that are assumed to be isolated. Robust weld detection therefore becomes one of the most critical diagnostics in an HV disconnect and pre-charge unit.
Practical weld detection combines several observations. The most common method compares pack-side and DC-link voltages after an open command and a defined delay: if either side remains close to nominal when it should decay towards zero, a welded or bypassed path is suspected. Insulation monitoring devices can provide additional insight by revealing unexpected conductive paths after contactor open. In systems with multiple series or redundant contactors, differences between the voltages seen at each node help localize which pole may be welded.
These observations are implemented using high-voltage divider networks feeding comparators or multi-channel ADCs referenced to a stable bandgap or shunt reference. After a disconnect command, a timing window defines when voltages and leakage currents are sampled and compared against limits. The resulting digital weld-detection signals feed a state machine in the BMS or a local controller, which distinguishes between transient anomalies and confirmed weld faults and maps them into explicit diagnostic bits such as weld suspected and weld confirmed.
Once welded states are represented as diagnosable flags, the HV disconnect unit can coordinate behaviour with the rest of the system. Weld suspicion can inhibit further close commands, block service or maintenance modes, and trigger controlled shutdown procedures. Confirmed weld faults can be latched until hardware inspection, ensuring that a contactor which no longer provides galvanic isolation is never treated as a safe disconnection point.
Surge and under-voltage comparators and protection windows
Surge, over-voltage and under-voltage events are best treated as gatekeepers for the HV disconnect and pre-charge unit. Only when the bus voltage lies inside an acceptable window should pre-charge or contactor closing be allowed. When the bus exceeds the upper bound or drops below the lower bound, comparators and window detectors must rapidly block new close commands and, when necessary, trigger opening or prevent reclosure after a fault.
The core implementation uses a resistive divider from the HV bus into one or more comparators. A lower threshold detects when the bus has not yet reached a safe level for pre-charge to finish or contactors to close, while an upper threshold detects over-voltage and surge conditions. Thresholds are typically defined as percentages of nominal DC voltage, for example a lower bound around 0.7 × Unom and an upper bound near 1.1 × Unom, with hysteresis to prevent chattering near the limits. A dedicated window comparator can integrate both thresholds and output a single window OK signal to the disconnect logic.
These comparators work alongside surge protection devices, TVS diodes and eFuses rather than replacing them. SPD and TVS elements absorb fast transients, while eFuses react to over-current. Comparators and supervisors interpret the resulting bus voltage profile and decide whether it is acceptable to close or keep holding mechanical contactors. When SPD or eFuse devices act, the comparator outputs provide a clear indication to the BMS or PCS that the HV bus is outside its normal operating window and that reconnect decisions must be delayed or blocked.
From an IC perspective, low-power comparators, window detectors and voltage supervisors paired with a stable bandgap or shunt reference implement these protection windows with minimal static loss. Their outputs drive simple logic that generates allow_precharge and allow_contactors signals, giving the HV disconnect unit a hardware-level admission control that complements the higher-level software strategy.
Isolated power and signal interfaces
Inside an HV disconnect and pre-charge unit, several different rails are needed to supply the coil drivers, sensing front ends, comparators, digital isolators and any local microcontroller or logic. Coil drivers typically require a 12 V or 24 V rail capable of handling high pull-in current and continuous hold current. Precision AFEs and comparators benefit from a clean, lower-voltage analog rail, while digital logic, timers and state machines run on standard microcontroller supplies. Some designs also reserve a small always-on rail to keep safety diagnostics alive when the main system is down.
These rails can be sourced from the main DC-DC converter of the PCS or cabinet, from a dedicated auxiliary AC/DC or HV DC/DC supply, or from a separate low-voltage battery. Using the main system supply simplifies wiring but may leave the disconnect unit powerless when that supply fails. Dedicated auxiliary converters or small backup batteries provide a more robust solution, allowing the disconnect box to open contactors, perform weld checks and report fault status even when the primary converter trips or the grid connection is lost.
Isolation requirements are driven by the maximum DC bus voltage, pollution degree, material CTI and applicable safety standards. Isolated DC-DC converters, sigma-delta current-measurement AFEs and digital isolators must provide appropriate basic or reinforced insulation, with creepage and clearance tailored to the working voltage. Common-mode transient immunity is critical when the disconnect shares a cabinet with fast-switching silicon carbide or gallium nitride stages, since poor CMTI can corrupt current measurements, comparator outputs and weld-detection diagnostics at precisely the moment they are needed most.
On the interface side, the HV disconnect unit exchanges information with the pack BMS, PCS and site EMS using both hardwired and bus-based links. Safety interlock loops and dry contacts convey simple permissive or fault states, often arranged so that a loss of power or broken wire defaults to a safe, inhibited condition. Isolated CAN, RS-485 or optical links carry richer status, such as pre-charge stages, contactor counts and weld-diagnostic flags. Digital isolators, isoCAN transceivers and optocouplers form the physical layer that transfers these signals across the high-voltage isolation barrier shown below.
Application mini-stories
Rack-level ESS cabinet example
Consider an indoor rack-level energy storage system built around an 800 V DC bus and roughly 250 kWh per rack. Each rack connects to a common DC-link on the PCS side and includes its own HV disconnect and pre-charge unit, plus a rack BMS. The DC-link capacitance is concentrated in the PCS and shared across several racks, so each rack must bring the shared bus gently up towards its pack voltage before closing main contactors.
A typical configuration uses a positive and negative contactor per rack, plus a mechanical pre-charge relay and resistor sized from the bus voltage and effective DC-link capacitance. A dedicated automotive-grade coil driver IC feeds the contactors and pre-charge relay, providing pull-in and hold-current regulation and basic diagnostics. Pack-side and DC-link voltages are measured through high-voltage dividers and a dual-channel window comparator or multi-channel ADC, allowing the controller to confirm pre-charge completion and detect welded contacts after open commands. A sigma-delta current sensor in the pre-charge path supervises current limits and thermal stress on the resistor.
An isolated auxiliary supply powers the coil driver, current sensor AFE and local logic, while isolated CAN or RS-485 connects the rack disconnect box to the rack BMS and the cabinet controller. This arrangement provides a clear set of diagnosable flags for pre-charge success, weld suspicion and contactor wear, without exposing the BMS directly to HV bus transients. Engineers can map the comparators, sigma-delta modulator and coil driver into a concise IC set and derive a repeatable schematic for each rack in a multi-rack ESS.
EV / commercial vehicle pack example
In a commercial vehicle traction battery, the HV pack often operates at 800 V or higher and experiences frequent charge and discharge cycles, high vibration levels and wide ambient temperature swings. The pack-integrated HV disconnect and pre-charge unit must handle higher inrush currents, faster load changes and strict functional safety targets. Main and negative contactors are sometimes complemented by additional contactors or service disconnects to support maintenance modes and crash isolation strategies.
Weld detection in this context combines voltage comparison, insulation monitoring feedback and redundant measurement paths. Dual HV measurement channels, each with its own divider and AFE or comparator chain, cross-check pack and inverter-side voltages after open commands. Insulation monitoring devices contribute an independent view of the conductive paths. High-CMTI isolated sigma-delta AFEs and digital isolators prevent switching transients from corrupting these measurements. The resulting signals feed a safety-class MCU that implements weld state machines aligned with the vehicle's ASIL targets.
The IC set for such a pack typically includes redundant coil driver channels, dual high-voltage measurement AFEs or comparators, high-isolation DC-DC converters, safety-oriented microcontroller families and robust digital isolators. Together they turn the HV disconnect unit into a diagnosable safety element: the pack can prove that disconnect commands have been executed, detect welded contactors and block unsafe reclosure events across the vehicle lifetime, while still fitting into a compact housing on the traction battery.
Design checklist and IC mapping
Use this checklist as a conversation starter with suppliers and FAE teams. Each item links back to the earlier sections on HV bus stresses, contactor and pre-charge topology, weld detection, surge and under-voltage windows, and isolated power and interfaces. The goal is to ensure that the disconnect and pre-charge unit is fully specified at both the system and IC level before layout and prototype builds.
- Bus voltage and DC-link capacitance defined: maximum DC bus level, worst-case rack or cabinet count and effective DC-link capacitance calculated, including PCS input capacitors and any cabinet-level film banks.
- Pre-charge strategy sized from worst-case inrush: pre-charge resistor value, number of pre-charge stages and timing chosen from the highest voltage and capacitance combination, with resistor power and energy checked for repeated attempts and fault conditions.
- Contactor count and coil ratings frozen: positive, negative and any service or intermediate contactors defined, including coil voltage, pull-in and hold currents, mechanical life and expected switching cycles.
- Coil driver rails and clamps verified: coil driver IC and supply rail sized for simultaneous pull-in at the lowest supply voltage and highest ambient, with flyback diodes, TVS clamps and driver absolute maximum ratings checked for all coil combinations.
- Weld detection paths defined and, where needed, redundant: at least one voltage-based weld-detection chain implemented (pack and DC-link sensing after open commands), with a second independent path or insulation-monitoring feedback added where functional safety or grid-code requirements apply.
- Diagnostic states mapped into clear flags: weld suspected, weld confirmed, pre-charge in progress, pre-charge failed and ready-to-close states defined, latched and exposed to BMS, PCS and EMS through interlock loops and isolated communication channels.
- Surge and under-voltage windows coordinated with BMS and PCS: UV and OV thresholds expressed as percentages of nominal bus voltage and aligned with the internal protection thresholds of the PCS, pack BMS and upstream protection devices to avoid conflicting trips or oscillatory behaviour.
- Window comparator dynamics tuned: comparator hysteresis, propagation delay and any digital filtering chosen to reject short transients while still blocking unsafe closing after surge, SPD action or eFuse intervention.
- Isolated power budget and thermal headroom checked: isolated DC-DC converters sized for worst-case coil pull-in, analog front-end, logic, isolator and auxiliary loads with margin, and verified against enclosure temperature, airflow and PCB copper area to keep junction temperatures within lifetime limits.
- Isolation ratings and creepage/clearance compliant: working voltage, surge ratings, insulation type (basic or reinforced) and creepage and clearance distances for isolated DC-DC converters, sigma-delta AFEs and digital isolators checked against the target system standards and material CTI.
- Interfaces and interlock loops fail safe: dry-contact or interlock circuits verified to default to a blocked or not-allowed-to-close state under loss of power, cable faults or isolator failures, and CAN or RS-485 physical layers validated for CMTI, ESD and common-mode voltage.
- IC families shortlisted and second sources identified: for each functional block below, at least one primary IC family and one alternative from a different vendor identified, matching voltage, isolation, temperature and functional safety expectations for the target ESS or vehicle platform.
| IC function block | Key parameters and design notes | Example vendor / series (non-exhaustive) |
|---|---|---|
| Coil driver | Coil supply voltage (12 V / 24 V), peak pull-in current, PWM hold capability, integrated current regulation, diagnostics (open/short coil, supply faults), operating temperature range and automotive or industrial qualification level. | Dedicated automotive coil or high-side drivers such as DRV110 / DRV120-class coil current controllers (Texas Instruments) or TLE9210-class multi-channel relay and valve drivers (Infineon) used as representative families for contactor and pre-charge coil control. |
| Current sense / sigma-delta modulator | Measurement range and shunt value, bandwidth, isolation rating (working and surge), common-mode voltage range, CMTI for fast bus dv/dt, output format (bitstream or SPI), offset and gain drift over temperature. | Isolated sigma-delta current-sense modulators such as AMC1306 / AMC1305 (Texas Instruments) or AD7403-class devices (Analog Devices), or isolated shunt amplifiers like ISO224-class family for pre-charge and weld-detection current monitoring. |
| Comparator / supervisor | Input common-mode range relative to sense dividers, integrated reference voltage, window comparator options, hysteresis behaviour, propagation delay, quiescent current and operating temperature. Mapping to UV/OV window thresholds and weld-detection voltage windows. | Low-power comparators and supervisors such as TLV6700-class window comparators or TLV6702-class dual devices (Texas Instruments), and supply supervisors like LTC2965-class voltage monitors (Analog Devices) used for HV bus window checks and auxiliary rail supervision. |
| Isolated DC-DC / gate driver supply | Output power budget for coils, AFEs, logic and isolators, isolation voltage, reinforced or basic insulation rating, efficiency, no-load consumption, CMTI and available safety certifications. In some designs, isolated driver supplies also support nearby gate drivers or IMD circuits. | Integrated isolated DC-DC converters such as UCC1205 / UCC12040-class devices (Texas Instruments) or ADuM5020-class isoPower converters (Analog Devices), as well as small industrial modules like RxxP2-class isolated converters (RECOM) for auxiliary rails in the disconnect box. |
| Digital isolator / optocoupler | Channel count, data rate, isolation rating and lifetime, CMTI for high dv/dt events, logic-side supply voltages, integrated CAN or RS-485 transceivers where required, and whether the device is qualified for automotive or industrial temperature and shock ranges. | Multi-channel digital isolators such as ISO77xx-class families (Texas Instruments) or ADuM140x-class devices (Analog Devices) for GPIO and SPI, plus isolated CAN / RS-485 transceivers such as ISO1042-class or ADM305x-class devices, and optocouplers like HCPL-3120-class parts (Broadcom) where a simple transistor output is sufficient. |
FAQs on HV disconnect and pre-charge design
This section collects common questions that appear when specifying an HV disconnect and pre-charge unit for battery energy storage or vehicle packs. Each answer is written from a hardware and IC point of view and links back to the detailed sections above for deeper discussion, examples and design checklists.
When is a single contactor with pre-charge acceptable, and when are dual or even triple contactors mandatory for safety and service isolation?
A single contactor with pre-charge is acceptable when bus energy, fault levels and safety requirements are modest and upstream or downstream equipment already provides additional isolation. As bus voltage, DC-link capacitance and safety expectations increase, designs move toward dual-pole or multi-contactor schemes. See sections H2-3, H2-4 and H2-5 for guidance.
At what DC-link capacitance or bus voltage does a single-stage pre-charge become impractical, and when should multi-stage pre-charge be used instead?
Single-stage pre-charge becomes impractical when the required resistor power, energy and allowable temperature rise exceed practical component or layout limits. Large DC-link capacitances or very high bus voltages favour multi-stage pre-charge to control inrush shape and heating. See sections H2-2 and H2-4 for how to frame these thresholds in real projects.
When is simple voltage-based weld detection sufficient, and when is it necessary to add current sensing or timing-based diagnostics?
Simple voltage-based weld detection is often sufficient when loads are small, discharge paths are well defined and voltage decay after an open command is predictable. As bus energy, parallel paths or safety requirements grow, adding current sensing and timing-based diagnostics becomes necessary. See section H2-5 for typical weld-detection architectures and trade-offs.
How far should surge and under-voltage comparator thresholds be placed from BMS and PCS protection limits to avoid chattering and conflicting trips?
Comparator thresholds for surge and under-voltage protection should sit far enough away from BMS and PCS limits to avoid chattering, yet close enough to protect contactors and pre-charge components. A small margin and suitable hysteresis are usually preferable to overlapping trip points. See sections H2-2, H2-6 and H2-9 for coordination strategies.
What is a safe strategy when pre-charge fails: should the system simply open and lock out, or is it acceptable to allow a limited number of retries?
Safe pre-charge behaviour balances hardware stress with availability. Many systems allow a small, fixed number of retries with enforced cool-down time between attempts, then latch into a fault that requires operator or remote intervention. Retry logic should never mask sustained faults. See sections H2-6, H2-8 and H2-9 for implementation patterns.
In which applications should weld detection and contactor diagnostics be treated as a safety function with ASIL-like requirements rather than a simple convenience feature?
Weld detection and contactor diagnostics become safety functions when a welded contact could expose personnel, vehicles or critical equipment to hazardous energy without other barriers. Traction batteries, safety-related UPS feeds and certain industrial drives often need ASIL-like thinking. See sections H2-5 and H2-8 for how redundancy supports these requirements.
When must the isolated supply for the HV disconnect be independent of the main DC-DC, and when can it safely be derived from the same converter?
An isolated supply derived from the main DC-DC is acceptable when its loss cannot compromise safe isolation and when shared fault modes are tolerable. If the disconnect must operate independently during converter faults, evacuation or firefighting scenarios, a separate auxiliary supply is preferred. See sections H2-7 and H2-9 for typical choices.
When does the common-mode transient immunity (CMTI) of sigma-delta current sensors and digital isolators become a limiting factor for the HV disconnect design?
CMTI becomes a limiting factor when fast-switching SiC or GaN stages share a cabinet or bus segment with the HV disconnect measurement chain. High dv/dt edges can corrupt sigma-delta modulators and digital isolators if ratings are too low, causing false weld flags. See sections H2-2, H2-5 and H2-7 for context.
How should responsibilities be split between cabinet-level and pack-level HV disconnect units when both are present in a system?
When both cabinet-level and pack-level disconnects exist, the pack-level unit typically protects the battery modules, manages weld diagnostics and supports vehicle or rack service. The cabinet-level unit coordinates site isolation, grid interaction and emergency shutdown. Clear responsibility split avoids duplicated logic. See sections H2-1, H2-3 and H2-8 for examples.
Is it better to share a current-sense path between pre-charge and normal operation, or to reserve a dedicated sensor for the pre-charge branch?
Sharing a current-sense path between pre-charge and normal operation reduces component count but can compromise resolution, range and timing for small pre-charge currents. A dedicated sensor on the pre-charge branch simplifies diagnostics and improves weld and fault detection. See sections H2-3, H2-4 and H2-5 for typical design patterns.
Which HV disconnect states should be carried over simple interlock or dry-contact loops, and which ones can remain on higher-level CAN or Ethernet links?
Interlock or dry-contact loops should only carry the most critical permissive and fault states, such as ready-to-close, emergency trip and weld confirmed. Richer diagnostic information can remain on CAN, RS-485 or Ethernet. This keeps the safety chain simple and robust. See sections H2-5, H2-7 and H2-9 for examples.
How can a single HV disconnect and pre-charge hardware platform be reused across projects with different bus voltages, DC-link capacitances and safety requirements?
A reusable HV disconnect platform starts from scalable building blocks: contactor ratings, divider ranges, pre-charge resistor footprints and isolated supplies that cover several bus classes. Parameterisation via component values and firmware allows adaptation to different voltages and DC-link sizes. See sections H2-2, H2-3, H2-8 and H2-9 for guidance.