A precharge resistor should be approved against the complete sequence of permitted starts and failed starts, not just one successful charge from room temperature. Give the resistor supplier the event waveform, starting thermal condition, repetition pattern, mounting arrangement and longest credible energized fault interval. Give the controller supplier the matching abort and retry requirements.
The purchasing deliverable is a shared duty specification linking the resistor, precharge switching device and controller revision. A resistance value and continuous wattage are only part of that specification. A successful first start does not establish whether the same assembly can tolerate repeated requests, a partially charged restart, or a load that prevents the expected voltage rise.
This guide concerns resistor-based DC-link precharge and its procurement evidence. It is not a live high-voltage test procedure or a completed safety design. Qualified system engineers must approve topology, protection, limits and validation. The hypothetical calculations below explain document comparisons; they do not establish a suitable SINAWATTS product or a universal retry policy.
Define the duty that is missing from a one-start calculation
Start by asking what can trigger another precharge event. Possibilities include a normal operator request, a controller restart, a service command, an interrupted startup or an automatic retry after a diagnosed fault. The project owner should identify which are permitted, prohibited or unresolved.
For every permitted trigger, record the minimum elapsed time since the previous event and the conditions that enable another attempt. Distinguish a timed delay from a temperature-based release or another approved condition. A label such as automatic recovery is not enough to specify the energy history.
The Sensata precharge white paper, checked on October 19, 2026, discusses overheating from repeated precharge attempts and unsuccessful charging with downstream loads active. It describes abort timing and event counting as possible controls. Those observations justify requesting a duty history; they do not provide a ready-made retry count or cooling interval for your assembly.
Use the connector hot-plug and precharge guide for the broader connector sequence. The present decision is more specific: what resistor and control evidence covers repeated or unsuccessful charging after that architecture has been selected?
Identify the resistor branch and the owner of every input
Mark the resistor branch on the controlled circuit drawing. Identify the source node, load node, bypass path, precharge switch, relevant sensing points and any protection included in the review. The evidence should refer to those node names rather than an informal description of the battery side.
Assign ownership of the input values. The system designer owns the operating envelope; the load supplier owns applicable input-capacitance and startup information; the resistor supplier owns its product evidence; and the controller owner owns the implemented sequence. Purchasing coordinates the records but should not invent missing limits.
Record nominal values separately from tolerances and bounding conditions. A nominal capacitor value, for example, does not automatically establish the largest installed total across every variant. Another load added to the same bus can change the duty even when the resistor and contactor part numbers remain unchanged.
Where a value is provisional, keep it marked as such in the quotation. Suppliers can explain assumptions and offer alternatives without implying that an unresolved application requirement has already been approved.
On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.
| Duty input | Record required for comparison | Typical unresolved gap |
|---|---|---|
| Source envelope | Applicable voltage range and source behavior | Only a nominal battery label |
| Connected load | Capacitance, tolerances and startup state | One inverter value without other bus loads |
| Initial condition | Load voltage and component thermal state | All calculations assume a cold, discharged system |
| Repetition | Event spacing, cluster length and permitted retries | Average daily start count only |
| Abort interval | Worst elapsed energized time and its components | Software timer without physical opening delay |
| Installation | Mounting and cooling configuration | Catalog wattage without its thermal conditions |
| Acceptance | Applicable pulse, thermal and functional evidence | A successful demonstration of one start |
Use the ideal charging model as a bounded worksheet
For an ideal constant source charging an initially discharged capacitor through a fixed resistor, the time constant is R times C. The initial current is source voltage divided by resistance. These elementary relationships help screen a proposed duty, but they do not include active loads, controller behavior or all physical parasitics.
For that same ideal model, resistor power falls exponentially during charging. Integrating that power over a complete charge gives one-half C times V squared. Increasing resistance reduces initial current and stretches the event, while the ideal total dissipated energy remains the same for the same capacitance and initial voltage.
These statements are model assumptions, not sufficient component selection rules. Ask the design owner whether the supplied waveform comes from a calculation, simulation or controlled measurement, and whether the model represents the actual source, connected loads and control states.
For a partially charged capacitor, preserve the initial capacitor voltage in the analysis. Do not reuse a fully discharged waveform without checking applicability. Conversely, do not assume every hot restart has less stress simply because the bus is expected to retain charge; its actual voltage and the resistor's starting thermal state are separate inputs.
Work through a hypothetical normal-start and failed-start comparison
Assume an explicitly hypothetical system has an ideal 400 V source, an initially discharged 2 mF capacitor and a 100 Ω resistor. Ignore other resistances, active loads and thermal changes in resistance for this arithmetic example. None of these values are a product recommendation or customer test result.
The ideal time constant is 0.2 s. Initial current is 4 A and initial resistor power is 1,600 W. After a long enough complete charge, the resistor has dissipated approximately 160 J. That energy alone does not describe the resistor's peak-power limit, permissible voltage, mounting requirement or repeated-pulse capability.
Now compare two deliberately different event models. In the first, the capacitor charges normally. In the second, an idealized fault holds the load-side voltage near zero for 0.5 s, leaving approximately the full source voltage across the resistor. The second is a simplified fault-energy screen, not a prediction that every downstream fault behaves this way.
On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.
| Hypothetical event | Simplified electrical result | Procurement implication |
|---|---|---|
| Normal complete charge | Approximately 160 J with decaying power | Request evidence for the actual pulse shape and initial temperature |
| Load-side voltage held near zero for 0.5 s | Approximately 1,600 W for 0.5 s, or 800 J | Review the fault interval and interruption evidence separately |
| Five fully discharged normal starts | Approximately 800 J total across the cluster | Thermal outcome still depends on spacing and cooling |
| One normal event every 10 s indefinitely | Approximately 16 W average in the ideal model | Average power alone does not establish pulse or peak-temperature acceptance |
The failed-start model deposits five times the energy of one complete normal charge. The five-start cluster has the same arithmetic total, yet its thermal behavior need not match a single 800 J pulse. Timing matters, and neither comparison can be turned into a temperature estimate without applicable thermal evidence.
Suppose an offer includes only a single-pulse curve starting from the supplier's reference temperature. The buyer cannot treat the five-start cluster as covered merely because each event is individually below that curve. Ask for confirmation of the complete repetition and cooling condition.
The next procurement action is concrete: attach both event descriptions, identify the model limitations, and ask which are supported by the offered part and installation. The supplier's response should separate demonstrated capability, calculation-based interpretation and unresolved validation.
Read pulse evidence with its temperature and waveform conditions intact
A pulse rating is not a free-standing number of joules. The documentation needs the event duration or waveform and the conditions under which the rating applies. Preserve whether the claim covers one pulse, repeated pulses, a particular temperature or a specified mounting arrangement.
Vishay's Pulse Capabilities for Thick Film Power Resistors illustrates separate checks for limiting voltage, pulse energy, repetition-related average power and thermal conditions. Its worked example concerns a named resistor family and installation assumptions. It supports the review method; it does not transfer a pulse capability to another resistor or authorize replacing a real waveform with an arbitrary rectangle.
Ask for the actual document revision and the offered resistance value. A family brochure can be useful context, but the supplier should identify which curve, table or application statement supports the specific offer. Where a curve is difficult to interpret, request the manufacturer's written interpretation.
Keep exclusions visible. A statement covering normal charging may explicitly exclude fault hold, reduced cooling or rapid restarting. Such an offer can still be reviewed, but its price must be compared against the same required scope as alternatives.
Define restart clusters instead of relying on starts per day
A daily average can conceal a severe short cluster. Twenty starts spread through a day are a different requirement from twenty attempts in a short interval. The RFQ should specify the most demanding permitted sequence, not only the expected normal usage.
Create a sequence record with an event number, initial state, energization profile, elapsed time before the next request and release condition. If the pattern includes pauses of different lengths, preserve them. A statement that the resistor cools between events needs a defined basis.
Ask the design owner to identify how normal operator actions interact with automatic retries. A controller that permits several retries per command may experience more attempts if a new external command resets the counter. The intended policy and the implemented counter behavior must agree.
Also define recovery after power interruption. The record should state whether relevant history survives a reset, how the next state is determined, and what prevents an unintended fresh allowance of attempts. These are questions for controller review, not instructions to bypass a protective function.
Send your precharge waveform, restart pattern and cooling requirements with the circuit revision. Request a quotation that names the evidence needed for both successful and aborted charging.
Treat abort latency as a complete interval
The resistor remains exposed until the actual current path has changed, not merely until software decides that the event should stop. Request an end-to-end interval covering sensing, filtering, decision logic, communication where used, output response and the switching device's applicable behavior.
Use a common time reference for the evidence. The start of the controller timer may differ from the physical beginning of branch current. A report should explain how those times are aligned and what uncertainty or tolerances remain in the comparison.
For a hypothetical timing review, assume synchronized records define four consecutive, non-overlapping intervals: 0.35 s from branch-current onset to a fault criterion, 0.04 s for subsequent filtering, 0.03 s for decision and output response, and 0.08 s until current extinction. The total is 0.50 s, matching the invented fault duration used above. At the assumed constant 1,600 W, budgeting only the first 0.35 s would represent 560 J instead of 800 J. These figures are illustrative allocations, not measured device delays or recommended settings. Real records may contain overlapping intervals, which must not be added twice; use common event markers and the actual waveform.
The abort condition must also be meaningful. An elapsed-time limit and an expected voltage-rise envelope answer different questions. Ask which faults each criterion is intended to detect, what operating variations it must tolerate and how an invalid sensor reading is handled.
For the precharge switching device, distinguish normal opening duty from opening during an unsuccessful charge. The current, voltage and circuit state may differ. A carry rating does not settle that interruption question; the responsible engineer needs applicable switching evidence.
Require confirmation of branch state after transition
The system should not infer every physical state from a command alone. Ask what evidence confirms that the main path has reached its intended state and that the precharge branch has transitioned as required by the approved architecture.
This is particularly relevant when the main contactor does not behave as commanded, the precharge switch remains conductive, or feedback is unavailable. The purchasing specification should identify the diagnostic coverage and the unresolved cases rather than demanding a generic fault-safe label.
Use the contactor weld-detection and auxiliary-contact guide for evaluating feedback evidence. A feedback contact, bus-voltage observation and branch-current observation are different signals; none should be assumed to prove more than the approved diagnostic design establishes.
Ask for a state-transition table linked to the firmware or controller revision. Include the condition for allowing load operation and the conditions that inhibit another start. This makes a later controller substitution reviewable without reopening the entire mechanical quotation.
Specify the thermal installation actually being purchased
A resistor supplied loose, attached to a bracket, or integrated into a cooled assembly can have different evidence requirements. Identify which supplier owns the mounting surface, thermal interface, fastening arrangement, local airflow or coolant condition where applicable.
Do not infer a thermal rating from an aluminum case or a large external size. Ask which documented installation conditions are required for the offered performance, and whether the purchasing scope includes the relevant mounting parts and assembly controls.
Specify the relevant starting condition in terms the supplier's evidence can use. Ambient air, housing temperature and internal element temperature are not automatically interchangeable. The supplier and system designer should agree how measured or calculated temperatures support the intended acceptance.
Record adjacent heat sources and enclosure conditions when they form part of the application envelope. The required evidence may need an assembly-level review even when each component individually has a familiar catalog rating.
Compare supplier offers against a common duty packet
Send each bidder the same circuit, waveform set, restart sequence and installation assumptions. Require a response against each requirement using supported, excluded or awaiting clarification, with a reference to the applicable evidence.
One bidder may propose a larger resistor; another may require a longer recovery interval; a third may exclude the fault event. These are different technical offers. Purchasing should make the tradeoffs visible rather than compressing them into one wattage and one unit price.
On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.
| Offer element | Useful supplier response | Follow-up when absent |
|---|---|---|
| Single event | Supported waveform, conditions and document reference | Request the exact pulse interpretation |
| Repeated events | Approved event pattern and thermal basis | Request a cluster assessment |
| Failed charge | Maximum supported exposure and assumptions | Align with end-to-end abort evidence |
| Cooling | Required installation and temperature conditions | Resolve assembly ownership and missing parts |
| Switch coordination | Relevant making and abort-opening conditions | Obtain component and system review |
| Reporting | Traceable part, revision and test or analysis record | Define deliverables before sample approval |
Document the reason for the selected option. If a control restriction is part of acceptance, it belongs in the controlled system requirement and verification plan. It should not remain an email footnote that disappears when software is updated.
Plan validation around distinguishable questions
Write the question beside each proposed validation event. A cold normal start checks one set of conditions. A hot repeat cluster checks another. A failed voltage rise, a sensor fault or an interrupted transition addresses a different part of the design.
The competent validation owner should define safe facilities, instrumentation and procedures. Purchasing's role is to ensure that the delivered evidence covers the purchased requirement and that any excluded events are explicitly dispositioned. This article does not instruct a buyer to create a high-energy fault.
Require synchronized records sufficient to relate branch behavior, relevant voltages, controller states and thermal observations. Ask for the necessary sampling and measurement capability rather than prescribing an arbitrary instrument specification.
Keep the physical specimen, resistor lot, switch variant, mounting configuration and firmware revision linked to the report. A later favorable test on a different assembly does not silently close an earlier exception.
Make pass criteria and investigation rules visible before testing
Agree what constitutes acceptable electrical completion, temperature behavior, resistance stability, physical condition and required diagnostic response. These criteria must come from the approved application and component evidence; there is no universal limit that fits every precharge assembly.
Define how borderline or incomplete results are handled. Missing temperature data, an invalid sensor channel or a fixture interruption may require a limited repeat. Preserve the original evidence and the reason for that repeat rather than replacing an inconvenient record.
A successful sequence is not proof of unlimited endurance. State the number and scope of validated operations and the basis for any additional life claim. Avoid marketing language that promises permanent prevention of welded contacts or resistor overheating.
When a result differs from the model, examine the assumptions before changing the limit. The investigation may identify additional connected capacitance, an active load, a timing mismatch or different thermal conditions. Record the actual cause if established; otherwise retain the uncertainty.
Control substitutions and field resets after approval
Treat a different resistance value as a system change, even if its wattage is higher. It can alter the event waveform and completion timing. A change in resistor construction, mounting kit or thermal interface also requires review against the evidence on which approval was based.
Controller changes deserve equal attention. A shorter startup deadline, new automatic retry, altered sensor filtering or reset behavior can change the imposed duty without any change to the power components. Link the approved duty packet to the controller release process.
For service replacement, provide the exact approved configuration and any required installation instructions from the responsible manufacturer. Do not instruct technicians to defeat a retry lockout to restore availability. A repeated fault should enter the approved diagnostic and service process.
Maintain a concise revision register listing what changed, affected requirements, evidence reviewed and approval disposition. This lets purchasing handle routine repeat orders quickly while identifying changes that genuinely need engineering attention.
Keep commercial terms tied to the evidence scope
Ask quotations to separate components, mounting parts, sample quantities, agreed documentation and any application-specific assessment. A lower price may simply omit the thermal installation or validation report assumed in another offer.
The OEM and ODM inquiry page is a place to describe the assembly need. It does not establish that a specific resistor, contactor, precharge controller or test service is available. Obtain written confirmation for the proposed configuration and evidence.
Specify ownership of unresolved requirements and the date needed for each decision. This can shorten quotation exchanges without pretending that an unknown operating condition is already settled.
No stock, price, MOQ, delivery time, certification or factory-testing capability is established by this guide. Those terms require an actual supplier response for the requested scope.
FAQ
Is a resistor with a higher continuous wattage automatically safer for precharge?
Not automatically. Compare its permissible voltage, pulse waveform, repetition and thermal installation against the actual duty. A continuous rating under one cooling condition does not by itself establish acceptance for a different short pulse or an aborted charging event.
Can I approve repeated starts using a single-pulse energy curve?
Only if the applicable manufacturer evidence and engineering assessment cover the complete repetition and starting conditions. Otherwise the curve answers a narrower question. Ask for the required recovery condition or a supported repeated-event assessment rather than inventing a delay.
Does a partly charged DC link always make the next attempt easier?
It can change the electrical event, but the resistor may also start hotter and other load states may differ. Record both electrical and thermal initial conditions. Do not treat assumed residual voltage as a substitute for a defined restart requirement.
Is a software timeout enough to define failed-precharge energy?
No. The exposure includes the applicable sensing and decision delays and the physical response that ends branch current. Use an agreed end-to-end interval and waveform. The software setting is one input to that assessment, not automatically the entire interval.
Can I use the normal precharge current as the switch's abort rating?
Not without applicable evidence. The switch may open under a different voltage and current condition during an aborted event. Ask the system and component owners to confirm the relevant opening duty and protection arrangement for the exact offered device.
What if a supplier offers a larger resistor instead of a retry limit?
Compare the supported operating envelope, installation and cost. Either proposal still needs an approved system-level duty specification. A larger part does not justify unlimited retries unless the actual evidence supports the entire permitted sequence.
What is the minimum information needed for an RFQ?
Provide the circuit revision, source envelope, connected capacitance and load state, event waveforms, restart clusters, abort timing, thermal installation, candidate switch details and required evidence. State quantities separately and identify every assumption awaiting engineering confirmation.
Send a repeated-duty precharge RFQ
Assemble the normal-start waveform, failed-start exposure and permitted restart sequence into one controlled packet. Include the installation drawing and controller revision, then request a response that identifies supported conditions, exclusions and required validation.
Send your precharge resistor duty and abort-evidence requirements. A useful quotation should explain how the offered configuration matches the complete operating history, allowing purchasing to compare real scope and engineering to approve the remaining application decisions.