The practical answer.
Choose the useful-voltage boundary from the complete device and its critical events, not from a generic coin-cell discharge endpoint. A buffer capacitor can support a transient, but its effective capacitance, ESR, leakage and recharge opportunity must be assessed. Firmware should distinguish a temporary dip, a recoverable state and an end-of-service decision without creating a repeated reset loop.
- Measure the rail at the sensitive load during the event.
- A datasheet capacity endpoint is not automatically the device's permitted cutoff.
- Ideal capacitance arithmetic is a starting bound, not a validated component choice.
Define which function must survive the lowest voltage
A device may retain memory at one rail voltage, execute code at another and need a higher voltage to start a radio or sensor. Record those requirements from the actual component and system documents. Identify the node and event for each limit. If the design team uses the battery terminal for one threshold and the processor rail for another, the difference must be explicit rather than hidden inside a single low-battery label.
The manufacturer capacity-test endpoint describes its specified test, not a promise that every device functions down to that value. Evaluate the complete path, including contacts, isolation, regulation and transient behavior. This article develops the cutoff and recovery decision rather than repeating an introductory explanation of average current and pulse loads. The useful result is a boundary the team can verify on its actual circuit.
Allocate a voltage-drop budget before choosing a capacitor
Write the starting voltage, minimum permitted load voltage and contributions from the source and current path. TI's coin-cell white paper discusses capacitor support under particular experimental load profiles. Its results support investigating buffering, not importing a universal capacity improvement or pulse-current permission into an unrelated product. Use the measured event and selected cell conditions to define the question the capacitor is intended to solve.
A deliberately simplified example assumes a capacitor must supply an additional 20 mA for 3 ms while allowing 0.20 V of droop. The ideal relationship C = I times t divided by delta V gives 300 microfarads. If an assumed 1 ohm ESR creates an immediate 0.02 V step, only 0.18 V remains for capacitive droop, raising the ideal requirement to about 333 microfarads. These are hypothetical circuit inputs, not Jetray ratings, a recommended component or a validated test result.
| Property | What to verify | Why it affects the decision |
|---|---|---|
| Effective capacitance | At actual voltage, temperature and tolerance | Label value may not represent the installed condition |
| ESR and path impedance | Relevant event and temperature | An immediate voltage step uses part of the margin |
| Leakage | Relevant operating and storage states | A steady drain can offset transient benefits |
| Recharge interval | Repeated-event sequence | The next event may begin before recovery |
| Mechanical and circuit integration | Approved placement and safe isolation | A circuit change can introduce new current paths |
Include leakage and recovery between events
A capacitor is not a source of free energy. Its charge must be restored through the actual circuit, and its leakage can add continuous demand. As conditional arithmetic, 0.5 microamp maintained for one 365-day year corresponds to 4.38 mAh. That is not an expected value for every capacitor or a measured device loss; it shows why the leakage specification and environment belong in a low-power budget.
Capture a sequence of permitted events, including the shortest expected interval, rather than only the first pulse after a long rest. Record whether the rail recovers and whether the load changes its behavior as voltage falls. If the device retries after a failed transmission, the recovery interval may be shorter than the nominal application schedule. Include that state in validation instead of assuming a fresh, fully recovered buffer before every event.
Separate warning, action and recovery thresholds
A software policy should explain when it warns, which activities it limits and when it permits normal service to resume. Define filtering, event timing and hysteresis so a brief dip does not cause rapid state changes. Do not invent one universal voltage difference for hysteresis: the correct policy depends on the circuit, measurement uncertainty, supported functions and verified behavior.
Pay attention to startup. A system can stop an event, observe voltage rebound, restart and repeat the same high demand. A simple comparison against recovered no-load voltage may then create an apparently endless cycle. Describe an approved recovery condition and user indication rather than repeatedly attempting work that the current supply state cannot support. A captured reset reason and event marker can make this failure mode much easier to identify.
Validate the device and hand over the conditions to sourcing
Use the approved cell model and mounting arrangement, representative circuit population and relevant environmental states. Preserve voltage at the sensitive rail, the event trigger, current profile and firmware response. An emulator is useful for repeatable threshold work, but final cell-related claims require the selected battery in the representative device. State which results come from simulation and which from actual hardware.
Jetray's CR2032 and CR2025 pages are reference-format inquiry routes. They do not establish that the two formats are interchangeable or that either supports a particular burst. Give the team the actual load and cutoff record and ask for the exact proposed manufacturer's evidence. The product decision should preserve cell identity, circuit and firmware revisions; changing any of these can alter the voltage boundary even if the enclosure remains unchanged.
- List the minimum voltage requirements of the functions that must remain usable.
- Measure the relevant rail during complete events and recovery intervals.
- Allocate a conditional drop budget and assess effective capacitor properties.
- Include repeated events, retries, leakage and approved environmental conditions.
- Validate warning, activity restriction and recovery logic without an automatic reset loop.
- Freeze the approved cell, circuit and firmware identity with the results.
Frequently asked questions
Can a larger capacitor always fix brownout?
No. Check effective capacitance, ESR, leakage, recharge opportunity and the rest of the current path. The cause may also be a contact, regulator or firmware issue.
Should firmware use the capacity-datasheet endpoint as cutoff?
Only if the actual device requirements and validation support it. The manufacturer's capacity-test endpoint and your useful-function limit are different definitions.
Does a voltage rebound mean the battery is ready for another attempt?
Not necessarily. Verify the loaded event and recovery policy. Rebound followed by repeated startup demand can create a reset loop rather than useful service.
Sources & further reading
- TI SWRA349: coin-cell load response and capacitor investigation ↗
- Panasonic primary lithium handbook: backup-voltage and circuit-design considerations ↗
AI-assisted editorial circuit-design education. Numerical examples are idealized planning calculations, not Jetray test results or an approved capacitor/cutoff design. No personal engineering review, guaranteed runtime or platform recommendation is claimed.
Prepared for the coin cell batteries collection. General engineering guidance does not replace a model-specific datasheet or application validation. Source review: 5 November 2026.