The public data strongly confirms or exceeds the datasheet claim with controlled heat and complete context.
Methodology and fairness rules
Battery tests are only useful when the comparison is clear. This page explains how raw CSV logs are converted into public metrics and why early stops are flagged.
Datasheet accountability rating
BRD Score system
BRD Score is a 1 to 10 quick-read score built around one question: does the cell live up to the manufacturer datasheet under the datasheet limits? Raw Ah / mAh receives the strongest weight because it is the clearest capacity accountability number. The score compares completed capacity evidence against datasheet minimum and nominal capacity when those values are available. Low-current datasheet-style runs are preferred context, but completed real-load data is accepted when it proves capacity. Usable capacity above 3.0V is still shown beside it because it explains real-world runtime. If a datasheet has unclear temperature/current language, hides cold-weather operating limits, contradicts itself, or contains known errors, BRD applies a visible datasheet-quality penalty and explains the reason in the score notes. It rewards cells that meet or exceed the printed claims, and it penalizes cells that fall short, overheat at or below the true CDR, sag badly, stop early, or publish weak or missing cycle-life information. Oversize physical measurements are shown as fit warnings for users, but they do not reduce the BRD Score. It helps visitors scan the catalog, then read the actual curves and notes before making decisions.
Before any score is calculated, completed runs are grouped by base cell model, physical sample, test mode, current, cutoff voltage, and the mode-specific condition. Burst uses its exact duration. Dynamic / FPV uses its exact profile fingerprint and is kept as comparison evidence rather than a regular BRD Score input. Repeated runs are averaged arithmetically within each physical sample first, then the physical-sample means are averaged with equal weight. Missing samples are excluded rather than treated as zero; Unassigned records never enter the Average selector or mix with assigned sample averages. For a legacy cell that has public fixed-current evidence but no reviewed physical-sample assignment, BRD preserves the established score by applying the same algorithm directly to those Unassigned runs and labels the result provisional. Assigned equal-sample averages automatically replace that fallback when they become available. Repeated-condition collisions remain visible in migration diagnostics; BRD never chooses the best or latest physical sample.
Many manufacturers publish conservative limits to protect cycle life, warranty margin, heat margin, and safety margin. BRD may identify cells that show extra lab margin, but that does not rewrite the datasheet or make higher-current use safe. The manufacturer's voltage, current, charge, discharge, pulse-duration, and temperature limits remain the operating limits.
The cell performs very close to, or above, the manufacturer claim with only minor limits.
The cell is useful, but heat, sag, current proof, or missing data deserves attention.
The cell may work, but the public data does not cleanly support all datasheet expectations.
The public test data shows meaningful mismatch, excessive heat, early stops, or missing proof.
Weighted categories
Completed raw mAh evidence compared with datasheet minimum and nominal capacity. Low-current tests are preferred context, but the system accepts completed real-load capacity proof instead of blocking the score.
Peak temperature, including post-stop heat soak when available, but only from tests at or below the true datasheet continuous-current rating.
Voltage hold at the hardest fair current level using early and average loaded voltage, plus direct mV/A sag only when the captured start voltage is clean. Weak current coverage, incomplete tests, or no clean rest-to-load capture cap the sag score.
BRD Rated CC is measured from arithmetic physical-sample averages of sustained BRD constant-current discharge only. For the explicit legacy fallback, a qualifying published Unassigned sustained CC run may provide provisional proof until physical-sample identity is reviewed. Burst/Pulse, Dynamic / FPV profiles, and short timed pulls never count. The automatic value is the highest completed sustained CC condition that stays inside the datasheet max cell temperature. It may be higher than the manufacturer's continuous rating; when that happens BRD labels it as lab evidence and explains that the manufacturer may have chosen a lower public rating for cycle life, longevity, warranty, or safety margin. It is never permission to exceed the datasheet in a device or pack.
Uses the datasheet AC IR / impedance value as an electrical predictor, including estimated I*R voltage drop at the datasheet continuous-current rating. Lower IR generally means less voltage sag and less I2R heat, but live load tests remain the final proof.
Scores the datasheet cycle-life claim separately. Strong, quantified cycle claims score higher; vague or missing cycle-life information scores lower because lifespan context matters.
Rewards clear hosted datasheets with capacity, true CDR, temperature limits, AC IR, and operating temperature range. Missing cold-weather limits, narrow discharge temperature range, contradictions, and known datasheet errors receive an explained penalty.
BRD Score is not a safety certification, manufacturer approval, pack-building recommendation, or replacement for the charts. The raw curves, test date, cutoff rule, current level, temperature behavior, and BRD Notes remain the final context. Tests above true CDR remain visible as beyond-datasheet data, but they are not used to punish normal temperature scoring unless the datasheet itself claims that operating region. Evidence confidence is category-aware instead of a flat public-test count, but it is no longer a visible weighted score category. It acts as a confidence/capping guard so a cell with too little public evidence is not presented with false certainty. Low-current, balanced-current, and high-current cells each have their own fair current-band expectations: capacity proof, lower-load behavior, mid/high in-limit behavior, and a rated-current edge near the datasheet claim. A missing out-of-category test no longer reduces the regular BRD Score. Tests above true CDR remain visible as beyond-datasheet data, but they are scored separately as Hero Score. BRD Rated CC is a measured sustained-discharge lab result, not an operating recommendation. If BRD proves more current than the datasheet continuous rating, that extra margin is shown as lab evidence only and may indicate that the manufacturer selected a lower rating for cycle life, longevity, warranty, or safety margin. The automatic value is the highest completed sustained CC arithmetic sample average inside the datasheet temperature limit. Burst/Pulse tests, Dynamic / FPV profiles, and short timed pulls never define BRD Rated CC. Manual fit measurements remain visible as BE AWARE warnings because tight devices and holders matter, but fit no longer changes the BRD Score number.
BRD Hero Score, pulse claims, and temp-cut ratings
Some datasheets publish two different current ideas: a true continuous CDR and a higher temperature-limited current rating. Some datasheets also publish a pulse or burst number, for example 180A for 2 seconds. BRD Score judges normal behavior only at or below the true CDR. Tests above CDR, above pulse current, or longer than the datasheet pulse window are separated into BRD Hero Score. If the test is still inside a datasheet temp-cut rating, it is labeled as Temp-Cut Hero Score instead of ordinary continuous-current performance. Normal CC/no-temp-cut claims are judged by both decent voltage sag and controlled temperature. Max-current-with-temp-cut claims are different: heat is expected, so Temp-Cut Hero Score focuses more on voltage sag/hold and whether the cell survives the requested stress.
Safety rule: Hero and Temp-Cut Hero results are lab characterization only. They are useful for understanding voltage sag, heat, survival, and hidden margin, especially when a datasheet is conservative for cycle life. They are never an instruction, approval, or recommendation to exceed the datasheet in a pack, device, charger, flashlight, power tool, vehicle, or customer build.
How far above datasheet CDR, pulse current, or pulse time the public test went. Temp-Cut Hero uses 25% here.
Loaded voltage and measured sag at the highest Hero run. Temp-Cut Hero raises this to 45% because decent sag is the main question when heat cutoff is expected.
Whether the cell survived the requested stress window. A short pulse file can count when it beats the datasheet pulse duration.
Peak temperature during Hero runs. It is weighted lightly for beyond-datasheet Hero tests and shown as context only for Temp-Cut Hero tests.
Highest raw-mAh condition average delivered during Hero testing compared with nominal capacity. The condition input remains an arithmetic physical-sample average.
BRD Hero Score is not a pack-design recommendation, manufacturer approval, safety certification, or replacement for the datasheet. It is a separate celebration signal for viewers who want to understand voltage sag, heat, survival, and capacity when a cell is pushed beyond its datasheet in the BRD lab. A weak Hero Score does not punish the normal BRD Score, because the cell was outside its published operating claim.
CSV ingestion
- The ingestion pipeline reads tester data and builds normalized public charts.
- Base model and physical sample identity are stored separately. Missing legacy metadata stays Unassigned, and numeric model suffixes are never interpreted as samples.
- New run IDs and filenames include explicit SAMPLE01 through SAMPLE10 tokens, and CC and Burst records use separate condition identities.
- The Python ingestion tool detects header rows, voltage, current, power, capacity, energy, and temperature columns.
- Power, capacity, and Wh are calculated when the CSV does not provide them.
- Malformed rows and missing columns are logged.
Fairness checks
- Files ending above target cutoff are flagged as early stops.
- Files that do not reach 3.0V are flagged as incomplete for usable-above-3.0V ranking.
- Mixed test currents are shown as cross-condition comparisons.
- Sample averages require completed runs with matching base model, mode, current, cutoff, and mode-specific condition. Burst duration must match; Dynamic / FPV profile fingerprints must match. Curves are aligned only over their shared domain, using overlapping elapsed time for Dynamic / FPV profiles.
- Raw total mAh and Wh are not treated as fair rankings when cutoff voltages differ.
Standard chart order
- Voltage over Watt-hour
- Power over Watt-hour
- Temperature over Watt-hour
- Current over Watt-hour
- Capacity over Watt-hour
- Voltage over Time
- Power over Time
- Temperature over Time
- Combined Capacity Page, with Capacity over Time and Capacity over Watt-hour
Public result scope
Public viewing
- Visitors can view processed charts, rankings, summary metrics, and fairness notes.
- Result pages focus on interpreted test data, dates, safety flags, and datasheet context.
- The public comparison tools are designed for browsing and sharing results.
Future improvements
- Additional report formats may be added as the project grows.
- Special data questions can be handled manually when needed.
- Published charts remain the main public reference for comparison.
Important metric definitions
| Metric | Meaning |
|---|---|
| Raw capacity mAh | Total measured capacity to the file ending point. |
| Raw Wh | Total measured energy to the file ending point. |
| Usable capacity above 3.0V | Capacity delivered before the curve first reaches 3.0V. If the file never reaches 3.0V, it is marked incomplete. |
| Usable Wh above 3.0V | Energy delivered before the curve first reaches 3.0V. This is often the practical high-drain comparison anchor. |
| LAB Room Temp | Room temperature entered during ingestion, used for temperature rise when available. |
| Early loaded voltage | Voltage near the beginning of real load, useful for initial sag comparison. |