One number, several levels
A specification sheet that offers four brightness levels and one runtime figure is describing the level the factory chose to measure, and nothing else. The figure usually comes from the lowest setting, run in a room at a comfortable temperature, on a fully charged pack, until the driver cuts out or the light has fallen to a defined fraction of its start value. Change any one of those four conditions and the number moves, sometimes by a factor rather than a percentage. Yet the sheet presents the result as a property of the product rather than a property of the test.
For a buyer the practical consequence is that a comparison built on headline runtime figures is a comparison of test choices, not of products. Two lanterns that both advertise a long run may differ because one has a lower bottom setting, one has a looser cutoff rule, and one was measured in a warmer room. None of that is deception by itself. It is missing context, and missing context is what turns a purchase order into a dispute several weeks later.
A per-level table is not a luxury. It is the smallest document that makes the claim comparable, and it is the document a sourcing team should ask for in the first round rather than at sample approval.
What each runtime row has to carry
Each brightness level needs its own row, and each row needs to be more than a number. The level name as the supplier calls it is the minimum. Two suppliers can both label something "medium" while one is driving a single LED string and the other is driving a COB array at reduced current, and the two will behave differently at the end of the run.
The starting charge is the second field. A runtime measured from a full pack is the best case. A runtime measured from a pack at its shipping state of charge is a different and often more honest figure for a user who opens the box and switches on immediately. The record should say which one it is, and whether the pack was rested after charging or tested straight off the charger, because the difference between a rested pack and a freshly charged one can be visible in the first minutes.
The ambient temperature is the third. The lantern loses heat to the room the whole time, and a cold room raises the drive the control circuit applies, which drains the pack faster. The fourth is the load condition: free air, a tent wall, a table, or a close work surface. The fifth is the cutoff rule, which deserves its own treatment below because it is where most of the ambiguity hides.
The six fields a runtime row should contain
| Field | What belongs in it | Why it changes the answer |
|---|---|---|
| Level name | The supplier's own label, plus the drive state | Two labels can hide different drive configurations |
| Starting charge | Full, rested, or shipping state of charge | Sets the best case versus the realistic case |
| Ambient | The temperature held during the run | Cold ambient raises drive and shortens runtime |
| Load | Free air, tent wall, table or close work surface | A nearby surface changes how fast heat leaves |
| Cutoff rule | Low-voltage cut, thermal derate, or stated output fraction | Each produces a different end point on the same pack |
| Sample identity | Golden sample or production batch, with the date | Separates a design record from a batch claim |
Ask for the table, not the headline. A supplier who can fill six rows for every level has run the test and kept the record. A supplier who can only repeat one number has not, or has decided not to share it.
The conditions that move the number
Ambient is the largest single lever. A lantern in a cold tent fights a bigger temperature difference, so the driver spends more time at full duty and the pack drains faster than the same unit in a mild room. That is physics, not a defect, but it means a runtime figure without an ambient value cannot be compared with a competitor's figure.
The drive setting is the second lever, and it is the one that separates a fair comparison from an unfair one. A lantern held at its lowest level for the whole run will post a long figure that no user will actually experience, because a user typically starts at the highest level and steps down. The honest figure is a pair: the runtime at the highest level and the runtime at the level the user will settle on.
Thermal management is the third. As the pack and the LED heat up, a sensible driver derates to protect both, which the user experiences as a gradual dimming. A driver that instead maintains drive until the protection trips ends the run abruptly. Both are legitimate, and the choice belongs in the firmware note. A supplier reporting a long runtime should be able to say whether the run stayed inside the thermal envelope of the pack and the array.
Longest runtime ≠ most useful light
The boundary to state plainly: the longest runtime is not a measure of useful light. A lantern that holds a modest, stable output throughout can be more useful than one that starts bright, derates early and spends its last third barely visible. The second boundary is that a runtime figure is not a battery capacity claim. Capacity lives in the pack specification, and converting a runtime into a capacity expectation requires the drive power, the duty cycle and the losses, none of which a runtime figure contains. If a supplier uses runtime to imply capacity, ask for the pack datasheet instead.
A third boundary: a runtime figure is not a promise about every level. The longest figure may belong to the lowest setting, while the highest setting may run a fraction of that time.
Five questions for the supplier
- For each brightness level, what runtime was recorded, and what was the drive state behind that level? Level name plus drive state is the minimum for a comparable row.
- What was the starting charge, and was the pack rested after charging? A rested full pack and a freshly charged one can differ noticeably.
- What ambient temperature was held during the run, and where was the lantern positioned? Free air and a close work surface are different loads.
- What ended the run: low-voltage cutoff, thermal derate, or the point where the output fell to a stated fraction of its start value? The end point defines the number.
- Was the run continuous or intermittent, and which sample and batch does it represent? Continuous and intermittent runs are different tests, and a golden sample is not a batch guarantee.
Next step
Request the six-row table for every brightness level, plus the firmware note that explains the cutoff behaviour, and keep both with the sample approval file. When the numbers arrive in that shape, the comparison with a second supplier becomes mechanical rather than interpretive. Related reading: tent camping lighting for how runtime interacts with beam and colour in a tent, emergency backup lighting for the recharge side of the same cycle. To apply this to a live sourcing project, start at contact.
Part of the Lumizuno outdoor-lighting notes. Illustrative values only; lumens and ingress ratings must be read with their test conditions.