Battery tools

What it calculates

The minimum battery capacity for a voice alarm system to BS 5839-8:2023, Annex C: enough to run the system quiescent for the standby period (normally 24 hours), then broadcast at full alarm load for the alarm period (minimum 30 minutes), with the standard's 1.25 allowance for ageing and temperature. The result is the capacity at the 20 hour rate (C20) that the battery bank must equal or exceed.

The number the standard won't give you

Annex C requires derating factors taken "from the battery manufacturer's data", then admits in the same breath that this data is often not available. That is the number this calculator supplies. It derives the factors from whatever the datasheet does publish: the manufacturer's constant-current tables where they exist, Peukert's law where they do not, and it always tells you which source it used. The answer is only as good as its source, so the calculator ranks them for you, best first: manufacturer table, then a Peukert fit from rating points near the alarm duration, then a fit from long-duration points (floored for safety), then a conservative default.

The Peukert rated-capacity method used by this calculator follows the work of Chris Gibson of SmartGauge Electronics, whose writing on Peukert's law also shaped our own understanding of the subject. His original articles are well worth a read at smartgauge.co.uk.

Where quiescent current goes wrong

Quiescent current is the number that decides the answer. The standby period is normally 24 hours and the alarm period 30 minutes, so the idle current is multiplied by 48 times as many hours as the alarm current. Get the efficiency assumption wrong and the answer moves by about one per cent. Get the quiescent current wrong and it moves by hundreds.

It goes wrong in three ways, all of them on the datasheet rather than in the arithmetic.

Reading the standby current when the system will actually sit supervised. Where a manufacturer prints both, the supervised figure runs from twice to eighteen times the standby one.

Reading a per-unit current as per channel, or the reverse. A four-channel amplifier is a factor of four either way.

Reading a maximum, a fuse rating or a supply sizing figure as a consumption. If a stated current implies an amplifier efficiency above about 90 per cent, or above 100 per cent, it is a rating and not a measurement.

Where the manufacturer publishes nothing, measure on site with the mains isolated, and let the reading settle.

Which level to use, and when to move up

Start at Level 1. Two questions, a first estimate, and every assumption it made listed beside the answer, so you can see exactly what it guessed on your behalf.

Move up to Level 2 once you have the real system data: loudspeaker load, quiescent current, battery voltage. This is the level most design work needs.

Use Level 3 when you have the full picture: site measurements from an existing system, an unusual message cycle, or a specific battery whose datasheet you want the calculation built on.

You do not have to choose the right level first time. Start simple and climb as the information arrives; each level carries the same result forward and takes it further.

One thing worth knowing before you specify a battery. The lighter levels protect you by assuming the worst: Level 1's guesses are deliberately conservative, so its answer usually sits above the real requirement. Feed in measured data at Level 2 or 3 and the number often comes down, sometimes to a smaller, less expensive battery that still meets the standard in full. Refining is not only about a more accurate figure; it can stop you fitting more battery than the system actually needs.

Where the derating comes from

The tag beside the C_min figure tells you how firm the derating factors are. In order of preference:

table: the manufacturer publishes a constant-current discharge table and we read the 30 minute figure straight off it. Nothing is inferred. This is the best case.

peukert-fit (2-point): the datasheet gives two rating points: a current and the duration the battery holds it for. Peukert's law plots as a straight line on log-log axes, so two points fix the exponent k, and the derating factors follow from it.

peukert-fit (4-point, NP Shortform): the datasheet gives four rating points, so instead of drawing a line through two of them we fit the single line that lies closest to all four. That is a least-squares fit: the line whose total squared distance from the four points is smallest. It matters because a line through any two points gives a different k depending on which two you pick, whereas the fitted line uses all the published evidence. For the Yuasa NP range this gives k = 1.209.

k entered: you supplied the Peukert exponent yourself.

D1 and D2 entered: you supplied the derating factors directly and the calculator applied them unchanged.

k is floored at 1.20. A lower exponent than that would flatter the battery, and on a life-safety calculation we would rather be a little pessimistic than a little optimistic.

What does 14/14 mean?

The calculator checks its own maths whenever you run the self-tests: 14 scenarios with known correct answers, including the standard's worked examples, are recomputed and compared. 14/14 means every check passed. If it ever shows less, do not rely on the results, and please let us know.