Search productSearch post

Company Blog

Backup Power Design for LED, Emergency Lighting, and UPS Equipment

blogs 00

Most backup power conversations start in the wrong place. Someone picks a battery because the capacity number looks right, and six months later the fixture fails its annual inspection, or the UPS beeps and dies in eleven minutes instead of thirty. The capacity number was never the problem. The problem was that nobody asked what the battery actually has to do during an outage, and for how long, and in what temperature, and who signs off on all of it.

This article walks through backup power design for three kinds of equipment that live or die by their batteries: LED emergency lighting, dedicated emergency luminaires and exit signs, and small UPS units. We will cover what the codes ask for, how to size a pack with real math instead of guesswork, where the 21700 battery format fits and where it doesn’t, what good battery pack assembly looks like from the outside, and when a private label arrangement makes sense for a distributor or channel brand. There’s also a section on smart home devices, because that market keeps showing up on the same quotation sheets.

What the code asks of the battery

If your equipment is emergency lighting for a commercial building in North America, the battery is not really your decision to make freely. The building codes have already decided most of it for you.

NFPA 101, the Life Safety Code, requires emergency lighting along means of egress to run on battery or generator backup for a minimum of 90 minutes after normal power is lost. Illumination has to come back within 10 seconds of the outage. The equipment itself must be listed to UL 924, the standard for emergency lighting and power equipment, by a nationally recognized testing laboratory. And the system gets tested on a schedule: a 30-second functional test every month, and a full 90-minute discharge test once a year, with records kept. Failing the annual test is the classic way a building finds out its backup batteries were dead all along.

UL 924 also sets performance expectations that shape battery selection upstream. The battery must recharge so the unit is ready for another outage within about a day, and it must hold up across the temperature range of the installation environment. Stairwells get hot in summer and cold in winter. Equipment mounted outdoors or in parking structures sees worse.

What does this mean for you as an equipment designer or buyer? Three things:

  • The finished fixture carries the UL 924 listing, not the battery inside it. If you manufacture emergency luminaires, your certification work happens at the fixture level, with the battery installed. Swapping battery suppliers after listing usually means re-evaluation, not just a purchase order change.
  • The battery still needs its own paper: cell and pack level safety reports (UL 2054 or IEC 62133 depending on the market), UN 38.3 for transport, and an SDS. More on that in the documents section.
  • Design margin matters more in this market than almost any other. A consumer product that dies at 85 minutes instead of 90 gets a bad review. An emergency luminaire that dies at 85 minutes fails an inspection that generates paperwork, findings, and possibly fines.

For small UPS equipment the code pressure is lighter, but the customer expectation isn’t. A UPS that silently loses capacity over two years on float charge is a support ticket waiting to happen, and the buyer of a UPS is exactly the kind of person who tests it.

Sizing a pack: a worked example

Let’s do the arithmetic once, properly, so you can reuse the method.

Suppose you’re building an LED emergency driver that must hold a 10 W LED load at reduced output for 90 minutes, per the code minimum.

Step 1: energy at the load. 10 W × 1.5 h = 15 Wh. That’s what the LED needs.

Step 2: add the conversion chain. The battery doesn’t feed the LED directly. There’s a driver, an inverter or boost stage, wiring losses, and the BMS. At this scale, 80–85% end-to-end efficiency is realistic. 15 Wh ÷ 0.82 ≈ 18.3 Wh at the battery terminals.

Step 3: derate for reality. The battery must still deliver at end of life, at the low end of its temperature range, after years on float charge. Lithium cells typically age to 80% of rated capacity in service, and cold knocks another chunk off. A combined derating factor of 0.7 is a defensible starting point for a code-driven application. 18.3 Wh ÷ 0.7 ≈ 26 Wh of rated capacity.

Step 4: pick the configuration. If the driver runs from a 10.8–12.6 V rail, a 3S lithium-ion pack (11.1 V nominal) works: 26 Wh ÷ 11.1 V ≈ 2,350 mAh. A single 21700 cell is 5,000 mAh at 3.7 V — roughly the right energy per cell, so you’d either run 3 cells in series or reconfigure the electronics. If the driver tolerates a wide input, a 1S high-capacity cell with a boost stage is cheaper and simpler.

ParameterValue in this exampleNote
Emergency load10 W LEDReduced-output emergency mode
Required duration90 minutesNFPA 101 minimum; some specs ask for 120
Energy at load15 WhLoad × time
Energy at battery≈18.3 WhAfter conversion losses
Rated capacity needed≈26 WhAfter end-of-life and temperature derating
Pack example3S2P 21700, 11.1 V 5,000–10,000 mAhHeadroom for a higher-output variant

One habit worth building: size to the annual test, not to the datasheet. The pack that passes year one and fails year three wasn’t undersized on paper — it was undersized against its own aging curve. Every watt-hour of margin you add in step 3 buys you years at the end of service life.

Cell choice: where the 21700 battery fits

The 21700 battery format — 21 mm diameter, 70 mm long — has quietly become the default cell for mid-size backup applications, and for good reasons that have nothing to do with fashion.

First, energy per cell. Modern 21700 cells run up to around 5,000 mAh at 3.7 V nominal, which is roughly 18–19 Wh in a single can. Compared to the older 18650 format, you get 40–50% more capacity per cell, which means fewer weld points, fewer cells to buy and match, a smaller BMS, and fewer things in the pack that can fail. For an emergency luminaire or a compact UPS where the battery lives in a housing designed a decade ago for a lead-acid brick, that density is the difference between fitting and not fitting.

Second, supply chain maturity. 21700 cells are produced at enormous scale for e-mobility and power tools, which means stable supply, competitive pricing, and multiple qualified manufacturers. When your product has a ten-year service life, cell availability five years from now is a real procurement question, and the 21700 answer is reassuring.

Third, the format is versatile across chemistries. The same footprint exists in high-energy NMC and in LiFePO4 variants, and KBT supplies both from its 21700 cylindrical cell line. If you want to see what’s available, the 21700 cylindrical rechargeable lithium-ion and LiFePO4 battery cells page lists the current options.

Where does the 21700 battery not fit? Two places, mainly. Very large UPS systems and energy storage cabinets usually move to large-format LiFePO4 prismatic cells or rack modules — see KBT’s large-capacity LiFePO4 energy storage batteries for that end of the range. And extremely slim devices, like thin emergency exit signs or flat-panel luminaires, need pouch cells instead, because a cylinder is a cylinder no matter how good it is. Chemistry choice follows the duty cycle: NMC/Li-ion for energy density in confined housings, LiFePO4 where cycle life, thermal stability, and long float service outweigh size.

A practical note on matching: for packs with cells in parallel, buy matched cells from one production lot where possible, and insist that your supplier documents the matching criteria (capacity, internal resistance, voltage at rest). Mismatched parallel strings age unevenly, and the weak cell drags the pack into early failure — exactly the failure mode the annual 90-minute test is designed to catch.

What good battery pack assembly looks like

Two packs can carry the same cells and the same spec sheet and still have completely different service lives. The difference is almost always in the battery pack assembly — the part of the process that never shows up on a quotation line item.

Here’s what to look at when you evaluate a pack supplier’s assembly capability:

  • Cell interconnects. Spot-welded nickel strip is the standard for cylindrical cells; the thickness and material of the strip, and the weld quality, decide whether the interconnect handles the current and survives vibration. For higher-current packs, laser welding or nickel-plated copper enters the conversation. Ask what current path they’d use for your spec.
  • BMS placement and protection. The protection board should sit where it can sense the cells it protects, with balanced sense leads, and its cutoff thresholds should match your application — a UPS that needs every last percent wants different thresholds than a toy. Overcurrent, overcharge, over-discharge, short circuit, and (for some chemistries and certifications) temperature protection are the baseline set.
  • Insulation and mechanical fixing. Fish paper, PET, or nomex between cells and case; cells fixed so they can’t shift; leads strain-relieved where they exit. Backup equipment gets installed, bumped, and forgotten for a decade — the pack has to survive the installer, not just the outage.
  • Float and charging behavior. Emergency lighting packs sit on a trickle/float charger for years. The charger profile and the BMS have to be designed as a pair. A pack that’s great at discharging but drifts out of balance after two years on float will fail exactly when the inspection arrives.
  • End-of-line testing. Every pack should leave the line with a functional test — capacity check or at minimum a full charge/discharge screen on a sample basis per lot, plus a protection-function verification. Ask what the test report for your shipment looks like.

KBT’s own description of its business is worth quoting here, because it maps to this list: the company combines battery pack engineering, assembly and testing with cell manufacturing resources — which means the cells and the assembly are under one engineering conversation instead of two. You can read how they position it on the about page, and it’s a fair checklist for evaluating any supplier, including their competitors.

If you take one thing from this section, take this: when you compare quotes, ask each supplier to describe their assembly process for your specific pack, in writing. The one who can answer in specifics — weld parameters, BMS part number, test report format — is the one whose packs will still be passing the annual test in year five.

Private label lithium battery packs

Distributors and channel brands in lighting and power equipment often reach this decision point: sell someone else’s branded pack, or put your own brand on the product. Private label lithium battery packs are the middle path — the supplier manufactures and assembles the pack, and it ships with your branding, your packaging, your part numbers.

When does it make sense? Usually when three things are true. You have ongoing volume, not a one-off project. You have customers who ask for your brand on the replacement part (which is exactly the situation in emergency lighting replacement batteries and UPS replacement packs — facility managers buy the brand on the fixture). And you’d rather spend your capital on channel and service than on a factory.

If you go this route, a few things deserve attention before you sign anything:

  • Certificates don’t transfer with the logo. The cell and pack certifications (UN 38.3, IEC 62133, UL 2054, and the cell manufacturer’s own reports) belong to the products as manufactured and tested. Your brand on the label doesn’t make you the certificate holder, and it doesn’t make the finished equipment listed — that still happens at the fixture or UPS level. Ask explicitly whose name appears on each document, and get copies with your shipment records.
  • Labeling is a compliance surface, not just marketing. Ratings, warnings, chemistry marking, and batch traceability all live on that label. Agree on the artwork with the supplier’s compliance person, not just their salesperson.
  • MOQ and carton economics. This is where private label gets real. KBT’s published approach is that standard models can be ordered from a single carton, while custom pack MOQ depends on the cell, BMS, enclosure, connector, tooling, and certification requirements — and for custom projects, small-batch MOQ is supported so you can test demand before mass production. That’s a reasonable structure to benchmark others against. The details of KBT’s OEM project process, from requirement review through sample customization to batch production, are laid out on the support page.
  • Change control matters more with your name on it. If the supplier substitutes a cell or BMS revision mid-contract, your customers find out before you do. Put a notification clause in the agreement.

Private label is a good arrangement when the supplier is good. It’s a brand-damaging arrangement when the supplier is mediocre and your logo is the one on the failure. Choose accordingly, and let the assembly-quality questions from the previous section do the choosing.

The adjacent market: smart home backup

Battery backup conversations rarely stay within commercial lighting. The same distributors who quote emergency drivers get asked about residential products, and the most common crossover is a battery pack for smart home devices — hubs and gateways, smart locks, routers and mesh nodes, security sensors, and the small UPS-class units people now keep under their desks.

The requirements rhyme with commercial backup power, at smaller scale. A battery pack for smart home devices has a different job description than an emergency luminaire’s: a hub that bridges a lock, sensors, and cameras should ride through a utility outage for a few hours, not 90 minutes — long enough for the homeowner to notice, check the camera, and not lose the automation state. That’s typically a 2,000–5,200 mAh class lithium-ion pack with a simple BMS and a DC output connector, specified less by code and more by product marketing (“keeps your home secure for up to 8 hours during an outage”).

What changes at this scale:

  • Size and shape dominate. Consumer housings leave awkward spaces, so flat pouch cells or slim 1S/2S packs beat cylinders. Connector choice (JST, DC5521, USB-C input) becomes a spec-line decision, and getting it wrong means a redesign of the enclosure, not just the pack.
  • Charging is usually internal and always-on — the same float-charge longevity question as emergency lighting, but with less thermal mass and less engineering attention. It deserves the same scrutiny.
  • Certification expectations are lighter but real: IEC 62133 / UL 2054 class battery reports, UN 38.3 for shipping, and the usual SDS. Consumer retail channels add their own vendor requirements on top.

An example of this class in KBT’s catalog: the 11.1 V 5,200 mAh pack with DC5521 output and a 12 V charger, built for CCTV and LED lighting — a bridge-power pack for security and lighting equipment with the BMS and charger supplied as a matched set. And for larger residential backup, the 12 V lithium-ion pack with cable family covers the higher-drain end. The engineering logic doesn’t change from the commercial side; only the numbers do. Which is, honestly, the point of designing it well the first time.

Documents to ask for

Whatever the application — emergency luminaire, UPS, or smart home bridge power — the document set you should expect from a pack supplier is similar:

  • UN 38.3 test summary — mandatory for transport since 2020, per model.
  • Safety report for the battery: IEC 62133-2 or UL 2054, at cell or pack level as applicable. Which one your market and channel requires depends on the finished product’s certification route.
  • MSDS/SDS — needed by forwarders and by your own customers’ EHS departments.
  • Shipping classification and packing instruction reference (UN3480 or UN3481, and the corresponding IATA/IMDG packing instruction) — your forwarder will ask; it’s better if you already know.
  • Batch test records — capacity and protection-function screening for the lot your order came from.
  • Warranty terms with a failure analysis commitment — not just a duration, but what happens when a pack comes back.

And one boundary to keep straight, because it’s the most expensive misunderstanding in this market: battery documents are not equipment approval. A pack with a full IEC 62133 file and a clean UN 38.3 summary is a well-documented component. The UL 924 listing of an emergency luminaire, or the safety approval of a UPS, is evaluated at the finished equipment level, with the battery installed, by the equipment manufacturer. If you buy packs for equipment you certify, plan the re-evaluation whenever the battery changes. If you buy finished fixtures, ask the fixture maker — not the pack supplier — about the listing.

FAQ

Can I use a 21700 battery pack to replace lead-acid in an existing emergency luminaire?

Electrically, often yes — a 3S lithium pack approximates a 6 V or 12 V lead-acid rail, and the energy density gain is dramatic. But the fixture was listed with the original battery configuration. Replacing it changes the input characteristics, the charging circuit load, and possibly the emergency driver behavior, which puts the listing into question. Lithium conversions of code-governed equipment should go through the fixture manufacturer or a re-listing process, not through a swap on a ladder.

How long should a backup pack last in float service?

Plan around 3–5 years of useful float service for lithium-ion in typical indoor conditions, with LiFePO4 often extending beyond that in favorable temperatures. The practical guidance: build the derating from the sizing example into the pack from day one, and budget replacement on a schedule rather than waiting for the annual test to find the failures one unit at a time.

Is private label possible for small quantities?

Custom-packaged private label orders generally carry higher MOQs than standard models because of tooling, label setup, and packaging. KBT’s published structure supports small-batch MOQ on customization projects so a brand can validate demand first — if your volume is genuinely small, starting with a standard model under their brand and moving to private label once volumes justify it is usually the cheaper path.

Backup power is one of those product categories where the buyer hopes never to see the product work. That’s exactly why the engineering has to be unglamorous and a little pessimistic — sized for end of life, tested on a schedule, documented well enough that the inspector’s clipboard comes back clean. If you’re specifying a pack for LED emergency equipment, UPS units, or anything else that has to be there during an outage, you can contact KBT with the load, the duration, and the market, and have the configuration and its documents confirmed before you commit to a design.

The prev: The next:

Related recommendations