TL;DR: Choosing a charging IC without mapping it to the specific standard that governs your end product’s target market is one of the most common — and most expensive — compliance failures we see in portable and stationary BESS design.
TL;DR: A single missed test clause in IEC 62368-1 vs. IEC 62133-2 can delay market entry by 4–6 months and add $23,000–$47,000 in re-certification costs depending on CB scheme lab backlog.
Which Standard Actually Governs Your Charging IC — and Why Most Design Teams Pick the Wrong One #
The confusion starts early. A design engineer selects a charging IC, pulls the IC vendor’s compliance documentation, sees “IEC 62368-1 compliant” on the datasheet, and assumes the box is checked. It isn’t. IEC 62368-1 governs audio/video, IT, and communication equipment at the system level. It says nothing specific about the electrochemical behavior of the cell being charged. If your product is a portable power station or a compact BESS, the charging IC’s role in cell protection falls under a different standard hierarchy entirely — and conflating the two creates compliance gaps that surface late, usually during final certification testing.
The relevant standard stack for charging ICs in battery products typically looks like this:
- Cell-level safety: IEC 62133-2:2017 (portable sealed secondary lithium cells and batteries)
- Pack/system-level safety: IEC 62619:2022 (stationary and portable secondary lithium cells and batteries — safety requirements)
- Transport classification: UN38.3 (Recommendations on the Transport of Dangerous Goods, Manual of Tests and Criteria, Part III, Section 38.3)
- End-product system safety (portable): IEC 62368-1 Ed.3
- End-product system safety (stationary BESS): IEC 62477-1 / IEC 62040 series
Your charging IC sits at the intersection of IEC 62133-2 and IEC 62619 requirements — specifically the overcharge protection thresholds, temperature cutoff windows, and charge termination accuracy those standards mandate. Getting those parameters wrong at the IC selection stage means the pack-level certification fails, not the component test.
The standard selection matrix below reflects what we use internally during our AVL gate review for charging IC qualification:
| Product Category | Primary Standard | Charge IC Relevance | Market Mandate |
|---|---|---|---|
| Portable power station (consumer) | IEC 62133-2 + IEC 62368-1 | OVP, OCP, temp cutoff | EU (CE), UK (UKCA), Japan (PSE) |
| Portable power station (industrial) | IEC 62619:2022 | OVP, OCP, balancing current, NTC spec | EU (CE), Australia (RCM) |
| Compact stationary BESS (<10kWh) | IEC 62619:2022 + IEC 62477-1 | Charge termination accuracy, isolation | EU, US (UL 9540A optional but common) |
| E-mobility charging | IEC 62133-2 + UN38.3 | All protection layers | Global (transport + end-use) |
| OEM cell packs for export | UN38.3 (transport) + destination market | Cell-matching, SOC at shipment | Global mandatory |
The GB/T standards deserve their own treatment. GB/T 31241-2022 (portable lithium batteries) is the Chinese national equivalent of IEC 62133-2, but with tighter nail penetration pass criteria and a different short-circuit test duration. If you’re sourcing a pack from a Shenzhen-based pack house for domestic Chinese distribution, GB/T 31241 governs and it does not automatically satisfy IEC 62133-2 — the test protocols diverge enough that you cannot cross-claim one cert for the other market. GB/T 36276-2023 covers electrochemical energy storage for stationary applications and aligns more closely with IEC 62619 in scope, but the thermal stability test conditions differ by approximately 15°C in the abuse temperature profile.
The Root Cause Most Design Teams Miss: Charge Termination Tolerance Stacking #
Here is where I’d prioritize your attention before any other compliance topic.
IEC 62133-2 Clause 7.3.8 and IEC 62619 Clause 7.3 both specify overcharge test conditions, but neither explicitly states a voltage accuracy tolerance for the charging IC. That tolerance lives in the IC’s datasheet as a VOVP parameter — typically ±1% to ±2.5% depending on silicon process node and temperature coefficient. The standard’s test protocol charges the cell to 1.2× the manufacturer’s maximum charge voltage. If your IC’s VOVP threshold has a +2.5% tolerance at 60°C, and your cell’s absolute maximum voltage is 4.20V, you’re potentially delivering 4.20V × 1.025 = 4.305V before protection triggers. That’s 105mV above the cell’s absolute maximum. Some cells tolerate this briefly. Others begin lithium plating.
The cascade goes like this: marginal VOVP accuracy → occasional overcharge events at temperature extremes → lithium plating on anode → dendrite formation over 200–400 cycles → internal short risk. The cell doesn’t fail on the certification bench. It fails 18 months into field use, in the hands of an end customer. When the incident investigation traces back to the charging IC spec, the question from the certification body is always the same: “What VOVP accuracy did you verify at 85°C during design validation?” If your DVT report doesn’t include a temperature-swept VOVP characterization at 3 temperature points minimum (0°C, 25°C, 85°C), expect a mandatory retest request.
Confirmation method: apply a precision bench supply set to VOVP_nominal + 50mV (simulating an upstream regulation error), log the actual protection trip voltage using a 16-bit DAQ at 10ms intervals, across a temperature-controlled chamber sweep. Acceptance threshold: VOVP must trip before reaching VMAX_cell at all temperatures. Any IC that fails this at 60°C or above should be removed from your AVL regardless of datasheet claims.
We run this as part of our QC-IC-04 incoming qualification procedure for every new charging IC vendor. In our 2023–2024 review of 11 IC sources from Shenzhen and Dongguan BMS manufacturers, 4 showed VOVP drift exceeding 1.8% at 85°C — two of which had “±1% across temperature” printed on their datasheets.
Corrective Actions Ranked by Impact and Feasibility #
-
Revise IC selection criteria to include VOVP accuracy at Tmax. Pull the full electrical characteristics table, not just the headline spec. If the datasheet shows VOVP at 25°C only, request the temperature characterization data from the IC vendor. This costs nothing and eliminates the majority of overcharge tolerance risk before you build a single prototype.
-
Add a hardware OVP redundancy layer at pack level. A secondary OVP IC (e.g., a dedicated cell protector IC in series with the primary BMS FET path) provides a hard cutoff independent of the primary charging IC. This fixes roughly 80% of the residual overcharge risk cases we encounter but adds $0.18–$0.34 per cell to BOM cost and requires PCB layout changes. Worth it for any product targeting IEC 62619 certification.
-
Specify charge termination accuracy in the supplier brief, not just protection thresholds. Most procurement specs list VOVP but omit charge termination voltage accuracy (the CV phase termination, typically ±5mV or better for LFP, ±10mV for NMC). IEC 62133-2 Annex A test conditions assume the charger terminates accurately — if it doesn’t, the cell is being cycled outside its validated window every cycle.
-
Map your standard requirements before IC sourcing, not after. This sounds obvious, but we see it backwards constantly. The right sequence is: define target markets → identify mandatory standards → extract charging IC-level requirements from each standard → then search for ICs. Most teams start at the IC catalog.
-
Request the IC vendor’s own IEC 62368-1 component qualification report. This doesn’t cover cell-level compliance (as noted above), but it does verify the IC’s electrical isolation, creepage distance compliance, and protection circuit reliability under stress. Absence of this report from an IC vendor is a signal about their certification maturity overall.
Prevention — What to Specify Upfront to Avoid This Failure Mode #
In your purchasing specification for a charging IC, include explicit language for: VOVP accuracy at 0°C, 25°C, and 85°C (not just 25°C); charge termination current threshold (C/10 minimum for LFP); NTC thermistor response time and shutdown threshold (typically 60°C ± 2°C for portable, 45°C for some stationary applications per IEC 62619 Annex D); and short-circuit detection response time (<500µs for most pack-level compliance).
For stationary BESS applications targeting the European market, also reference IEC 62477-1 Clause 5.2.3 requirements for power electronic converter protection — your charging IC may interface with a converter stage that has its own compliance requirements independent of the battery standard.
Request the IC vendor’s application note for compliance-mapped designs in your target standard. A vendor who cannot produce one probably hasn’t done it.
Sourcing Guidance for Buyers #
When evaluating Chinese IC vendors and pack manufacturers in this category, the first document to request is the IC vendor’s DVT report with temperature-swept VOVP characterization. Its absence doesn’t always mean the product is non-compliant — some smaller Shenzhen-based vendors have good silicon but poor documentation practices. But it does tell you immediately whether they can support a compliance audit.
The qualification red flag specific to charging ICs: any manufacturer who presents a single CB test certificate covering multiple product SKUs with different cell chemistries. UN38.3 and IEC 62133-2 certifications are chemistry-specific and configuration-specific. A cert for a 3S LFP pack does not cover a 4S NMC pack from the same facility. If the cert numbers don’t match the exact cell count and chemistry of your order, it’s a shared certificate and should be treated as unverified.
For incoming inspection, we recommend a sample of 32 units per lot (per ANSI/ASQ Z1.4 Level II, AQL 1.0) with bench verification of VOVP at 25°C. Any lot showing more than 1 unit outside ±2% of the datasheet VOVP value fails the lot and triggers a supplier corrective action request. For BMS engineering considerations that affect how these protection thresholds interact with pack design, the protection hierarchy matters as much as the individual IC spec.
Understanding how these standards map to your cell technology selection is equally important — because the IC’s required protection thresholds are determined partly by the cell’s chemistry and partly by the standard’s abuse test conditions, and those two inputs have to be resolved together before you finalize your IC shortlist.
Published by compactbess.com Technical Team | Request a sourcing consultation
We ran into this exact standard-mapping issue on a compact BESS design last year — during IEC 62619:2022 compliance testing the certifying lab flagged our NTC thermistor placement because the cutoff response time at 60°C ambient didn’t meet the standard’s charge termination requirements, even though our FEA showed junction temps well within the IC vendor’s 85°C absolute max. Turns out the IC datasheet thermal spec and what 62619 actually mandates for pack-level cutoff response are two completely different things, and we’d been designing to the wrong reference.
On the IEC 62477-1 isolation requirement for compact stationary BESS — are you seeing labs interpret the reinforced insulation creepage distances differently for charging IC primary-to-secondary paths versus the pack’s main contactor isolation, or has that been consistent across CB scheme labs you’ve worked with?
Ran into the OVP threshold tradeoff hard on a 48V rack-mount unit destined for both EU and Australia RCM — the IC we’d designed around had a 4.25V/cell OVP that was fine for IEC 62619 but our cell vendor’s datasheet upper limit was 4.20V, so we were technically sitting in a gray zone that the RCM assessor flagged immediately. Tightening to a 4.18V cutoff with a secondary hardware latch added $1.40 per unit in a discrete comparator circuit but we couldn’t find a single charging IC with the right threshold granularity that didn’t also balloon our BOM in other ways.
UN38.3 Section 38.3.4.6 (the overcharge test) is one that catches teams off guard when their charging IC’s OVP threshold sits right at the boundary — we had a 21700-based pack where the IC’s 4.25V cutoff passed IEC 62619 cell-level review fine but the UN38.3 test lab applied a 1.2x charge voltage multiplier during abuse conditioning and the cell went into thermal runaway at cycle 3. Tightening to 4.20V OVP solved it but required a full re-spin of the balancing topology.
The standard-stack framing in this piece is accurate but the part that burned us was IEC 62133-2 cell-level vs. IEC 62619 pack-level attribution during a CB scheme audit — the lab wanted explicit documentation showing which protections were being claimed at which layer, and our charging IC vendor’s compliance letter didn’t distinguish between the two. Took us about 11 weeks to get a supplemental test report from the IC manufacturer that satisfied the auditor’s traceability requirement, which was 11 weeks we absolutely didn’t have budgeted.
SOC estimation accuracy doesn’t get enough attention in these standard-mapping discussions, but IEC 62619:2022 clause 6.2 indirectly forces the issue — if your coulomb counter has a ±3% integration error that compounds over a 200-cycle aging curve, your actual OVP headroom shrinks in ways the IC vendor’s static threshold spec doesn’t account for. We’ve seen this bite on a 14S LiFePO4 pack where the BMS was reporting “safe” SOC but electrochemical reality had drifted enough that the protection layer was operating on stale data.
IEC 62619:2022 clause 7.2.4 tripped us up specifically on charge termination current threshold — the standard’s end-of-charge current floor interacted badly with our IC’s programmable termination resistor resolution, and at pack level the accumulated error across 16 cells pushed us outside the acceptable window during testing. Wasn’t an OVP issue at all, just termination accuracy, which doesn’t get much airtime in IC selection discussions but absolutely should appear in your standard-mapping table alongside the OCP and OVP columns.
Thermal cutoff interaction with charge termination caught us about 18 months into a 72-unit field deployment of man-portable lithium packs for a generator-start application in high-altitude desert environments. The NTC we’d spec’d was accurate at the calibration temperature but the resistance-temperature curve drift at sustained 45°C ambient pushed the apparent temperature reading low enough that the IC was allowing top-of-charge cycles the IEC 62619 cutoff window would have blocked if the thermistor were reading correctly. Didn’t show up in bench testing because nobody ran the NTC at prolonged thermal soak before characterizing the charge curve.
The standard-stack hierarchy the article lays out is exactly right, and the IEC 62368-1 datasheet checkbox trap is real — we lost about 11 weeks on a 1.2kWh portable industrial unit because the IC vendor’s compliance note referenced 62368-1 system-level testing and our compliance team didn’t catch that 62619:2022 pack-level OCP validation was still completely open.
One thing we haven’t seen discussed here is how SMBus/I2C address conflicts play out when you’re stacking multiple BMS ICs in a multi-string telecom cabinet — we had a 48V/200Ah backup system where two ISL94216 front-ends shared a bus and the integrator hadn’t implemented the ADDR pin strapping correctly, so both devices were ACKing on 0x28 and the host firmware was silently reading stale coulomb-count registers from whichever device drove the bus last. The pack sailed through IEC 62619:2022 bench testing because it was single-string during cert, and the address collision only manifested in the 4-string field configuration three months post-deployment.