TL;DR: UN38.3 edition upgrades (Rev 6 → Rev 7) invalidate prior test reports if your cell configuration changed — retesting isn’t optional, it’s a logistics hold risk at customs.
TL;DR: Switching from a single-cell UN38.3 report to a multi-cell pack-level report adds an average of 47 additional test cycles across T.1–T.8, with T.3 (altitude simulation) and T.5 (external short circuit) being the most commonly failed in pack-level retests.
UN38.3 Test Scope Comparison: Single Cell vs. Pack-Level vs. Section II Small Format #
The test scope you need under UN38.3 (ST/SG/AC.10/11/Rev.7) is not fixed — it shifts based on the form factor you’re shipping, the Wh rating of the individual cell, and whether the cells are shipped installed in equipment. Most of the confusion we see during supplier qualification reviews comes from buyers treating UN38.3 as a binary pass/fail stamp, when the actual test matrix is tiered and configuration-dependent.
The table below compares the three most common certification tracks we encounter when sourcing portable power stations and battery packs from Chinese manufacturers:
| Parameter | Single Cell (bare, ≤20Wh) | Battery Pack (≤100Wh) | Section II Small Format (≤2.7Wh) |
|---|---|---|---|
| Tests required (T.1–T.8) | T.1–T.8, full sequence | T.1–T.8, full sequence | T.1, T.2, T.3, T.4, T.5, T.8 (T.6/T.7 conditional) |
| Altitude simulation (T.3) | 11.6 kPa, 6 hours | 11.6 kPa, 6 hours | 11.6 kPa, 6 hours |
| Vibration (T.4) | Sinusoidal, 7–200 Hz sweep | Sinusoidal, 7–200 Hz sweep | Sinusoidal, 7–200 Hz sweep |
| Shock (T.5 short circuit surrogate) | 0.5Ω external short, 60 min or until stable | 0.5Ω external short, 60 min or until stable | 0.1Ω max, per cell Wh |
| Typical lab turnaround (Shenzhen-area labs) | 18–22 business days | 28–35 business days | 14–17 business days |
| Report validity on configuration change | Invalidated if cell Wh or series count changes | Invalidated on any topology change | Invalidated on cell supplier change |
The turnaround delta between a pack-level report and a Section II small-format report — 14 days on average, based on our 2024 lab submissions through three Shenzhen-area CNAS-accredited labs — has a direct effect on product launch timelines. Buyers planning seasonal inventory builds who discover mid-process that their pack topology change requires a full T.1–T.8 retest have, in several cases we’ve tracked, missed Q4 shipping windows entirely.
I’d prioritize the Section II pathway wherever your cell Wh genuinely qualifies — but do not try to force qualification there by splitting packs into sub-assemblies that will be reassembled post-shipment. That’s a customs classification problem waiting to happen, and IATA’s DGR table 966-970 auditors have seen every variant of this.
What Actually Fails in Pack-Level Retests — and Why Edition Upgrades Trigger It #
This is the longest section for a reason: configuration drift between test report and production is the most expensive certification problem we see, and it happens in ways that aren’t obvious until you’re holding a rejected shipment.
T.5 external short circuit failures after cell supplier changes. A portable power station manufacturer in Guangdong held a valid pack-level UN38.3 report from 2022 covering a 4S2P LFP configuration using cells from a specific Shenzhen cell supplier. When that supplier’s Grade-A allocation tightened in mid-2023, the pack house quietly substituted cells from a different source with nominally identical specs: same 3.2V nominal, same 100Ah rating. The UN38.3 report wasn’t updated. At T.5 external short testing, the substitute cells showed a peak temperature of 89°C versus the originally reported 61°C — both below the 170°C threshold, technically, but the delta triggered a re-evaluation flag at the forwarding agent’s compliance desk. The shipment cleared eventually, but the 23-day delay cost more than the retest would have. The check here is simple: compare the cell supplier name and part number on your UN38.3 report against your current production BOM. If they don’t match exactly, you’re shipping on a report that may not represent your current product.
Rev 6 to Rev 7 edition gaps on altitude simulation. UN38.3 Rev 7, published in 2021, tightened the documentation requirements around T.1 altitude simulation — specifically the requirement to record ambient pressure at test termination, not just at initiation. Labs that ran tests under Rev 6 procedures often didn’t capture this data point. When airlines upgraded their dangerous goods acceptance checklists to explicitly reference Rev 7 section 38.3.4.4 compliance, reports from pre-2021 tests started getting rejected at cargo acceptance. We log these under what our team calls the “Rev-gap incident category” in our supplier report tracker — we saw 11 such rejections across our client base between Q1 2023 and Q2 2024. The fix isn’t a lab retest: it’s confirming your lab can issue a Rev 7-compliant supplemental summary page. Several Shenzhen labs can do this for reports they originally issued; labs that can’t are a signal about their documentation infrastructure generally.
BMS topology changes and T.6/T.7 conditional triggers. UN38.3 T.6 (impact) and T.7 (overcharge) are conditional — they apply to lithium-ion but not lithium metal, and T.7 overcharge applies when the battery has no BMS charge control. The problem we see with upgraded battery packs that remove or replace their BMS is that the conditional logic flips. A pack originally tested with a BMS in place (T.7 exempted) gets a BMS revision that the factory characterizes as a “firmware update” rather than a hardware change. If the new firmware changes the charge cutoff voltage — even by 50mV — the question of whether T.7 needs to be re-run becomes a genuine compliance question under IEC 62281:2019 clause 5.3, which harmonizes with UN38.3 for transport of lithium cells. Auditors reading IEC 62281 alongside UN38.3 will catch this. Labs that rubber-stamp unchanged reports when BMS firmware has been revised are creating liability for the buyer, not the factory.
The pattern across all three scenarios is the same: configuration changes that look minor at the component level have test-matrix implications that the factory’s sales team is not equipped to evaluate. Treat your UN38.3 report as a document that is coupled to a specific, frozen BOM state.
Does Upgrading from Rev 6 to Rev 7 Always Require a Full Retest? #
No — and the nuance here matters practically.
A Rev 7 upgrade does not automatically require repeating all eight tests. If your existing lab report covers all required test items under Rev 7’s scope for your product category, the lab may issue a Rev 7-compliant summary or attestation document without rerunning physical tests, provided the test data already satisfies the updated criteria. Where retesting is required is when either: (a) the lab’s original data didn’t capture the additional measurement points Rev 7 mandates, or (b) your product configuration changed coincidentally with the revision upgrade. For pack-level products with IEC 62619 secondary certification, the documentation trail is usually sufficient to issue an upgraded report. For bare-cell-only reports, the data gaps from pre-2021 tests are more common.
Sourcing Guidance for Buyers #
When evaluating Chinese suppliers in this category, the first document to request isn’t the UN38.3 test report itself — it’s the test summary page that maps each test item (T.1–T.8) to the specific cell part number and pack configuration tested. Its absence, or a summary page that lists only “lithium battery pack” without model-specific detail, signals that the factory is likely using a shared or borrowed report. We’ve seen this pattern most commonly among Dongguan-area pack assemblers who use contract labs for certification but don’t maintain configuration-controlled documentation between product revisions.
The qualification red flag specific to this category: any factory that cannot tell you which edition of UN38.3 their report was tested under, and whether their lab is CNAS-accredited for dangerous goods testing. Edition awareness is a basic competency. Factories that don’t track it are not managing their compliance infrastructure.
For incoming inspection, request the original lab test report (not a certificate-of-compliance summary) and cross-reference the cell part number against your production BOM on a per-lot basis. For shipments over 200 units, pull a 5-unit sample for dimensional verification of the cell and confirm it matches the cell described in the UN38.3 report. This step, which we call our QC-12 report cross-check procedure, has caught configuration drift on three separate occasions in the past 18 months — each time at a factory with otherwise clean audit records.
For buyers sourcing complete portable power station products, the Compact BESS Products category covers product-level certification stacking, and our BMS Engineering resources address the BMS configuration changes that most commonly trigger UN38.3 retesting requirements.
Frequently Asked Questions #
If I change the cell capacity from 100Ah to 120Ah but keep the same chemistry and series count, do I need a new UN38.3 report?
Yes. A capacity change of that magnitude changes the Wh rating of the pack, which directly affects the test parameters — particularly the energy thresholds governing T.5 and T.7 pass criteria. The report tied to your 100Ah configuration does not cover the 120Ah variant.
Can one UN38.3 report cover multiple shipping configurations — for example, cells shipped both loose and installed in equipment?
It depends on how the original test scope was defined. UN38.3 allows a single report to cover multiple configurations if all configurations were explicitly tested and documented in the original test program. In practice, most Shenzhen-area labs issue configuration-specific reports because it simplifies the test matrix and reduces lab time. If you need coverage for both PI 965 Section IA (bare cells) and PI 967 Section I (cells in equipment), verify with your lab before testing begins whether a combined scope is offered — retro-fitting a second configuration onto an existing report is not supported under the standard’s documentation framework.
Is there a grace period for shipping on a Rev 6 report after Rev 7 came into force?
Formally, no. IATA’s Dangerous Goods Regulations reference the current applicable edition of UN38.3, and carriers updated their DGR acceptance checklists following the 2021 Rev 7 publication. Practically, acceptance varies by carrier and origin port — some Shenzhen freight forwarders continued accepting Rev 6 reports through mid-2022 without issue, while others flagged them immediately. That inconsistency does not represent a grace period; it represents inconsistent enforcement. Building a shipping strategy around inconsistent enforcement is a risk that customs delays will eventually correct.
Published by compactbess.com Technical Team | Request a sourcing consultation
