How to Read a UN 38.3 Test Summary Before Accepting a Battery Pack

The UN 38.3 test summary covers the pack on your dock only if four things match: the exact cell model, the pack model designation, the BMS revision, and the mechanical design of the enclosure. Beyond that, the page has to trace to an accredited test report by report number and date rather than to a certificate of compliance, and it has to list all eight test items from T1 through T8, including the ones marked not applicable. Miss any of those and the document is stale, no matter how good the letterhead looks.

I read these at the receiving table, not at a desk. Carton open on the pallet, pack still sitting in its foam, packing list on one side, laptop balanced on the other with the PDF open. Two strings of characters have to be identical: the model designation on the summary and the model designation printed or laser-etched on the pack. Not similar. Identical. I have watched a shipment sit for three days because one string ended in -2 and the other one didn't, and nobody could tell me which one was the real SKU.

Here is the order I go in.

The description block comes before the results

Most people jump straight to the pass column. That's the part that means the least on its own. The description block, the paragraph that names the cell, the battery, the nominal voltage, the watt-hour rating, and the physical construction, is what ties the results to a physical object. If it says "lithium-ion battery pack, 48 V, 20 Ah" and stops there, you have no idea whether the thing in the box is the thing that was tested.

What I want is the same specificity the supplier uses on their own drawing: cell manufacturer and cell model, pack model designation with revision, nominal voltage, rated capacity, and enough mechanical description to identify the enclosure. If the pack has a metal housing on the drawing and the summary describes a plastic one, that's not a typo. That's a different assembly.

Cells and packs are two separate claims

This is where a lot of summaries fall apart. The cell was tested. The pack is an assembly of tested cells plus a BMS, a housing, a harness, and whatever thermal interface material the factory is using that month. The manual treats a battery as its own article, so a cell-level summary does not cover your pack.

When a supplier forwards a page that describes a 21700 cell and nothing else, I ask one plain question: where is the battery-level summary? Sometimes it exists and nobody sent it. Sometimes it doesn't exist, and the supplier has been shipping for a year on the cell report alone. Both answers are useful. Only one of them is a problem.

The eight lines, and what a blank is hiding

The summary should list each of the eight test items: T1 altitude simulation, T2 thermal test, T3 vibration, T4 shock, T5 external short circuit, T6 impact or crush, T7 overcharge, T8 forced discharge. Not every item applies to both cells and batteries, so "not applicable" is a legitimate entry.

Blank is not. Blank means the person assembling the summary skipped a row. A summary that lists six tests and then jumps to a signature line tells me the document was built to satisfy a shipping clerk, not to survive a review.

I look at how results are recorded, too. "Pass" with nothing behind it is thin. A summary that references the report sections, or gives measured values, tells me somebody had the actual report open while they wrote it.

Lab, report number, date: follow the thread

Three fields decide whether the page is a document or a decoration. The test laboratory's name and address. The unique test report identification number. The date of the report.

The point of those fields is that they let you walk back to the report, so walk back. Ask for the report itself, under NDA if that's what it takes. A redacted report that still shows the test setup, the sample description, and the results is worth more than a clean summary with nothing behind it.

One thing buyers get wrong here: the summary is not issued by the lab. The lab issues the test report. The manufacturer writes the summary that condenses it. That means the summary is a manufacturer's declaration, and it inherits whatever the manufacturer chose to include. It is not a neutral document. It's a claim about a report, and the report is the evidence.

That's also why a certificate of compliance is not a substitute. A certificate says someone tested something and it passed. A summary says which report, from which lab, on what date, covering which item, with which result. One is a statement. The other is a trail.

Check the edition of the Manual of Tests and Criteria the testing was performed against as well. That field tells you how old the underlying work is, which matters more than the date on the summary page.

What changed since the ink dried

This is the mechanism that puts most stale summaries into circulation, and it rarely happens on purpose. A supplier swaps to a second-source cell because of allocation. The BMS vendor ships a firmware update that moves the overcharge cutoff. The enclosure goes from aluminum extrusion to stamped steel to save weight. The connector changes. Someone adds potting compound to pass an ingress test.

Any of those can invalidate the summary you're holding, because the summary describes a configuration, not a product family.

So when a supplier tells me the pack is "the same," I ask for it in writing against the specific change. What I want is one of two things: a new summary that reflects the current configuration, or a signed statement that identifies the change, names the report number, and says the existing testing still covers it, with the reasoning. Verbal reassurance doesn't survive a personnel change, and it doesn't survive a carrier asking questions at the airport.

If you're ordering a second batch, put the new summary next to the one in your file from batch one. Same report number, same date, same description, same revision. If any of those moved, ask why before you sign the receiving paperwork.

Where the summary stops

A UN 38.3 summary is a transport document. It exists so the pack can legally move, by air, by ground, by ocean, and it says the pack survived a defined set of abuse conditions once. It says nothing about how the pack behaves across 4,000 charge cycles, whether the BMS disconnects correctly at end of discharge, or whether the enclosure survives a drop onto a warehouse floor.

That's a different stack of paperwork. UL 2054 or UL 2271, depending on how the pack is classified and where it's deployed, plus whatever your own safety review requires. I have watched a program stall because everyone treated the UN summary as the safety certification and nobody had ever asked for the UL work. They are not interchangeable. One gets the battery onto the airplane. The other gets it into the application.

Keep the paper

The last thing I do is boring. I put the summary, the report reference, the date I received it, and the email where the supplier confirmed the configuration into the same folder as the PO. Two years later, when a pack comes back from the field with a swollen cell, that folder is the difference between a conversation about root cause and a conversation about who agreed to what.

If the summary on the dock doesn't match the pack, hold the shipment and send one email. Ask for the current summary, the report number behind it, and a written statement covering any change since the report date. It takes a day. It saves the kind of week you don't want to have.