Your Pack Assembler Warranty Does Not Automatically Cover Cell Defects

A pack assembler's warranty covers the assembler's work — the welds, the busbars, the enclosure, the harness, the crimps, and the pack-level functional test the unit passed before it shipped. It does not, on its own, cover the cells inside. Cell defects belong to the cell manufacturer's warranty, which runs to the assembler as the cell purchaser, not to you. A cell-lot failure arriving at month fourteen has to travel through the assembler to reach the party who actually owes the coverage, and it only travels if two things are true: the pack serial number can be traced to a specific cell lot, and the assembler can still recover from the cell maker under terms that were quoted back to back before the purchase order was signed. Both of those are procurement decisions. Both get made months before anyone knows a lot is bad.

I've watched that chain break on a conference call. A 48V pack came back from an AMR fleet at month fourteen with one swollen module, the cell was down, and we were counting lost production in hours. The assembler's account manager was on the line and said the right thing: "We stand behind our packs." I said fine, then send me the cell manufacturer's warranty. He said he'd have to check with his supplier. That pause — the one where a warranty administrator goes looking for a document — is the sound of a pass-through that does not exist. He wasn't lying to me. He just had nothing to pass through.

What I had on my desk afterward was two pieces of paper, and they described two different worlds. The pack warranty ran three pages, used the word workmanship eleven times, and handled cells in a single sentence that incorporated the cell manufacturer's warranty by reference, a copy of which was available upon request. The other piece of paper was a two-column table someone had pasted into a quote: duration, capacity threshold. That table is not a warranty. It is a summary of a warranty written by the party who benefits from you not reading the original.

What the assembler is actually warranting

Workmanship coverage is a real thing and it is worth having. It covers whether the pack was built correctly — weld integrity pulled and tested, fastener torque, enclosure seal, connector crimp, the harness dressed so it doesn't chafe. Most of the failures you'd catch in the first ninety days live here.

What it usually does not cover is a cell that was fine when it left the line and degrades faster than it should, or a lot that develops an internal short at month eighteen. Some assemblers do absorb cell defects into a pack-level warranty and simply carry that risk themselves. That is a different product and it is priced like one. Before you compare quotes, find out which one you're being sold, because a two percent delta on a pack price can be the whole difference between an assembler who eats a bad lot and a buyer who eats it.

If the pack warranty says cell defects are handled under the cell manufacturer's warranty, you have a pass-through whether or not anyone used that phrase. The word matters less than the effect.

Get the cell maker's document, not the summary table

Ask for the actual cell manufacturer warranty, current revision, attached to the quote as a PDF. Not a paragraph in the proposal, not a table, not a link that expires. Read four things in it.

When the clock starts. This is the one that quietly costs buyers money. If the cell warranty begins at the cell manufacture date, and the cells sat in a distributor's warehouse for five months and then in the assembler's build queue for two more, you have burned a third of a three-year term before the robot moved. If it begins on pack ship date or on your commissioning date, that's better for you and worth asking for explicitly. Get the definition in writing either way.

What the remedy actually is. A great many cell warranties pay a pro-rata credit toward replacement cells, not a free pack. At month forty of a five-year term you may be looking at a credit covering a fraction of cell cost, with the assembler's labor and freight landing on your side of the ledger. That's not a scandal. It's just not what people hear when they hear the word warranty.

What triggers the capacity fade threshold, and how it's measured. If the threshold is stated but the measurement method is not — whose equipment, what temperature, what discharge rate, how many cycles — you'll be arguing about the test, not the pack.

And the exclusions. Which brings me to the next thing.

Traceability from pack serial to cell lot

The claim starts with a number on a label.

Your pack carries a serial. The assembler's build record should tie that serial to the cell lot code, the cell manufacturer, the cell model, and the date those cells were received. That linkage is what lets a claim move at all. Without it, you're holding a failed pack and a shrug.

Ask for this before the PO, and ask for it as evidence: show me a traceability record from a pack you built last year, serial to cell lot. If they can't produce one for an existing build, they will not produce one for yours when it matters. Programs are most exposed when the assembler buys cells opportunistically from distribution, when a build runs across two cell lots, or when the cell model changed mid-program without anyone writing it down. A twelve-module pack assembled from two lots turns a clean claim into a conversation about which modules are covered.

I have a standing request in my RFQs: name the cell manufacturer and the cell model in the agreement, and require written notice plus requalification of warranty terms before either changes. Allocation moves. Assemblers swap cells. If the substitute cell isn't qualified under the same duration, the same capacity threshold, and the same exclusions, your pass-through coverage changed and no one sent you a letter about it. When a change is proposed, ask for the new cell maker's warranty document. A note that says the new cell is equivalent is not a warranty.

The exclusions that decide the claim

Two exclusions do most of the work in real disputes.

The first is disassembly. Opening a pack typically voids cell-level coverage. Picture the failure: a swollen pack comes off a robot at 2 a.m., and the first competent person who touches it wants to know what's inside. If they open it, you may have converted a covered cell defect into a customer-caused event. So write into the pack agreement who is permitted to open a failed pack, under what procedure, and who authorizes it. That sentence costs nothing at negotiation and saves a claim later.

The second is operation outside the cell specification. That includes voltage limits, continuous discharge current, and temperature range. The uncomfortable part is that this exclusion is usually adjudicated with your own data. The BMS logs the duty cycle, the current, the temperatures. If your AMR routinely pulls more continuous current than the cell datasheet allows, or the pack runs hot in a summer aisle, the cell maker will find it, and the log is the exhibit. Keep your logs. Produce them on request with a claim number and a stated purpose, not into a general email thread.

What belongs in the RFQ

A short list, because this is the part anyone can copy:

• The cell manufacturer's warranty document, current revision, attached to the quote.
• A written statement of the beginning-of-warranty date and the total duration.
• A sample traceability record showing pack serial to cell lot to cell manufacturer.
• Named cell manufacturer and cell model, with a change-notice and requalification clause.
• For a cell-related failure: who diagnoses, who authorizes, who ships, who pays freight in both directions, what the turnaround commitment is, and what you get back — new pack, remanufactured pack, modules, or a credit.
• Whether the buyer's recovery runs against the assembler regardless of what the assembler collects, or only to the extent the assembler collects.

That last one is the whole article in a single line. If your claim runs against the assembler no matter what, you have a pack-level warranty covering cell defects and you should price it that way. If it only runs as far as the assembler recovers, you are sharing risk with a company whose leverage over its cell supplier may be thin — and whose willingness to fight for your claim at month thirty depends on whether you're still buying from them at month thirty.

I've had both work. I've also had a small assembler tell me honestly that they couldn't get pass-through terms because they didn't buy enough volume to ask, and we structured a higher pack price with cell defects carried at the pack level instead. That's a legitimate answer. What isn't legitimate is a proposal that implies coverage without a document behind it.

One more thing, and it's the unglamorous one. Keep the file. Pack serials, cell lot records, the warranty documents as originally issued — not as summarized in a final quote PDF — commissioning dates, freight paperwork, BMS logs. When the failure comes, whoever opens that folder at that moment decides whether you have a claim or a story. I have never once regretted fifteen minutes of filing at receipt. I have regretted not having it.