Technical Evaluation & Sample Request Guide for Lithium-Ion vs LFP Chemistry

What to Lock Down Before You Contact a Supplier #

Most engineers reach out to a Shenzhen or Dongguan cell supplier with something like “we need 100Ah LFP prismatic samples.” That’s not a spec — it’s a shopping request. Before any inquiry goes out, the following parameters must be defined internally and stated explicitly in your RFQ document.

For LFP prismatic cells: nominal voltage (3.2V), rated capacity at the discharge rate your application actually uses (not the factory’s preferred 1/3C), maximum continuous discharge current, and cycle life target with the specific test conditions (C-rate, temperature, depth of discharge, retention threshold). For NMC cylindrical or pouch: add upper cutoff voltage explicitly, because factories default to 4.2V and some 811 chemistries need 4.25V — but will degrade significantly faster if held there without proper BMS cutoffs.

If you skip the discharge rate spec, you’ll receive datasheet capacity numbers tested at 0.2C or 1/3C, which tell you very little about performance at 1C or 2C. That gap matters. A 100Ah cell from a second-tier Shenzhen pack house may deliver 97Ah at 0.2C but only 88Ah at 1C — a difference that cascades directly into runtime calculations for your product.

Include operating temperature range requirements in the inquiry, not just nominal. A cell that performs well at 25°C can show 20-25% capacity reduction at -10°C, which is a distribution risk if your product ships to northern European or North American markets. BMS engineering considerations for low-temperature performance are tied directly to which chemistry you’re evaluating, so coordinate both workstreams in parallel.

Comparison: LFP vs NMC Sample Evaluation Parameters #

The criteria for requesting and testing samples differ meaningfully between LFP and NMC chemistries. This table captures what we specify in our internal EVL-03 Sample Request form before contacting any new supplier.

For most portable power station applications — 500Wh to 5kWh, cycling daily — LFP wins on longevity and thermal margin. We’d specify LFP for any product that cycles more than 300 times per year or operates in an uncontrolled thermal environment. NMC makes sense where energy density is genuinely constrained by physical volume, and the application sees fewer than 200 cycles per year with consistent thermal management. Don’t let a supplier convince you to use NMC 811 in a budget portable product just because they have excess stock — the BMS requirements to handle it safely cost more than the cell savings.

The Parameter That Changes the Evaluation Outcome #

Standard comparisons between LFP and NMC cover voltage, energy density, and nominal cycle counts. The factor that actually shifts sourcing decisions after sample testing is lot-to-lot impedance consistency, and almost no one measures it systematically during the first sample round.

Here’s why this matters operationally. You test 12 cells from Lot A, everything passes. You approve the supplier and run a 500-unit pilot. By the time those cells go through your pack assembly process, 8-12% show internal resistance values 15-20% higher than your approved sample baseline. That’s not a defective cell in the conventional sense — it still passes capacity check. But in a 4S or 8S configuration, the high-impedance outlier generates more heat under load, ages faster, and triggers premature BMS protection trips within 6-9 months.

We’ve tracked this specifically across 18 incoming lots from four Dongguan-area LFP suppliers over a 14-month period. The worst performer showed an impedance standard deviation of 0.047 mΩ across a 50-cell sample at goods receipt — more than triple what the Grade-A specification promises. The best performer held ±0.008 mΩ across comparable sample sizes.

The practical implication for your sample request process: ask for cells from three different production lots, not three cells from one lot. State this explicitly in your inquiry. A supplier who refuses or can’t accommodate it is telling you something important about their manufacturing control. Cell technology selection guides cover grading methodology, but lot consistency is separate from grade — Grade-A can still show poor lot-to-lot control at a small manufacturer.

One scenario worth flagging from the NMC side: a 2023 shipment of 21700 NMC cells from a Shenzhen-area supplier showed first-lot capacity of 4,847 mAh (rated 5,000 mAh), which cleared our 3% tolerance threshold. Second lot, same supplier, same part number: 4,612 mAh — an 8.9% drop that failed our incoming threshold and would have caused a firmware-calculated runtime error of approximately 11 minutes in a target product. No process change was disclosed. The supplier blamed “raw material variation.” We closed the AVL entry after that.

What to Do With Samples After They Arrive #

Incoming sample evaluation has a defined sequence. Skipping steps or running them in the wrong order wastes time when a cell fails early and you can’t determine which failure mode triggered it.

Step one: dimensional and weight check before any electrical testing. For LFP prismatic 280Ah cells, allowable dimensional variance from the datasheet is ±0.5mm on length and width, ±0.3mm on height. Weight variance beyond ±2% on a fresh cell is worth flagging — it may indicate inconsistent active material loading, which will show up in cycle testing but isn’t obvious from a single capacity measurement.

Step two: open circuit voltage check and rest period. Cells should arrive at 30-50% SOC for transport compliance under UN 38.3 Section 38.3.2.3. Check OCV immediately on receipt, then again after 24 hours at rest. A cell showing OCV drop of more than 5mV in 24 hours at room temperature has a self-discharge problem.

Step three: capacity verification at your specified discharge rate. Not the factory’s rate. Run a full charge-discharge cycle at 0.5C with the voltage limits specified in your BMS configuration — not the factory’s default test limits. Record actual delivered capacity, then calculate deviation against rated capacity.

Step four: AC impedance measurement at 1kHz, 25°C, 50% SOC. This is the lot consistency benchmark. Run all cells in the sample set and calculate standard deviation, not just mean.

Step five: brief cycle screening — minimum 10 cycles at 1C/1C before committing to longer formation testing. You’re not running a full lifetime test here; you’re screening for early anomalies that indicate manufacturing defects. A cell that shows more than 1.5% capacity deviation between cycle 3 and cycle 10 needs investigation per IEC 62660-1 cycle testing methodology.

The whole incoming evaluation sequence takes 12-18 days if run consecutively. If you’re on a tight design-in schedule, run dimensional/OCV checks and impedance in parallel on arrival day, and start capacity cycling immediately after the 24-hour rest. You can compress the timeline to 8-10 days without losing the critical data.

One brief list of common first-shipment red flags worth watching:
– Cells with OCV above 3.45V for LFP on arrival (suggests improper formation or age)
– Capacity test run by supplier at 0.2C presented as “rated capacity at standard conditions”
– UN38.3 report with a test completion date more than 18 months prior to your sample shipment
– Swollen or visibly deformed pouch cells — acceptable deformation tolerance for pouch NMC is less than 0.3mm expansion from nominal thickness at 50% SOC

After evaluation passes, set a 90-day milestone for the design-in decision. If you haven’t completed at least 50 cycles of qualification data within 90 days of receiving samples, the evaluation should be paused and the supplier status marked inactive in your AVL.

For the transition from sample approval to production supply agreement, the minimum documentation package should include: the signed cell specification sheet with your agreed test limits (not the factory’s generic datasheet), lot traceability records for the qualification samples, the full UN38.3 report with cell serial numbers matching your evaluated configuration, and a written process change notification agreement. The last item is non-negotiable. A supplier who won’t commit in writing to notify you before changing anode, cathode, or electrolyte formulation is a supplier you cannot qualify under IEC 62619:2022 clause 5.4.1 or manage safely under any systematic safety certification framework.

Sourcing Guidance for Buyers #

When evaluating Chinese suppliers in this category, the first document to request is the UN38.3 test report — and ask specifically for the internal cell serial number range tested, not just the cell model name. If the supplier sends a report where the tested configuration (cell count, arrangement, voltage) doesn’t match what you’re buying, that report is not applicable to your product. A supplier who resists this request, or sends a report with blank serial number fields, typically doesn’t have an original test — they have a borrowed or shared certificate.

A qualification red flag specific to LFP/NMC cells: if the supplier’s capacity specification changes between the quotation stage and the sample datasheet without explanation, it’s a signal of inconsistent production grading. A 100Ah cell that becomes a “nominal 98Ah” cell by the time samples ship indicates the supplier is selling off-grade stock into your qualification pipeline.

For incoming inspection sampling, we use a minimum AQL 1.0 Level II plan on dimensional and OCV checks, which for a 500-unit production lot translates to a sample size of 50 cells. Capacity verification is done on a 10% sample with a ±3% tolerance against the agreed rated capacity at 0.5C. Any lot showing more than 2 non-conforming cells out of the capacity sample is quarantined and returned, regardless of whether the mean capacity is acceptable.

What specific capacity deviation threshold should I specify in an LFP sample request inquiry?
Specify ±3% of rated capacity at 0.5C discharge rate. That’s the threshold we use in incoming inspection for Grade-A cells. At 0.2C you can tighten it to ±2%, but that discharge rate doesn’t reflect real application loads for most portable energy systems.

How many samples should I request from a new cell supplier?
For an initial evaluation, 20 cells minimum — but request them from at least two separate production lots if the supplier has them. A single-lot sample of 5 or 10 cells gives you no visibility into manufacturing consistency, which is often the actual failure mode, not the cells themselves.

Is a UN38.3 test report sufficient to approve a new cell supplier?
No. UN38.3 covers transport safety, not application performance or cycle life. It’s a necessary compliance document, not a performance qualification. Treat it as the floor, not the ceiling. You still need capacity, impedance, and cycle data specific to your operating conditions.

Does chemistry choice — LFP vs NMC — change the evaluation timeline significantly?
It depends on your cycle life target. LFP qualification at 2,000-cycle targets requires a longer screening period by definition; you can compress it with accelerated testing at 45°C, but that introduces its own interpretation challenges. For NMC at 800-cycle targets, the screening timeline is shorter but thermal abuse testing becomes more important. The total elapsed time from inquiry to design-in decision is roughly similar — 10 to 14 weeks — if you run the process correctly.

What if a supplier refuses to provide cells from multiple production lots for evaluation?
That’s a meaningful signal. Suppliers with mature manufacturing processes and lot traceability systems can pull multi-lot samples without difficulty. A refusal usually means either the supplier lacks lot traceability (a process maturity problem) or they know there’s variance between lots they’d rather you not see before purchase. Neither is a supplier you want anchoring your production supply chain.

---

Published by compactbess.com Technical Team | Request a sourcing consultation