Power Banks & Portable Chargers — Technical Specification Overview

What the Spec Sheet Tells You vs. What the Product Actually Does #

Power bank spec sheets from Chinese manufacturers follow a predictable pattern: capacity in mWh or Ah at the cell nominal voltage (usually 3.6V or 3.7V for NMC, 3.2V for LFP), peak output current, and a port configuration list. What they almost never disclose: the discharge rate at which capacity was measured, the ambient temperature during testing, or the conversion efficiency of the boost circuit between the cell and the USB output.

That gap matters enormously at the procurement level. A 20,000mAh power bank using NMC cells rated at 3.7V holds 74Wh of cell-level energy. Factor in a boost converter efficiency of 91-93% (realistic for a mid-grade IC like the IP5328P or similar), and you’re delivering roughly 67-68Wh at the 5V output rail. At 1C discharge, real-world delivered energy often drops further — we’ve logged 63.4Wh on units claiming 74Wh, not because the cells were bad, but because the BMS cut off at 3.1V rather than 3.0V and the boost circuit was thermally derated at 2.4A continuous.

This is a calibration and firmware problem, not a cell problem. And it’s one that most outgoing QC processes at Chinese pack houses never catch, because they test at 0.2C into a resistive load — conditions that hide both the voltage sag and the thermal derating behavior.

The spec sheet tells you what a product could do under ideal lab conditions. Incoming inspection tells you what it does under load, which is the only number your end user will ever experience.

Symptom Identification — Reading Failures Back to Their Root Causes #

Three symptoms show up repeatedly in field returns and pre-shipment audits:

Symptom 1: Delivered capacity 15-25% below nameplate. The LED indicator shows full, the device charges for the expected number of cycles, but the phone (or test load) pulls less energy than the rated output suggests. This maps to three possible causes: (a) Grade-B or relabeled cells with actual capacity 10-15% below the stamped rating, (b) BMS cutoff voltage set too conservatively (3.1V vs. 3.0V on NMC, 2.8V vs. 2.5V on LFP), or (c) boost converter efficiency below spec due to component substitution on the PCB.

Symptom 2: Unit gets noticeably hot during output at 18W or above. Surface temperature above 42°C during PD 18W output is a red flag, not a normal operating condition. Root causes: undersized MOSFET on the boost stage, missing or undersized thermistor on the cell, or BMS with no output current derating curve. Shenzhen-based pack houses building to sub-$4 BOM targets routinely omit the NTC thermistor on the output stage.

Symptom 3: Cycle life degrades sharply after 200-300 cycles at 1C/1C rates. The unit was sold with a 500-cycle claim. After eight months of daily use, capacity retention is below 75%. This almost always traces back to cell grade — specifically, the difference between what the datasheet calls “Grade A” and what a qualified cell supplier actually ships under that designation.

The table below maps symptoms to likely causes and recommended diagnostic steps:

Root Cause Deep-Dive — BMS Firmware Calibration Is the Misdiagnosed Problem #

When a power bank underperforms, the first instinct is to blame the cells. That’s usually wrong. In our QC-07 incoming inspection procedure, we separate cell-level failures from system-level failures as the first triage step — and system-level failures, driven by BMS firmware, account for roughly two-thirds of the underperformance cases we’ve logged.

Here’s the mechanism. A power bank BMS uses a combination of coulomb counting and open-circuit voltage (OCV) lookup tables to estimate state of charge. The OCV table must be calibrated specifically to the cell chemistry and, critically, to the actual capacity of the cells installed. When a Dongguan-based BMS manufacturer ships a standard firmware build — which most of them do, because custom firmware costs time and requires cell characterization data — that firmware contains a generic OCV table, often derived from a 2,600mAh NMC cylindrical cell even if the product uses 3,000mAh cells or a different form factor entirely.

The result is systematic SOC error. The BMS believes the pack is at 15% SOC when it’s actually at 8%. It cuts off output at what it thinks is a safe floor, leaving 200-400mAh of usable capacity on the table. Or worse — the opposite error, where the BMS allows discharge below the safe floor because the OCV curve is shifted, accelerating lithium plating and capacity fade.

This problem is invisible on a simple capacity test at 0.2C. You need to run a full characterization: charge to 4.20V at 0.5C, rest 30 minutes, discharge at 1C to 2.75V cutoff, log voltage and current at 10-second intervals, and compare the measured Ah against the BMS-reported SOC at each point. A well-calibrated BMS should track within ±4% across the full SOC range from 20% to 90%. We reject any BMS board where SOC error exceeds 7% at the 20% SOC point — that’s the failure mode that causes end-user complaints about sudden shutdowns.

The confirmation threshold for BMS firmware miscalibration: SOC error > 6% at any point between 15% and 85% SOC, measured during a 1C discharge with a 10-second logging interval.

Fixing this after production means reflashing firmware, which requires the BMS manufacturer to provide a programmer interface and cooperate on cell characterization — a negotiation most buyers can’t win post-delivery. Specifying it upfront, as a requirement in the purchase order, is the only reliable path.

Corrective Actions Ranked by Impact and Feasibility #

1. Require cell-level capacity verification before pack assembly. Ask for incoming inspection records showing actual measured capacity (mAh at 0.5C, 25°C) on each cell lot. This costs the factory roughly $0.08-0.12 per cell in testing time and adds 24-48 hours to production lead time. It eliminates the Grade-B cell substitution risk and catches mislabeled capacity upfront. This fixes the root cause in cases where underperformance is cell-driven.

2. Specify BMS SOC calibration requirements in writing. Include a clause requiring SOC error ≤ 5% from 20-90% SOC, verified by the test protocol described above. For BMS engineering guidance specific to portable applications, this threshold is achievable with any BMS using dedicated cell characterization firmware, but it requires the supplier to run a cell-specific calibration cycle before flashing production units. Few factories do this unless contractually required.

3. Add a 15W sustained output burn-in test to incoming inspection. Run each sample unit at 15W output for 20 minutes and log surface temperature. Reject any unit exceeding 44°C on the cell surface or 47°C on the boost converter area. This catches the thermal derating and MOSFET sizing failures before they reach your customer.

4. Audit the BOM for component substitution. Request the approved vendor list for the boost IC, the MOSFET, and the protection IC. Cross-reference against what’s actually installed on the PCB. Component substitution mid-production is common at factories under cost pressure, and a substituted MOSFET with a higher on-resistance will cause exactly the thermal and efficiency failures described above. This requires board-level inspection and is labor-intensive, but for high-volume orders it’s worth it.

5. Switch to an LFP-based design if cycle life is the priority spec. NMC power banks from mid-tier Shenzhen factories typically deliver 400-500 actual cycles at 80% retention under 1C/1C conditions. LFP equivalents, even from the same factories, routinely hit 1,800-2,200 cycles under the same conditions. The weight penalty is real — LFP energy density is roughly 200-250 Wh/kg vs. 250-300 Wh/kg for NMC — but for professional tools, field equipment, or any application with a 3+ year product life expectation, the trade-off favors LFP.

Specification Comparison — Three Sourcing Tiers Across Key Parameters #

The table below reflects values from qualified supplier audits and incoming inspection logs, not manufacturer claims. “Mass market” refers to standard consumer-grade OEM, “mid-tier OEM” to products with traceable cell sourcing and basic BMS customization, and “spec-grade” to products with full cell characterization, custom BMS firmware, and UN38.3 test reports tied to actual serial numbers.

The price gap between mass market and spec-grade is $5-8 per unit at 20,000mAh. For a consumer brand selling at $25 MSRP, that’s a significant BOM percentage. For a B2B buyer supplying field technicians or medical personnel where a sudden shutdown causes operational failure, it’s a rounding error on the cost of one incident.

Prevention — What to Specify Before the PO Is Signed #

Put the following in the technical specification attached to your purchase order: minimum delivered capacity at 1C discharge (not at 0.2C), BMS SOC accuracy requirement (±5% from 20-90% SOC), cell-level test record format, and maximum surface temperature at rated output. Also specify that the safety certification documentation must reference the exact cell lot numbers and BMS firmware version used in the production batch, not a generic certificate from a prior configuration.

For UN38.3 compliance, the UN38.3 Transport Testing Requirements specify testing on production-representative samples — not prototype configurations. Request the actual test report, not just the certificate. The document to request: a signed CoC (Certificate of Conformance) with cell manufacturer, cell model, cell lot number, BMS firmware version, and pack configuration, all matching the production sample.

Sourcing Guidance for Buyers #

When evaluating Chinese suppliers in this category, the first document to request is not the CE certificate — it’s the cell incoming inspection record. A factory that tracks actual measured cell capacity by lot number has a QC process worth engaging. A factory that can’t produce this record is buying cells on price and trusting the stamp on the wrapper.

The qualification red flag specific to power banks and portable chargers: a factory that offers BMS customization but can’t produce a cell characterization curve for your specific cell. Customizing protection thresholds without cell-specific OCV data is guesswork, and guesswork in BMS firmware shows up as SOC errors and premature cutoffs in the field.

For incoming inspection, test a minimum sample of 8 units per 500-unit lot at 1C discharge into a calibrated constant-current load. Log delivered capacity, peak surface temperature, and SOC accuracy at the 20% and 50% points. Reject the lot if more than 2 of 8 units fall outside the contracted capacity tolerance of ±4%. This sample size follows a simplified version of IEC 62133-2:2017 acceptance criteria, which governs secondary lithium cell safety for portable applications.

Also cross-reference any PD or fast-charge claims against USB Power Delivery specification r3.1 — many Shenzhen factories implement a proprietary fast-charge protocol and label it “PD compatible” when it only triggers on their own cable/adapter. That’s a compliance gap that matters in regulated markets.

For a broader view of how cell selection affects system-level performance in compact storage products, the cell technology documentation covers grade classification and cycle life testing methodology in more depth.

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

---