LIFEPO4 CHARGER BOARD BASED ON CN3058E

A 3.6 V set point is not a de-rated 4.2 V set point

Both chemistries use the same four-stage profile: precharge, constant current, constant voltage, terminate on taper. Texas Instruments puts the CV point at 3.9–4.2 V for Li-ion against 3.5–3.65 V for LiFePO4, with precharge thresholds near 3.0 V and 2.0 V respectively (LiFePO4 Design Considerations, SLUAAR1). A 26650 LiFePO4 cell specification agrees from the cell side: charge at constant current to 3.65 ±0.05 V, hold 3.65 V, and — the sentence that matters — charge voltage “should not be above 3.85 V/cell to avoid over-charging”.

The gap between a 4.2 V charger and a 3.65 V cell is 0.55 V. A 4.2 V CV source holds a LiFePO4 cell at 4.2 V while the input is present, and the taper logic sees nothing wrong: current falls to the termination threshold and the charger reports success half a volt over spec.

The reverse pairing fails quietly. A 3.6 V charger on a Li-ion cell stops where a LiFePO4 pack would be full; the cell never reaches its own CV point and the cycle ends short of capacity. Worse, the CN3058E fast-charges anything above 2.5 V on the FB pin, so a Li-ion cell resting at 2.8 V — below the conditioning threshold its own charger would apply — gets full programmed current instead of a trickle.

What the CN3058E fixes and what it leaves to the board

It is a single-cell LiFePO4 linear charger with an on-chip P-channel pass MOSFET, so no blocking diode and no sense resistor are needed. Regulation is 3.6 V at 1.5%, specified as 3.55–3.65 V over −40 to 85 °C, so a worst-case part sits at the cell’s own 3.65 V ceiling and never above it. Input range 3.8–6 V, 6.5 V absolute maximum on any pin, current programmable to 1 A. Below 2.5 V on FB it precharges at a tenth of the fast-charge current; it terminates at a tenth of the CC value; it restarts when the cell falls 0.3 V below regulation. An internal loop pulls current back near 135 °C junction.

The external parts, one pin at a time

• ISET sets current by ICH = 1218 V / R_ISET: 1.218 kΩ is 1 A, 2.44 kΩ is 0.5 A, with 1% metal film. The pin sits at 1.205 V in CC mode and 0.12 V in precharge. Stray capacitance narrows the choice: with the ISET pole above 200 kHz, R_ISET < 1 / (6.28 × 2e5 × C) puts a 10 pF node at about 80 kΩ.
• BAT needs 4.7–10 µF to ground. The datasheet calls this the compensation capacitor for the voltage loop, not bulk storage.
• VIN takes 1 µF ceramic at the pins; with a long cable the datasheet recommends a series diode or a TVS against the hot-plug spike.
• FB is a Kelvin sense input drawing about 3 µA, meant to land on the cell terminal rather than on BAT, so wiring and pass-FET drop stay outside the loop.
• TEMP compares an NTC divider against windows at 45% and 80% of VIN and suspends charge outside them. Grounding the pin disables it — and removes the only temperature limit in the system.
• CHRG and DONE are open-drain, good for 10 mA; tie the unused one to ground.

An optional resistor from FB to BAT raises the set point by Vbat = 3.6 + 3.61e-6 × Rx: 14 kΩ gives 3.65 V, 166 kΩ reaches 4.2 V. The charger permits that; the cell specification does not.

Current, heat, and the linear charger’s ceiling

A linear charger dissipates the difference between input and cell: P = (VIN − VBAT) × I. On 5 V charging a 3.2 V cell at 1 A that is 1.8 W; on 6 V it is 2.8 W, because at 1 A every input volt is a watt.

The datasheet lists thermal resistance for the SOP8 as TBD, so junction temperature cannot be computed from it; only the 135 °C regulation point and the 150 °C maximum junction are given. That matters, because foldback reduces current silently — CHRG and DONE carry no hint that the loop is limiting, so a board that cannot spread 1.8 W just looks slow. The exposed pad must be soldered to ground copper with vias; the datasheet is explicit that skipping that contact raises thermal resistance. Even the current setting is nominal: the 1 A configuration is specified from 850 to 1150 mA.

What the bare cell still needs around it

This is a charger, not a protector. It has no over-discharge cutoff, no short-circuit protection across the cell terminals, and no over-voltage comparator independent of its own loop: if the pass FET fails short, VIN lands on the cell and nothing inside the package stops it. A protected cell, or a separate protector with back-to-back FETs, covers that. Temperature is conditional too — the cell specification forbids charging below 0 °C and allows only 0.1C from 0 to 5 °C, and the TEMP window enforces that only if an NTC is fitted and R1/R2 match the real thermistor.

Documented, and not

Everything above is documented in the references. Die temperature on your board, the current the layout delivers before foldback, the termination current at the cell and the window your divider produces have to be measured on the assembled unit.

References

• Consonance Electronic, CN3058E 1A LiFePO4 Battery Charger, datasheet REV 1.1.
• Texas Instruments, LiFePO4 Design Considerations, application note SLUAAR1, July 2023.
• Guangdong Superpack Technology, IFR26650EC-3.6Ah Lithium Iron Phosphate Battery Approval Sheet.
• Texas Instruments, Multi-Chemistry Charger in Low Battery Power Applications, application note SLUAAH5, November 2022.