TP4056

It is a linear-versus-switching and integration question. The TP4056 costs a few cents, needs four external parts and charges at up to 1 A, but wastes (VIN - VBAT) x I as heat and has no power path, so the system browns out when you pull the battery. The BQ24295 is an I2C-controlled 1.5 MHz synchronous buck charger with NVDC power path, up to 3 A charge current, BC1.2 adapter detection and OTG boost — the right choice for phone-class products at roughly twenty times the part cost. The IP5328P goes further, integrating a 5 A charger, an 18 W boost output, fuel gauge and fast-charge protocols for power banks.

Connect one resistor from PROG to ground; the IC regulates that pin to 2 V in constant-current mode, and the datasheet table gives 10 kΩ for 130 mA, 5 kΩ for 250 mA, 3 kΩ for 400 mA, 2 kΩ for 580 mA, 1.5 kΩ for 780 mA, 1.33 kΩ for 900 mA and 1.2 kΩ for 1 A. Never set more current than the cell's specification allows — most 18650 cells are rated 0.5C to 1C, so a 2,000 mAh cell at 1 A is fine but a 500 mAh LiPo is not. Termination happens when current falls to C/10, so a smaller PROG resistor also terminates sooner in absolute terms.

It is inherent to linear charging: at 5 V input and a 4.0 V cell charging at 1 A, the IC dissipates about 1 W, and at 4.2 V float it is still 0.8 W. The SOP-8 paddle and a ground pour spread it, but the case will feel hot to touch. Safe, yes, provided the die stays inside thermal regulation (which reduces current near 145 °C) and the pack temperature is monitored. If the enclosure is sealed or ambient is high, reduce the PROG current, add copper and airflow, or switch to a switching charger such as the BQ24295.

No to both. It is a single-cell device whose 4.2 V float is referenced to the BAT pin, so a 2S pack needs a dedicated charger such as the BQ2084 or a balancing topology. And it has no power-path management: the load is fed through the battery terminal, so with no pack connected the output sits at 4.2 V through the internal pass device and the status pins can behave oddly (the datasheet notes the green LED blinking when a capacitor is present but no battery). If you need system-plus-charge operation, use a power-path part.

For any product that ships to consumers, yes. Grounding TEMP disables temperature qualification, which is what most cheap modules do, and it means a pack can be charged at 0 °C — a condition that plates lithium and permanently damages cells — or above 45 °C, where thermal runaway risk rises. Wire a 10 kΩ NTC (103AT-2 class) from the pack to TEMP with the divider arranged so charging suspends below 45% or above 80% of VIN for more than 150 ms, and place the thermistor against the cell, not on the PCB.

Bench-test, because visual inspection alone is unreliable on such a cheap part. Charge a known cell with a 1.2 kΩ PROG resistor while logging current and voltage: genuine parts hold close to 1 A in constant-current mode, terminate at C/10 near 4.2 V, and float at 4.137 V to 4.263 V — anything above 4.263 V is a reject on safety grounds. Clones frequently terminate at 50% to 60% of the set current, drift high on float voltage, ignore the TEMP window or lack thermal regulation, which shows as a case temperature that keeps climbing. Marking checks (laser etch, matte leads, mould quality) filter out re-tinned pulls. We test every lot this way and provide the charge curves with shipment.

MOQ is 100 pcs loose for prototyping or one reel (2,500 pcs) of TP4056-42-SOP8-PP for production, with assembled charging modules available in packs of 10 or 100. Shenzhen stock ships in one to two business days; factory channel replenishment is two to four weeks, which is unusually short because NanJing TopPower and its clones are produced in very large volumes.