TP4056 Thermal Budget and Charge-Current Setting

One resistor sets the current. The input voltage, the battery voltage and the copper set the temperature — and the temperature decides whether the resistor value survives contact with reality.

1. What the resistor actually sets

The PROG pin is a current sink. The charger mirrors that current into the battery, so the relationship is a simple inverse:

Ibat (mA) = 1200 / Rprog (kOhm)

1.2 kOhm -> 1000 mA
2.4 kOhm ->  500 mA
10  kOhm ->  120 mA

The datasheet confirms this with two operating points: 2.4 kΩ gives 500 mA typical, 1.2 kΩ gives 1000 mA typical. The absolute maximum BAT pin current is 1200 mA, so do not go below 1 kΩ even though the formula would keep going.

Why "just fit a bigger resistor" is only half a fix

Lowering the current solves the heat problem but costs time, and the trade is not linear in the way people expect. Going from 1 A to 0.5 A halves the dissipation but doubles the CC phase. For a 2000 mAh pack that is the difference between a 2.7-hour charge and a 5.4-hour one — often unacceptable in a product that promises a full battery overnight.

The correct answer depends on which constraint binds. If the enclosure is the limit, cut current and accept the time. If the charge time is the limit, change the topology: a 2 A switching charger at 90 % efficiency puts 0.33 W into the board instead of 1.3 W, and finishes in half the time.

2. Where the heat comes from

A linear charger is a controlled resistor between the input and the battery. The voltage it drops is the difference between the two, and the current is the charge current:

P = (Vin - Vbat) x Ibat

5.0 V in, 3.7 V cell, 1.0 A:  P = 1.3 W
5.0 V in, 3.7 V cell, 0.5 A:  P = 0.65 W
6.0 V in, 3.7 V cell, 1.0 A:  P = 2.3 W

That last line is the one that surprises people. A 6 V source is only 1 V more than 5 V, but it adds 77 % to the dissipation at the same current. Solar panels, 6 V wall adapters and unregulated supplies all sit in that trap.

The battery voltage is not a constant

All of the tables above assume 3.7 V, which is the nominal voltage. In reality a discharged cell sits near 3.0 V and a full one at 4.2 V, so the dissipation changes during every charge:

5 V in, 1 A:
  Vbat = 3.0 V  ->  P = 2.0 W   (worst case, start of charge)
  Vbat = 3.7 V  ->  P = 1.3 W   (typical)
  Vbat = 4.2 V  ->  P = 0.8 W   (end of CV phase)

This matters for two reasons. First, the first minutes of a charge are the hottest, so a design that survives a full charge from 3.7 V may still throttle on a deeply discharged pack. Second, when the current tapers in CV, the heat falls — which is why a charger that is too hot at the start usually settles down later and gives a false sense of safety.

Worst-case sizing should therefore use the empty-battery voltage, not nominal. The difference at 1 A is 0.7 W, which is more than half of the typical figure.

3. What thermal regulation does — and what it does not

The TP4056 includes a thermal feedback loop. When the die gets too hot, it reduces the charge current rather than letting the junction run away. The datasheet describes this as thermal regulation and sets the absolute maximum junction at 145 °C.

This is a safety net, not a design margin. Two consequences follow:

• The charger may never deliver the current you programmed, and you will not get an error flag — the battery just charges slowly.
• The part can still run hot enough to discolour the PCB or affect nearby electrolytic capacitors.

The θ values are planning estimates for an SOP-8 with and without a thermal pad. Measure your own board — but the shape of the table is what matters: at 1 A, the difference between a good pad and no pad is the difference between comfortable and marginal.

4. Charge time: the number users actually feel

Charge time is capacity divided by current, plus a correction for the constant-voltage phase. From CV entry to C/10 termination typically adds 30–40 %:

t (h) = (C / 1000) / I x 1.35

1000 mAh at 1.0 A:  1.35 h
2000 mAh at 1.0 A:  2.70 h
3500 mAh at 0.5 A:  9.45 h

This is why asking a TP4056 to charge a 3500 mAh pack feels disappointing. The part is not broken; the arithmetic is simply against it. At 0.5 A the last third of the charge takes as long as the first two thirds combined.

5. Input voltage is the hidden variable

The datasheet operating range is 4.0 V to 8.0 V, and the part will work across all of it. Efficiency, however, is roughly the battery voltage divided by the input voltage:

eta = Vbat / Vin x 100 %

5.0 V in:  74 %
6.0 V in:  62 %
8.0 V in:  46 %

At 8 V input, more than half the energy becomes heat. The absolute maximum input is 8 V, but the practical maximum for a 1 A charge is 5.5 V — and 5.0 V is better still.

6. What goes wrong in the field

Case 1: a 6 V solar panel that "heats very badly"

A designer charged a 3500 mAh cell from a 6 V, 9 W panel and reported the IC heating so badly it was uncomfortable to touch, at any Rprog from 1 kΩ to 4.7 kΩ (source). At 6 V and 1.2 A, dissipation is 2.76 W. Even with a good pad that is a 138 K rise, so the thermal loop throttles hard, and the package still runs near its limit.

Fix: move the input to 5 V, or accept a much lower charge current. A solar source makes this worse because the panel voltage rises when the load is light.

Case 2: paralleling chargers to charge a bigger pack

Another common question is whether two TP4056s can be paralleled to get 2 A into a 4000 mAh pack (source). In principle the current adds, but termination does not: each charger measures its own current, so one can terminate while the other keeps charging. Two independent CC/CV loops fighting over one cell is not a design.

Fix: if you need 2 A, use a charger rated for it, or a switching charger — not two linear parts in parallel.

Case 3: "it takes forever to finish"

Slow completion is usually one of three things: the CV phase doing its job, a thermal loop holding current below the programmed value, or a termination threshold that is never reached because the battery leaks. Check the actual current during the CV phase before changing Rprog.

A useful diagnostic sequence: measure current at 3.9 V (mid CC), at 4.15 V (CV), and after the charge LED changes state. If the mid-CC current is below 80 % of the programmed value, the thermal loop is active. If the CV current never falls below C/10, the cell or the protection board is leaking. If both look correct but the total time is long, the pack is simply bigger than the charger.

One more trap: many modules place the two status LEDs across the input rail with 1 kΩ resistors. Two LEDs at 5 V add about 6 mA — negligible for heat, but they make the board look warmer than the charger itself, which sends people chasing the wrong component.

7. Choosing a topology

Who should not use the TP4056: any design charging above 1 A; any input above 5.5 V without accepting throttled current; any sealed enclosure with a pack over 2000 mAh; and any product that must operate while charging, where a power-path charger is the correct part.

What the ready-made modules get wrong

Most TP4056 boards in circulation are 1 A modules with a 1.2 kΩ PROG resistor, a USB connector and two LEDs. They are convenient and they are usually thermally optimistic.

• Copper: the pad is often connected to a small pour, not a plane. θ is closer to 80 °C/W than 50 °C/W.
• Current: 1.2 kΩ means 1 A into any pack, including a 600 mAh cell where 0.5C would be more appropriate.
• Input: micro-USB and USB-C boards often omit the 5.1 kΩ CC resistors on USB-C, so a C-to-C cable will not deliver power at all.
• Protection: the DW01 + dual MOSFET pair protects the cell, not the charger. It does not limit charge current.

If you use a module, measure the charge current and the board temperature before shipping. If you design your own, treat the pad and the PROG resistor as the two components that decide whether the product works in the field. SOP-8 EP variants exist for exactly this reason and are worth the small price difference — see the exposed-pad version.

8. Board-level details that matter more than the resistor

• Thermal pad: the SOP-8 with exposed pad is not optional at 1 A. Connect it to a ground pour with vias; without it, θ rises by tens of K/W.
• Input capacitor: 10 µF ceramic close to VCC. Long leads add inductance and the charger can oscillate on a weak adapter.
• Battery capacitor: follow the datasheet value; too small causes instability at CV entry.
• PROG resistor: 1 % metal film, short trace. A drifting resistor changes both current and termination.
• Status LEDs: their current comes from VCC, not the battery, but they add to the thermal load of the same small board.

For a related thermal calculation on a linear regulator, the same arithmetic appears in the AMS1117 thermal budget guide.

Why the last 20 % takes an hour

During CV, the charger holds 4.2 V and the current decays exponentially. The cell is already at roughly 70–80 % capacity when CV begins. If you plot current against time, the area under that tail — not the CC phase — often accounts for a third of the total charge time.

Three things stretch the tail further:

• Cold cells: internal resistance rises, the CV transition happens earlier at a lower current.
• Aged cells: increased self-discharge means the current may never fall to the C/10 threshold, so the charger keeps topping up.
• Measurement point: if the charger senses voltage at the board rather than at the cell, trace and protection-FET drop makes it think the cell is fuller than it is.

Practical consequence: for a 3500 mAh pack, a 1 A setting may still take 5–6 hours in a warm enclosure because the thermal loop keeps pulling the current down during CC. Measure the CC current, not just the programmed value.

9. Verification on the bench

10. Design review in three questions

What a scope shows that a multimeter hides

A multimeter reports the average charge current and will happily show 1 A while the part is thermally throttling in bursts. Put a current probe or a shunt on the battery lead and watch the first 60 seconds. A clean CC phase is a flat line. A throttled charger shows a sawtooth: current rises, the die heats, the loop cuts it back, the die cools, and the cycle repeats every few seconds.

That waveform is the single most useful diagnostic for this part. If you see it, the resistor value is irrelevant — the thermal path is the limit.

1. What is (Vin − Vbat) × Ibat? Above 1 W, the pad and the copper decide whether 1 A is real.
2. How long will a full charge take? If it is over 6 hours, change the current or the topology before shipping.
3. What is the worst-case ambient? Add 15–20 °C for a closed plastic case, then check Tj in the table.

For incoming inspection and date-code checks on chargers, see our quality and traceability process. For volume pricing and alternates, the battery charging collection lists what we hold in stock, including SOP-8 EP variants.

Two builds that look identical can behave differently because of the regulator vendor. We stock several TP4056 sources, including this ESOP-8 part and the standard SOP-8 version; check drop-out and thermal behaviour per source rather than assuming they are interchangeable.

11. FAQ

What resistor gives 1 A on a TP4056?

1.2 kΩ. The relationship is Ibat (mA) = 1200 / Rprog (kΩ), so 1.2 kΩ gives about 1000 mA. Do not go below 1 kΩ; the absolute maximum BAT current is 1200 mA.

Why does my TP4056 get hot even at 500 mA?

Because heat depends on the voltage across the part, not just current. At 5 V input and a 3.7 V cell, 500 mA dissipates 0.65 W. In a small board without a thermal pad that can be a 50 K rise, and in a sealed box the ambient adds to it.

Can I use a 6 V or 9 V adapter?

The operating range is 4.0–8.0 V, so 6 V works electrically but doubles the dissipation at the same current. A 9 V adapter exceeds the absolute maximum input of 8 V and must not be used.

Does the thermal regulation protect the chip?

It limits the die temperature by reducing current, which protects the part. It does not protect your charge-time promise, and it gives no error output — the battery simply charges slower than programmed.

Can I parallel two TP4056s for 2 A?

Not recommended. Each charger runs its own CC/CV loop and its own termination detection; one can stop while the other continues. Use a single charger rated for the current, or a switching charger.

Why does charging slow down near the end?

That is the constant-voltage phase. Once the cell reaches 4.2 V, current falls exponentially until it hits C/10 of the programmed value. For a 1000 mAh pack at 1 A, expect the last 30–40 % of the time to deliver only a fraction of the capacity.

12. Sources

• TP4056 datasheet (NanJing Top Power ASIC) — 4.0–8.0 V input, 1200 mA BAT maximum, 145 °C junction maximum, thermal regulation, 4.2 V ±1.5 %, C/10 termination.
• Texas Instruments, BQ24040 datasheet — linear charger comparison and power-path context.
• Microchip, MCP73831 family datasheet — small linear charger comparison.
• Electronics StackExchange, TP4056 heating very badly (2018).
• Electronics StackExchange, Charging a 4000 mAh Li-ion battery using multiple TP4056 ICs (2018).