INA219 vs INA260: which power monitor?
Same job — current, voltage and power over i2c — but a genuinely different philosophy: the INA219 measures across your shunt resistor, the INA260 brings its own. That one choice decides everything else.
Side by side
INA219 vs INA260: the spec sheet, condensed
| Property | INA219 | INA260 |
|---|---|---|
| Shunt resistor | external (0.1 Ω typical on breakouts) | built in — 2 mΩ, factory-trimmed |
| Current range | ±3.2 A with 0.1 Ω; re-shunt for more | up to 15 A continuous, fixed |
| Bus voltage | 0–26 V | 0–36 V |
| Setup | calibration register math required | none — direct mA/mV/mW readings |
| i2c addresses | 0x40-space, 4 on typical breakouts | 0x40–0x4F, all 16 selectable |
| Typical board | a few dollars; Gravity wattmeter version | costs more; generic header modules |
The reasoning
INA219's external shunt vs the INA260's built-in one
The INA219 is a precision voltmeter across a resistor you choose. Ship-with-the-breakout 0.1 Ω gives ±3.2 A, but nothing stops you soldering a 10 mΩ bar and reading a motor at 30 A — you just recompute the calibration register. Power comes from that flexibility, and so does the friction: get the calibration value wrong and every reading is confidently, silently wrong.
The INA260 deletes the whole problem. TI trims the 2 mΩ shunt at the factory, so the registers speak plain engineering units the moment it powers up — read, display, done. The cost of that convenience is a hard ceiling: 15 A continuous through the internal shunt, and no way to trade range for resolution. It's also the better fit for higher-voltage rails (36 V vs 26 V).
The rule: known load under 15 A and you value your afternoon → INA260. Unusual current ranges, tight budgets, or a Gravity-ecosystem build → INA219. Profiling an ESP32's WiFi spikes? Either works — that use case made the ESP32 roundup with the INA260 for the zero-math setup.
Get one
Questions
Power monitor FAQ
Can the INA219 measure more than 3.2 A?
Yes — that limit belongs to the usual 0.1 Ω breakout shunt, not the chip. Swap in a smaller external shunt (say 0.01 Ω) and recalibrate, and the same INA219 reads tens of amps. That flexibility is exactly why it survives against newer parts.
Why does the INA260 need no calibration math?
Its 2 mΩ shunt is inside the package and characterised at the factory, so the chip hands you finished numbers: current in 1.25 mA steps, bus voltage, and power, straight from registers. The INA219 can't know what shunt you soldered, so you must program a calibration value first.
Can I use both on the same bus?
Comfortably — both live in the 0x40 address neighbourhood, but each is address-selectable (typical INA219 breakouts offer four addresses, the INA260 sixteen), so give them different addresses and they coexist. Handy when you want one on the battery and one on a motor rail.
High-side or low-side — which do these support?
Both parts are happiest high-side (sensing in the positive lead), which keeps the load's ground honest; both also tolerate low-side sensing. The INA219 handles bus voltages to 26 V, the INA260 to 36 V — neither belongs on mains or anything above that.
What about measuring more than one rail?
TI's INA3221 puts three high-side channels behind one i2c address — one chip for battery, 5 V and 3.3 V rails at once. Less precise than a dedicated INA260 per rail, but far tidier for a power dashboard.