Accuracy

DHT11 is ±5% RH / ±2°C; DHT22 ±2–5% / ±0.5°C; good I2C parts (SHT31, HTU31, AHT20, HDC1080) hit ±2% / ±0.2–0.3°C. Match accuracy to whether you need calibrated data or just a rough reading.

Interface

Single-wire (DHT) is dead simple but one sensor per pin and slow. I2C lets you chain several sensors on two wires and read fast — the better choice for anything beyond a beginner project. BME280 also offers SPI.

Power Draw

For battery and coin-cell projects, sleep current matters most. The HDC1080 sleeps at ~100 nA; Si7021 and SHTC3 are also frugal. DHT sensors and always-on BME680 gas heating are thirstier.

Extra Channels

Need more than T&H? BME280 adds barometric pressure (and altitude); BME680 adds a VOC gas channel for air-quality projects. If you only need temperature and humidity, skip the extras and save money.

Environment & Filter

For dusty, humid or condensing places, choose a part with a PTFE filter (SHT31, Si7021) or a sealed cabled probe (AM2302/AM2301B). Bare breakouts belong indoors, away from splashes and direct sun.

The DHT22 (AM2302) is the sensor most people meet first, and for good reason. The 3-pin module has the pull-up resistor built in, so you just wire power, ground and one data line and load the Adafruit DHT library. It reads the full 0–100% humidity range and works down to −40°C — a big step up from the DHT11 — making it the go-to for greenhouse, weather and learning projects where you want real data without fuss.

This little combo board pairs the modern AHT20 humidity/temperature sensor with a BMP280 pressure sensor on one I2C bus, giving you BME280-style 3-in-1 data for noticeably less money. The AHT20 is genuinely accurate (±2% RH, ±0.3°C) and far more reliable than the old DHT parts, and the two chips sit at different I2C addresses so they coexist happily. You load two libraries instead of one, but for a budget weather node it’s hard to beat the value.

The blue DHT11 is the cheapest way to put temperature and humidity into a project, often sold in multi-packs for the price of a coffee. It wires up exactly like a DHT22 and uses the same library, so it’s perfect for first Arduino lessons, classroom kits and throwaway prototypes. Just respect its limits: 0–50°C only, no sub-zero readings, ±2°C and ±5% RH — fine for “is it humid in here?” but not for anything you’d call data.

The BME680 takes everything the BME280 does — temperature, humidity, pressure — and adds a MOX gas sensor that responds to volatile organic compounds, letting you build indoor air-quality monitors and “is the room stuffy?” dashboards. It speaks both I2C and SPI and runs on the same Bosch ecosystem. The gas channel needs a burn-in period and gives a relative air-quality index rather than calibrated PPM, but for hobby IAQ projects it’s the most capable single sensor here.

The Si7021 from Silicon Labs is a tidy, reliable I2C temperature/humidity sensor that sips power, making it a favorite for battery-powered ESP32/Feather nodes and wearables. It reads ±3% RH and ±0.4°C, includes a PTFE filter to keep the sensing element clean, and runs happily on 3.3V or 5V. It only measures temperature and humidity — no pressure or gas — but it does that job cleanly and efficiently with a well-documented Adafruit library.

If you like the simplicity of a DHT but want real accuracy, the HTU31 (TE’s third-gen HTU sensor) is the upgrade. For about the same price as a DHT22 it gives you ±2% RH, ±0.2°C, a clean I2C interface with two selectable addresses, CRC-checked data so you can trust each reading, and an on-board heater to evaporate condensation. It’s the sensor to choose when “good enough” isn’t, but you don’t want to pay Sensirion money.

When battery life is everything, the HDC1080 from Texas Instruments is the part to know. It combines ±2% RH and ±0.2°C accuracy with a sleep current of roughly 100 nanoamps — low enough to run for years on a coin cell. The common CJMCU-1080 breakout puts it on a tiny I2C board for very little money. Build quality on generic boards varies, and it’s temperature/humidity only, but for a sleepy LoRa or BLE sensor node nothing here sips less power.

The AM2302 is a DHT22 in a chunky plastic body on the end of a ~23 cm cable, with the pull-up resistor already inside — purpose-built for placing the sensing element away from your board, outside a box, or near a vent. That cabled form factor is what makes it the easy choice for outdoor weather stations and remote logging. Note that the original Adafruit AM2302 is being phased out in favor of the sealed AM2301B (wired AHT20), which is more accurate; if you want the freshest design, consider that successor.

Just Starting Out?

Grab a DHT11 multi-pack to learn on, or jump straight to the DHT22 for real range and accuracy with the same simple library.

Weather / Room Monitor?

The BME280 adds barometric pressure for forecasting and altitude, all on I2C/SPI — the best all-rounder for environmental projects.

Need Real Accuracy?

The SHT31-D (±2% / ±0.3°C) or the cheaper HTU31 (±2% / ±0.2°C) give you data you can actually log and trust.

Battery-Powered Node?

The HDC1080 sleeps at ~100 nA; the Si7021 is also frugal. Both are ideal for LoRa/BLE sensors that must last on a coin cell.

Air-Quality Project?

The BME680 adds a VOC gas channel on top of temp/humidity/pressure — the one to pick for indoor air-quality dashboards.

Outdoor / Remote Spot?

Use the cabled AM2302 (or its sealed AM2301B successor) so the probe sits away from your board, sheltered from sun and splashes.

DHT22 or BME280 — which should I pick?

If you only need temperature and humidity and want the absolute simplest wiring, the DHT22 is fine. But the BME280 costs only a little more, adds barometric pressure, uses I2C/SPI so you can chain it with other sensors, reads faster, and is more accurate. For most projects beyond a first lesson, the BME280 is the better long-term buy — just make sure you receive a real BME280 and not a BMP280.

Why does my “BME280” show no humidity reading?

You almost certainly received a BMP280, not a BME280. They look nearly identical and share most of the register map, but the BMP280 has no humidity sensor at all. Check the chip marking or the I2C chip-ID register, and buy from a reputable seller (or a branded Adafruit/SparkFun board) if humidity is non-negotiable. This swap is the single most common complaint with cheap modules.

Is I2C better than the single-wire DHT protocol?

For anything beyond a beginner project, yes. I2C lets you put several sensors on the same two wires (SCL/SDA), reads quickly, and has robust, well-maintained libraries. The DHT single-wire protocol needs precise timing, allows only one sensor per data pin, and limits you to a reading every 1–2 seconds. The DHT’s only real advantages are simplicity and cost — which is exactly why it’s still great for learning.

How accurate do I actually need?

It depends on the job. A fan or thermostat trigger is happy with ±2°C / ±5% RH (DHT11 territory). A weather station or comfort monitor wants ±0.5°C / ±2–3% RH (DHT22, BME280, Si7021). Calibration, scientific logging or HVAC validation needs ±0.2–0.3°C / ±2% RH — that’s where the SHT31, HTU31, AHT20 and HDC1080 shine. Buying more accuracy than you need just costs money and, sometimes, power.

Can I use these sensors outdoors?

Only with care. None of these breakouts are waterproof, and condensation or direct sun will ruin readings. For sheltered outdoor use, pick a part with a PTFE filter (SHT31, Si7021) and mount it in a vented radiation shield out of the rain, or use a sealed cabled probe like the AM2302/AM2301B. Keep the PCB and electronics inside an enclosure; expose only the sensing element to the air.