WelkinAir quality sensor board

What it measures

  • CO₂ — 400–5000 ppm; ±(50 ppm + 5% of reading), pressure-compensated
  • VOC + NOx — 1–500 relative index
  • Barometric pressure — ±3 Pa relative (≈ ±0.25 m), ±50 Pa absolute
How it connects — power + I²C inside the board
HOST MCU I²C host WELKIN SENSOR BOARD XC6220 3.3 V LDO 3.3 V VIN 3.4–5.5 V EN CE I²C SDA/SCL SCD41 CO₂ photoacoustic I²C 0x62 ≤205 mA pulse SGP41 VOC + NOx (MOX) I²C 0x59 ~3 mA continuous BMP390L barometer I²C 0x76 ~3.4 µA
amber = power  ·  teal = I²C bus  ·  dashed = EN (control)
Board layout — to scale
thermal-relief slitSDASCLVDDGNDEN▸1SCD41CO₂SGP41BMP390LXC622016.0 mm34.0 mm
Top view · 16 × 34 mm
6.5 mmothers < 1 mm
Side · 6.5 mm tall

SCD41 on a milled thermal-isolation island at the vent end.

Specs

BlockPartSpec
CO₂ sensorSCD41400–5000 ppm · ±(50 ppm + 5% of reading) · photoacoustic (optical, true CO₂ — not NDIR or eCO₂) · pressure-compensated
VOC + NOx sensorSGP41VOC + NOx indices, 1–500 each · relative MOX trend via Sensirion Gas Index Algorithm · not absolute ppb
Pressure sensorBMP390L±3 Pa relative (≈ ±0.25 m) · ±50 Pa absolute · feeds CO₂ pressure compensation
LDOXC6220Low-noise 3.3 V regulator · ≤30 mV p-p ripple for the SCD41
Power3.4–5.5 V → 3.3 V~215 mA peak (all sensors) · ~18 mA average all measuring · ~0.2 mA idle · ~0.1 µA off (EN low, XC6220 shutdown)

FAQ

Why is the SCD41 better than an eCO₂ sensor?

An eCO₂ sensor does not really measure CO₂. It measures VOCs (gases from cooking, cleaning, people, and so on) with a small heated plate, and then it guesses a CO₂ number from them. The guess uses one idea: that CO₂ and VOCs go up together indoors. But this is often wrong. If something releases VOCs — cooking, cleaning, alcohol, a marker pen — the eCO₂ value jumps, even when the real CO₂ has not changed. And if CO₂ slowly builds up in a quiet room with no VOCs, the sensor cannot see it.

The SCD41 measures CO₂ directly and gives a real value in ppm (400–5000). It reacts only to CO₂, not to other gases. This is the number you need when you decide to open a window or turn on ventilation. In short: eCO₂ is a guess, and the SCD41 is a real measurement.

Why a 1–500 index and not an exact VOC/NOx value?

The SGP41 has a small heated surface, and its electrical resistance changes when gases touch it. The problem is that it reacts to many gases at the same time. Also, "VOC" is not one gas — it is hundreds of different ones. So the sensor cannot give an exact value (in ppb) for one single gas.

There is a second problem. The sensor's normal level slowly changes with humidity, temperature, and age, so a fixed calibration from the factory would become wrong over time. To fix this, Sensirion's Gas Index Algorithm runs on your host (it is free, BSD-3 license). It learns the sensor's own recent normal level and gives a number from 1 to 500. About 100 means "normal for you", and a higher number means the air is getting worse.

Can it work on its own?

No. It needs a host — a computer or microcontroller (MCU) — to read the data over the I²C bus. The board has no processor of its own.

Can I turn it on and off to save power?

Yes, but the two sensors are different. The SCD41 (CO₂) is OK for periodic sleeping, in single-shot mode it uses only about 0.5 µA between readings, so you can wake it only when you need a CO₂ value. The SGP41 (VOC/NOx) works better when it stays on: its algorithm needs hours to learn the normal air level, and the NOx part needs about 10 seconds to warm up after each start. If you turn off its power, it loses this progress and has to start again.

Just leave a note