When the number is wrong

Four symptoms

Pick what your Serial Monitor shows.

The first three are the sketch or the wiring, and each has one fix. The fourth is physics, and the fix is where you put the board.

nan: nothing answered

nan is the only error this library reports. It means no forty bits with a good checksum came back: nothing on the wire, the wrong pin, or a sensor that did not wake. The TK39 is the TK38's circuit, so when every reading is nan in the DHT11 book is the checklist, in order.

Two causes are particular to this board. On 3V3 the DHT22 is at the bottom of its 3.3 to 6 V range, so a long thin supply wire has nothing to spare. And a sketch that says DHT22 on a TK38 sends a start pulse the DHT11 usually ignores.

A steady wrong number: the type

0.00 C 2.00 % in a warm room, every time, with no nan: the sketch says DHT11. The wrong type shows why the checksum passes. Change one word.

The same line again: the cache

Asked more often than every 2000 ms, Adafruit's library hands back its previous reading without touching the wire. The numbers are real and old; they hold, then jump. A failure is cached the same way, so a retry inside two seconds gets the same nan. Every two seconds times the loop so it cannot happen.

Warm and dry: the placement

A sensor warmer than the room reads warm, and it also reads dry. The water in the air is the same; at the sensor's higher temperature it is a smaller fraction of what the air could hold. At 23.4 °C and 51.2 %, two degrees warm costs about 6 % of humidity, three times the DHT22's ±2 %.

The DHT22 itself draws 1 to 1.5 mA while measuring (§7), for about 5 ms every few seconds, so it does little to warm itself. What warms it is its neighbours: a regulator, a processor, a USB socket, a breadboard next to them. The datasheet's §8 asks for it to be mounted as far as possible from parts that change its temperature. How much a given board warms it has not been measured for this book; move it away on longer wires and compare.

Wet: the element

Water on the humidity element reads as saturated air, so a sensor that has had dew on it, or has come in from the cold, can read at or near 100 % until it dries. The datasheet asks you to avoid using it where dew forms, says it recovers from conditions outside its range at the cost of faster ageing, and gives a reconditioning procedure: two hours at 50 to 60 °C under 10 % humidity, then five hours at 20 to 30 °C over 70 %.