When it doesn't work

Work down the list. Each section starts with the cause that turns out to be right most of the time.

Nothing happens at all

No light anywhere, nothing on the screen.

1. Is the Nano actually in its socket, the right way round? The USB outline is printed on the silkscreen at the bottom edge of the board. The USB connector goes there.
2. Is the board getting 5 V? Measure between the VCC and GND pins of any module header. Expect 4.8 – 5.2 V.
3. Solder bridge across the Arduino socket. Those 30 pins are the tightest spacing on the board. Look along the back at a shallow angle, in good light.
4. The Nano itself is dead. Unplug it from the board entirely and plug it into your computer with USB. Its own power LED should light. If it does not, the Nano is the problem, not your soldering.

The upload fails

avrdude: stk500_recv(): programmer is not responding

This is the single most common problem and it is almost never a fault.

Change Tools → Processor to ATmega328P (Old Bootloader) and upload again. Most inexpensive Nano clones ship with the older bootloader, and the IDE defaults to the new one.

Port not found / no port in the Tools menu

The CH340 USB-to-serial chip used on clone Nanos needs a driver on Windows and older macOS. Install the CH340 driver, unplug, replug.

Also check your USB cable — a surprising number of cheap cables are charge-only and carry no data lines at all. Try a different one.

Upload starts and then fails partway

Unplug everything from the module headers and try again. A sensor drawing a lot of current during the upload can brown out the board.

The screen

Backlight is on, but the display is blank or shows a row of white blocks

Wrong I²C address. Change LCD_ADDRESS in the sketch:

const byte LCD_ADDRESS = 0x3F;   // was 0x27

The two common addresses for these backpacks are 0x27 and 0x3F, and there is no way to tell from the outside which one you have.

If neither works, upload an I²C scanner sketch and read the address it finds.

Backlight is on, contrast looks wrong — very faint or solid blocks

There is a small blue potentiometer on the I²C backpack, on the back of the display. Turn it slowly with a screwdriver while the board is running.

Nothing at all, not even a backlight

Check the module's 4 pins against the header: GND, VCC, SDA, SCL, counting from the top of the board. If it is plugged in one position off, or backwards, nothing happens.

The screen works but the text is garbled or freezes

Usually a loose SDA or SCL joint. Reflow the four pins of the LCD header.

The LEDs

They are lit when they should be dark, and the colours look inverted

The J5 jumper and the COMMON_ANODE setting disagree with each other.

Change one of the two — not both.

One colour never lights

1. The LED is rotated. Pull it out and check the long leg is in position 2 of the socket.
2. The resistor for that colour has a bad joint. R4/R5/R6 serve LED1; R7/R8/R9 serve LED2. Reflow both ends.
3. Test it in software — set that one pin high in setup() and see if the colour appears.

No LED lights at all

Is the jumper cap on J5? Without it, the LEDs' common leg is not connected to anything and neither LED can light. This one catches everybody once.

The sensors

Temperature and humidity read nan, or the screen says the sensor failed

1. DHT11 plugged in backwards. The three pins are VCC, DATA, GND counting from the top of the board.
2. The middle pin is not making contact. These modules have long thin pins that bend easily.
3. You are reading it too fast. A DHT11 needs about 2 seconds between readings. The supplied sketch already respects this; if you have edited SENSOR_INTERVAL_MS, put it back to 2000 or more.

The CO reading is nonsense, or wildly high in clean air

1. Give it time. The MQ-7 has an internal heater that must come up to temperature. The screen says warming up for the first 20 seconds; the reading is meaningless before that, and for a brand-new sensor the first few hours of use are unreliable while it burns in.
2. It is not calibrated. MQ7_R0 defaults to 10.0 kΩ, which is a guess. Follow the calibration steps in the README.
3. The module is plugged in wrong. Pin 1 of the MQ-7 header goes to A7 and must receive the module's analogue output; pin 2 goes to D12 and takes the digital one. MQ-7 breakout boards differ between suppliers — read the labels printed on your module rather than trusting the pin order. (The pin legend on the schematic itself has these two swapped — see hardware-reference.md §7.)

GAS ALARM appears with no sensor plugged in

It should not — D12 is configured with its internal pull-up for exactly this reason. If it happens anyway, check for a solder bridge near the MQ-7 header.

The buttons

Nothing happens when I press

1. Wrong button. SW1 = next screen, SW2 = previous, SW3 = °C/°F. They are marked on the silkscreen.
2. A cold joint on the switch. Reflow all four legs.
3. The switch is rotated 90°. A 4-leg tactile switch fits the board in two orientations, but only one of them connects the pair the board uses. If it went in without any force, it may be the wrong way. Desolder, rotate 90°, refit.

One press registers as several

The debounce window is too short for your switches. Increase it in the sketch:

const unsigned long DEBOUNCE_MS = 60;   // was 40

Power

The Nano gets hot

Disconnect immediately.

• USB and the screw terminal are both connected. This board has no protection between them. Use one or the other.
• The screw terminal polarity is reversed. + and − are marked on the back.
• There is a short between 5 V and GND. The front of the board is a +5 V plane and the back is a GND plane, so a stray leg or a blob can bridge them easily. Measure resistance between a VCC and a GND pin with the power off — anything under about 100 Ω means you have a short.

The screen dims or the board resets when the MQ-7 warms up

The MQ-7's heater draws around 150 mA. If you are powering everything from a laptop USB port through a long thin cable, you can run out of headroom. Use the screw terminal with a proper 5 V supply, or a shorter cable.

Still stuck

Open the Serial Monitor at 9600 baud and watch what the board says as it boots. It prints CubeSat booting..., then reports readings once a second. That output usually tells you which stage is failing.

If you think you have found a fault in the board rather than in your build, please open an issue with a photo of both sides of your assembled board.