When it shows nothing
Six symptoms, and which half of the problem each one lives in. A dark display is nearly always two crossed wires; a colon that will not light is nearly always the wrong panel or the wrong bit.
Start with where, not why
Four places a fault can be: the sketch, the four wires, the supply, or the panel that is fitted. Almost every question about this display is answered by working out which one before changing anything.
The two that account for most of it
Dark. CLOCK and DATA crossed over. The chip only begins listening when DATA falls while CLOCK is high, so with the wires swapped that pattern never arrives and it ignores every byte — silently, undamaged, indistinguishable from a board with no power. Swap them before you suspect anything else.
No colon. In order: the panel is a digit one and has points instead; the
mask is 0b10000000, which is the first digit and lights nothing on a clock
panel; or the number is 0 with leading zeros off, which makes the released
library drop the dots altogether.
Things that are not faults
An I²C scanner finding nothing is correct — there are no addresses in this protocol.
Levels 3 to 7 looking the same is correct — they are 10 to 14 sixteenths, and all the real dimming is at 0 to 2.
Digits missing from a photograph is correct — the chip lights one digit at a time, and a fast shutter catches it.
A number that stays on the display after the board resets is correct, and it is the one to keep in mind, because it is also what a crashed sketch looks like.
Make it admit when it is lying
The display has no idea whether your sketch is alive, and it will hold the last thing it was told indefinitely. Every finished build should have one of these:
- a colon or a point that blinks once a second, which stops when the sketch stops;
- four dashes whenever the value is unknown or stale, rather than the last good reading;
setBrightness(n, false)to blank the display on an error, which keeps the digits in the chip and brings them back when the error clears.
None of them costs more than a line. All of them turn a confident wrong number into an obviously blank display, which is the difference between a project that is wrong and a project that says so.
When it does not work
Measure VCC and GND at the display's own pins rather than at the board's header — a jumper that looks seated can be sitting on the plastic beside the pin. If VCC is there and the display is still dark, swap CLOCK and DATA anyway: crossed wires are indistinguishable from a dead board and damage nothing.
If the same bar is missing on every digit, that is one segment line, which points at the display's solder joints rather than anything in your code. If it is one whole digit, that is its common. Both are unusual on a factory-soldered board; check first that the sketch is not writing the pattern you are seeing.
Long thin leads and a tight fold are where intermittent contacts live. The library's default 100 µs delay is generous enough for a metre of wire, so speed is rarely the cause — start with the connector, then the cable.
A byte was corrupted in transit and the chip acted on it as a command. Anything that shares a supply with a motor or a relay can do this. Re-sending the whole display once a second costs nothing and repairs it automatically.
The same idea with sixty-four LEDs instead of thirty-two, when seven bars stop being enough.
The 8x8 LED matrix →Edit this page — content/books/tm1637-display/when-it-shows-nothing.mdx
Questions about this product
See what other owners have asked, and read their solutions.
This page covers several products. Choose yours to see the right discussions.
Discuss this article
Ask about this page. The answer stays here, on the page it belongs to, for whoever hits the same wall next.