If powering a MatrixPortal and associated LED matrix solely through the USB port, you might encounter a situation where the image appears stable when only a few pixels are lit, but brighter images with more pixels lit may exhibit “ghosting” or “sparkling” artifacts.

This is not a software bug. Please do not report it as a software bug.

This is a USB power issue. It does not affect everyone because there are so many different USB power sources and permutations. What follows is an explanation and some workarounds.

The Source of Gremlins

Not all USB ports are created equal. The ports on many computers are designed to provide 5 Volts at a maximum of 1 Ampere (1A)…or sometimes as little as 500 mA (0.5A). Likewise with USB wall chargers: small ones might deliver 1A or less, and even beefier models seldom exceed 2–2.5A.

LED matrices can be power hungry. And the more pixels in use, the hungrier they are. When a matrix demands more power than is available through the USB port, this can result in a momentary brown-out condition, and the sort of visual glitches described above. This can be exacerbated by cheap, thin USB cables that have a lot of resistance.

Complicating issues, the voltage and current from a USB port are not rock steady. This is especially true of USB battery banks. The buck/boost converter circuit in these devices creates a sort of “buzz” in the output; on average it might be 5V 2A, but the instantaneous voltage and current will oscillate around this. This is no problem for a typical steady use like charging a phone, but LED matrices are weird…it’s not just that they’re hungry, but like the supply voltage, the LED current draw also oscillates they’re actually flashed on and off hundreds of times a second to produce an image.

Some Solutions to Try

If you encounter the aforementioned phenomenon, here are some things to try, starting from the very simple and working our way up…

• Try swapping the USB cable if the one you have seems thin. That free cable included for charging a mouse or Bluetooth speaker might be unsuitable for the amount of current we need. Look for something beefy, like a tablet or laptop charging cable. Those luxury braided cables are also sometimes a sign of “substance.”
• If powering from a computer’s USB port: try using a sizable (2A or better) USB wall charger instead.
• If powering from a USB wall charger or USB battery bank: Try swapping out for any others you might have on hand. More current (higher amperage) is always nice, but that’s not the whole story. As explained above, the output filtering (or lack of) can affect the steadiness of the output voltage and current.

Those are the easy “switch something out” fixes. If none of those address the issue, now things get progressively more complex…

• Try powering the MatrixPortal and LED matrix separately: USB for the MatrixPortal, and a 5V DC “wall wart” supply (2A or preferably more) for the matrix, using one of these screw terminal power adapters (i.e. don’t use the screw terminals on the MatrixPortal, use this instead):

• If the project must be powered from a USB battery bank and you’ve already tested some others: try powering the LED matrix and the MatrixPortal separately. This might involve two power banks, or one with multiple ports might suffice (most enforce per-port current limits), you’ll need to experiment. Cables exist to convert USB type A or C to a DC plug that’s compatible with the screw terminal adapter above — one would power the LED matrix through this, and the MatrixPortal through a normal USB cable. Just make sure the cable provides 5V output; some are “booster” cables to higher voltages that could destroy the matrix.

• If nothing else seems to work, the last line of defense is better filtering on the power supply output. This involves adding one or more capacitors across the + and – terminals until the image stabilizes. Unfortunately there is no one-size-fits-all solution here…it requires experimentation with an assortment of caps on hand. One might start with a 1000 µF cap, and work up or down from there. If something seems promising but imperfect, two or more with different values might be needed in parallel (e.g. 100 µF + 1000 µF).