Charlieplexing a LED « LeafLabs Garden

LeafLabs Garden » Support » Maple Hardware Support

Charlieplexing a LED

(12 posts) (4 voices)

Started 5 years ago by MessiahAndrw
Latest reply from gbulmer

MessiahAndrw
Member

Hello,
I'm trying to charlieplex a large array of LEDs through the Maple. However to start with I am just trying to control 2 LEDs via 2 pins like in this diagram:

http://upload.wikimedia.org/wikipedia/commons/a/ad/Complementary_Drive.png

What I tried to do is connect an LED from pin 0 to 1, and another LED from pin 1 to 0, then alternate which is glowing by running:

void loop()
{
pinMode(0, OUTPUT);
pinMode(1, INPUT);
digitalWrite(0, HIGH);
delay(1000);
digitalWrite(0, LOW);

pinMode(0, INPUT);
pinMode(1, OUTPUT);
digitalWrite(1, HIGH);
delay(1000);
digitalWrite(1, LOW);
}

But nothing happened.... I tried using just 1 LED from pin 0 to 1..
void setup()
{
pinMode(0, OUTPUT);
pinMode(1, INPUT);
digitalWrite(0, HIGH);
}

void loop()
{
}

But still nothing happened.. However by connecting from pin 0 to ground, the LED was able to glow. I ruled out that this was an issue with pin 1, because I tried other combinations of pins and they didn't work.

Is there something I am missing, or cannot the Maple do charlieplexing?

Posted 5 years ago #
MessiahAndrw
Member

I've solved this problem, rather than use input, I must use OUTPUT_OPEN_DRAIN.

Posted 5 years ago #

soundcyst
Member

yeah, you definitely want them both to be outputs, but you shouldn't need to change the pin mode in the loop.

try:

void setup()
{
    //config directions
    pinMode(0, OUTPUT);
    pinMode(1, OUTPUT);

    //init to both off
    digitalWrite(0, LOW);
    digitalwrite(1, LOW);
}

void loop()
{
    digitalWrite(0, LOW);
    digitalWrite(1, HIGH);
    delay(1000);

    digitalWrite(0, HIGH);
    digitalWrite(1, LOW);
    delay(1000);
}

Posted 5 years ago #

gbulmer
Moderator

I don't believe it is essential to change the pinMode when using two pins to drive two LEDs, but it would do no harm when both LEDs are off.

It is not essential to change the pinMode of pins when driving a simple matrix of LEDs (single LEDs), but again, it would do no harm.

But, I have never found a way to drive a matrix of 'Charlieplexed' LEDs (i.e. LEDs which are 'paired' to be driven with inverted signals) without setting some pins to high-impedance, i.e. INPUT.

I think a good place to start is driving 8 LEDs with 4 pins. It is also important to ensure their are no pull-up or pull-down resistors active, otherwise there will be a dim LED glow even when the pin is in INPUT mode.

Posted 5 years ago #
soundcyst
Member

right, changing the pin direction only comes into play when there are more than two pins in use so that only two pins are source/sink-ing at a time.

http://en.wikipedia.org/wiki/Charlieplexing#Charlieplexing

if less than two pins are source/sink, then it won't work, and if more than two pins are source/sink, more than one LED will light up at a time.

so, my comment above about changing pin directions is really only applicable to the 2-led version

Posted 5 years ago #
poslathian
LeafLabs

Be careful driving current into an output pin. For example if you wire two GPIO's together, both as outputs and drive a 1 into a 0, you risk frying the internal gpio circuitry.

If you are trying to drive LED's in a matrix style, I recommend wiring them up with each row sharing a single cathode and each column sharing a single anode - then using one transistor per row to source current through that row, and one transistor per column to drain that current to ground. Err I dont have a diagram but I can provide more info if needed, we did this in an 8x8 grid for some project a while back and it worked fine.

Posted 5 years ago #
MessiahAndrw
Member

Charlieplexing proved promising at first with its higher LED-count-per-pin (n(n-1) LEDs to pins) that conventional LED matrices, but by binary selecting the row you can dramatically save on pin count (p(2^n)). I use the 15 PWM pins to power the column pixels, and 6 other GPIO to binary select the row to ground (leading to a maximum of 64 rows even though not all rows are used).

The system I've implemented allows up to 960 LED channels (I use the term 'channel' rather than 'LED' because I'm using RGB LEDs) even though the number actually used are a lot less. I haven't done a large scale test yet since I need to construct the board the LEDs will be placed on.

Posted 5 years ago #
poslathian
LeafLabs

wow 960 channels off of 21 GPIO's? can you post a simple schematic? Im not sure I fuly understand how youve managed to pull that off. Very interesting...

Posted 5 years ago #
MessiahAndrw
Member

Here you go:

http://img40.imageshack.us/img40/5043/ledmatrix.png 1208x844 - 54Kb.

Posted 5 years ago #
gbulmer
Moderator

Have you wired this up and got it working? Does it look okay?

Each LED row will have a mark:space ratio of 1:63, so I'd expect it to be dim.

When I've run 8 columns x 16 rows (using a couple of 8x8 LED matrix displays) they seemed noticably dim.

So it wouldn't need all wiring up, it'd just need to run the LEDs at a mark:space of 1:63.

Posted 5 years ago #
MessiahAndrw
Member

I haven't wired it up since I'm sending my Maple back due to my USB connector coming off.

Anyway, the code for the PWM and timing interrupt is done, and I was able to test it with just two LEDs connected being lit at 1:64. The RGB LEDs I'm using are bright enough to make me squint when constantly powered, and also because our perception of brightness increases logarithmically not linearly 1:64 appears to be about a 3rd of the normal brightness. This is acceptable because I'm building a display where I just want to 'pixels' to change colour and not glow.

Posted 5 years ago #
gbulmer
Moderator

The natural log of 64 is over 4 (and most things in nature seem to be natural logs :-), so isn't it less than 1/4 the brightness?

That aside, if you've tested it, and are happy, everything is fine; that's all that I was trying to alert you to.

Posted 5 years ago #

RSS feed for this topic

Reply

You must log in to post.