Tlc5940 multiplexing code

Those FETs require 10V to switch on the gates. Don't be fooled by the gate threshold voltage that is just where it starts to cinduct. I am not sure if that would cause your reported problem because it seems they are on all the time rather than not on.

Grumpy_Mike: Those FETs require 10V to switch on the gates. Don't be fooled by the gate threshold voltage that is just where it starts to cinduct. I am not sure if that would cause your reported problem because it seems they are on all the time rather than not on.

I'm glad you mentioned that as the data sheet did confuse me quite a bit. The gate threshold voltage shows a min-max of -2.0 to -4V and the following was in the comments on the product page:

The product description also says: The FQP27P06 is a very common MOSFET with very low on-resistance and a control voltage (aka gate voltage) that is compatible with most 5V microcontroller or mechanical switches. This allows you to control high-power devices with very low-power control mechanisms.

But again, I'm so ignorant with this stuff. Earlier you indicated that this looks like it would work, but was just over spec. But is it now looking like it won't work at all without 10v and the product page just has some mis-information?

I had a very hard experience with 5940mux library, for this reason I tried to
improve the basic TLC5940 library.

My goal is drive a lot of rgb led using tlc5940 and plex all with UDN2981.

Initially, using the basic tlc5940 library, trying a multiplex, the problems were the interferences between leds.

I solved perfectly the problem using

while (tlc_needXLAT);

after the call of

Tlc.update();

This permit to wait the completing of PWM cicle, in this way I solved the problem!

This is the engine of my multiplex code:

void ligthLedMux() {    
  for (byte j=0; j<digMax; j++) {
    for (byte k=0; k <= nLedMax; k++) {
      if (ListaTransistor[k] == dig[j]) {
        ligthLed(k,ValoriRGB[k][0],ValoriRGB[k][1],ValoriRGB[k][2]);
      }
    }
    
    Tlc.update();
    while (tlc_needXLAT);  //wait for PWM cicle

    for (byte l=0; l<digMax; l++) {      // change multiplex channel
      digitalWrite(dig[l],HIGH);
    }
    digitalWrite(dig[j],LOW);
  }

}

GianfrancoPa: I had a very hard experience with 5940mux library, for this reason I tried to improve the basic TLC5940 library.

I greatly appreciate the reply, but if I understand correctly, the library and setup from Grumpy Mike which I'm trying to get working is different from the Tlc5940Mux library and setup from Alex Leone. It seems like I'm really close to getting this working so I'd rather try to figure this out before trying a whole different setup. Besides, his setup seems less documented so I think my chances of getting it working would be smaller.

As for additional troubleshooting:

I tried looking at the serial monitor to see if there was any obvious problems there, but nothing displayed. I did set the speed to 57600.

I've checked the hardware over and over, but can't figure out what/if I screwed up. I still just get solid rows of unique colors, but not individual LEDs. I'm still hunting for FETs more like what Grumpy Mike used, but am still unclear if the FETs I'm using now are definitely a problem.

I figured it out!!

Of course it was something really dumb on my part. I had wired up the cathodes incorrectly. Now I'm going to try to scale this puppy up and wire in some photoresistors to make it interactive!

Here's what I had before:

Here's the correct hookup:

Bigger versions: http://i.imgur.com/C43Wy.jpg http://i.imgur.com/ZYYWm.jpg

I figured it out!!

Well done. I bet that mixed feeling of triumph and what an idiot I was, is good.

It is still what I feel when I solve a problem. It’s a bit addictive. :slight_smile:

I scaled it to 8X8 in my Hexome With one TLC5940 per colour so that you can have separate current control allowing you to balance the colours.

With your default 4x4 setup I'm left with D5 and D13 pins open. Can I use one of the analog pins to get the 7th row going or do I have to do something else?

'm left with D5 and D13 pins open

Not quite follow what you are saying, sorry.

Grumpy_Mike:

'm left with D5 and D13 pins open

Not quite follow what you are saying, sorry.

Sorry for being unclear. With everything wired up as per the 4x4 schematic all the digital pins are used except D5 and D13, so I figure I can use those to control FETs #5 and #6, but I'm not sure where to hookup FET #7.

Oh I see, yes you can use these to control other FETs and also use the analogue pins to control others. A better way would be to use a 74LS42 as in this project http://www.thebox.myzen.co.uk/Hardware/Econo_Monome.html to drive more FETs with fewer lines.

Grumpy_Mike: Oh I see, yes you can use these to control other FETs and also use the analogue pins to control others. A better way would be to use a 74LS42 as in this project http://www.thebox.myzen.co.uk/Hardware/Econo_Monome.html to drive more FETs with fewer lines.

An in doing so, I figure i'd need to change the line in TLC5940Multiplex.cpp:

static int8_t rowPin[4] = {8, 2, 11, 7};   // pins to enable or turn on each row, low = on

to

static int8_t rowPin[7] = {8, 2, 11, 7, 5, 13, 16};   // pins to enable or turn on each row, low = on

Did I do that right and is there anything else I'd need to update?

That looks OK, did you also set the pin mode to output for the extra pins?

Grumpy_Mike:
That looks OK, did you also set the pin mode to output for the extra pins?

I’m trying to but am a little confused.

If I put pinMode(A0, OUTPUT); into TLC5940Multiplex.cpp I get an error like

'A0' was not declared in this scope

I also tried it in void setup within Arduinome_4x4_RGB.pde but I still have the same row and columns not lighting up.

Lastly, I noticed this in TLC5940Multiplex.cpp

// iniilise multiplex row selects and turn them all off (off = 1 as we have a P channel FET)
    for (i = 0; i<4; i++){

I tried changing the 4 to 7, but then just one column lit up.

You already have:-

static int8_t rowPin[7] = {8, 2, 11, 7, 5, 13, 16}; // pins to enable or turn on each row, low = on

Which is using the A2 pin (that's the 16) so all you have to do is to use:- pinMode(16, OUTPUT);

Grumpy_Mike: You already have:-

static int8_t rowPin[7] = {8, 2, 11, 7, 5, 13, 16}; // pins to enable or turn on each row, low = on

Which is using the A2 pin (that's the 16) so all you have to do is to use:- pinMode(16, OUTPUT);

I did this in void setup() in Arduinome_4x4_RGB.pde but I'm not getting output for pin A2 or D13. From looking at the code it seems it's only setup for 4x4 RGB (48 LEDs), but my setup is now 7x7 (147 LEDs). I took several passes trying to scale the code for 7x7 but had no luck.

Do you have the code you used for your 8x8 project available online?

Yes it is the library code you have to hack as well. Unfortunately the 8X8 monome is a different arrangement and so the code won't help you. I have never done more than a 4:1 multiplex on the TLC9540. What you have to do is to increase the count before you wrap round. At the moment this is done with an increment and an AND with 3 instruction. That will only work for multiplex lines that are powers of two. You need to do something like:- index++; if(index >= 7) index=0;

Grumpy_Mike: Yes it is the library code you have to hack as well. Unfortunately the 8X8 monome is a different arrangement and so the code won't help you. I have never done more than a 4:1 multiplex on the TLC9540. What you have to do is to increase the count before you wrap round. At the moment this is done with an increment and an AND with 3 instruction. That will only work for multiplex lines that are powers of two. You need to do something like:- index++; if(index >= 7) index=0;

Okay, I'm going to try to take it slow and do as much research on my own as to not drive you crazy while I figure this out. :)

A couple of questions to help get me going though..

1) Are there any resources/documentation on multiplexing in general or this driver that I can reference?

2) I'm trying to wrap my head around the LED buffer code. If I read the comment correctly, each LED is 12 bits (0-4095) and each hex number is 1 byte (8 bits). Does this mean that each led 'shares' a byte with its neighbor? Also, what does 'bb', 'bg', 'gg' etc refer to? I figure this is some simple math thing

 * Each value is 12 bits (0-4095).  24 bytes is 192 bits, which is 16 led's * 12 bits each.  Each of the numbers in the array below is 1 byte (0-255).
 *    For example,
 *
 *  | 1st byte | 2nd byte | 3rd byte | 4th byte | 5th byte | 6th byte | ...
 *  | LED 16 value   | LED 15 value  | LED 14 value   | LED 13 value  | ...
 */

// Change this array initilisation to get a diffrent switch on pattern

 static uint8_t ledBuffer[] = {
    // row 1 (botom)
   //     lower LEDs not used        bb    bg    gg    rr    rb    bb    gg    gr    rr    bb    bg    gg     rr    rb    bb      gg    gr    rr
    // row 2
    0x0, 0x0, 0x0, 0x0, 0x0, 0x00, 0x00, 0x00, 0x00, 0x40, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x00,  0x40, 0x00, 0x00,   0x00, 0x04, 0x00, // red row

Also, this doesn't work yet, but do you mean something like this?

    scanindex++;
    if (scanindex >=7) {
        scanindex=0;
    }
    
    scan = (scanindex) & 3; // Increment row count
    bRow = pBuffer[scan];   // point at next row in the bufer precalculated to save time

Also, what does ‘bb’, ‘bg’, ‘gg’ etc refer to?

that is the distribution of bits in the array and so is the answer to your second question
so in the blue (b) is defined by the first two nibbles (4 bits is a nibble two nibbles in a byte) of the first byte and the most significant nibble of the second byte. That second byte has it’s least significant nibble as the most significant nibble of the green brightness with the two lease significant nibbles in the third byte.
so in binary you have
bbbb bbbb - bbbb gggg - gggg gggg
First byte — second byte ---- third byte
This puts together two 12 bit numbers in three bytes.

scan = (scanindex) & 3; // Increment row count

This is the bit you have to remove as that constrains scan to a number between 0 and 3

Some good news and a question.

After tinkering all day I finally ‘got it’ and was able to scale and layout a new ledBuffer. The trouble was that I was only getting the last 4 rows to light up. After some trial and error I found if I comment out the code to get the state of the push-buttons, the rest of the rows turn on as expected. I don’t have buttons on my project. Do you know why that bit of code would interfere?

Here’s the code I commented out:

// read the state of the push buttons
    for(uint8_t i = 0 ; i<multiplexColoums; i++){
        buttonPressed[i][scan] = digitalRead(switchInput[i]);	
    }

Full code:
Modified driver: http://pastebin.com/HQa1mWvb
Modified sketch: http://pastebin.com/zXKJjJ8P

Well done for getting it, my Wife looked at my last post and said "that is just rubbish!"

It looks like you hadn't expanded the keyboard buffer to take the extra non existent keys and that was over writing your LED buffer.