This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TLC5947 strange behavior

Other Parts Discussed in Thread: TLC5947

I'm trying to make LED driver with TLC5947 and found some strange problem when prototyping.

My prototype schematic is very simple - TLC5947 with connections:

  • BLANK and GND - to ground.
  • OUT0, OUT1 and OUT2 - to 3 lines of white Cree CLA1A LEDs, 6+6+7 LEDs in series, other end of LEDS - to +24V
  • IREF - to GND through resistor 2.2k
  • SOUT - not connected
  • SIN, SCLK, XLAT - to MCU (atmega168)
  • VCC - 5v via DC-DC converter (grounds of input and output connected together, +input connected to +24v)

I can make LEDs work, can write brightness and so on, BUT if I touch with my multimeter probe (or steel tweezers) to almost any place on the board, LEDs status changes randomly. It can turn off, can change brightness and so on.

I have tried to connect all input pins (SCLK, SIN and XLAT) to GND, add capacitors to input of DC converter (100uf), add decoupling capacitors (10uF + ceramic 10nF) between GND and VCC line - still same effect - changing of brighness randomly. I gave tried to connect VCC to power of MCU (without DC converter), even replaced power supply 24V - with no luck.

May be you know some adivces for me to get stable working of TLC5947 in my case?

Thanks in advance!

  • Hello Oleg,

    When you observe the problem with using a measurement unit, is the other end of the unit connected to the same ground as the ground of the TLC5947.

    It sounds as if you have a ground issue in your system and normally if there is somewhere ground shift the only possibility to overcome this issue is a new layout. At least in my experience if there is ground shift problem, nothing helps but a layout change.

    Best regards,
    Brigitte

  • No, other end of multimeter is not connected to anything and floating. Same with tweezers (which connected to me :)

    Wow, I impressed that this problem has its own name - 'ground shift'!

    Can you recommend me a new layout or may be some rules I should folow? May be you know a reason WHY 'ground shift' happens? Shift happens.. hmm :)

    This wil help make a new board layout, that will not cause problems, not just handreds of blind tries of layout changing.

    Thanks!

  • Hello Oleg,

    Please find here an example how a board could look like. We cannot guarantee the functionality in every environment, but at least during demonstration of the part it works fine.

    2766.TLC5947_schematic.pdf

    3872.TLC5947_Layout_PCB.pdf

    As you can see, the whole board is flooded with GND and the top and bottom layer GND is connected with vias. At the end you have to minimize the parasitic components that you add with the layout. The board adds parasitic resistance, capacitance and inductance and the smaller these components are, the lower their influence to the performance of the board.

    Here are some general analog layout rules that help to optimize a layout:

    1. Please keep in mind that current always flows in loops and needs GND as return path. This has to be minimized as well.

    2. Check which paths have high transient voltages or currents (dV/dt or dI/dt) and make these tracks as short as possible.

    3. Check which paths have high currents going through and minimize these current loops as well.

    4. Make all analog traces as wide as possible to minimize resistance and inductance.

    5. Place decoupling components as close as possible to the IC and route these with wide traces.

    Best regards,
    Brigitte