6x TLC5940 32 rgb leds

There are three voltage sources. The 5v that powers the chips only, the 5v that powers the output circuits, and the arduino 12v. I'm not sure why it fry's some of the chips while it should work. 50x chips just arrived in the mail, I guess I just have to play around with it.

I'm sure i had some problem similar to that when i started using tlc9540's. i solved my problem by running the arduino with the same power supply as the tlc9540 chips. I had the grounds connected and couldn't find any problem. But for some reason when i got everything using the same power supply it when away.I seemed to have problems for ages and i couldn't pin it down for days it just seemed random. It turned out to be the length of the lead to the power supply, all the power leads in fact had to be as short as possible. It works fine now. Ive got 12 tlc9450's running. I did notice that it can seem to work on something like the night rider demo but when they were all on and pulling a lot more currant the arduino would lock up i think that was caused by the power lead lengths.

I have a bunch of 3w rgb leds with 350ma per led and a HengFu with two 20amp 5v outputs. The r3 resistor I used a 1.5ohm 2watt.

Can you post a schematic of that? It seems like you are trying to pull too much power through the chip.

Here is a circuit for one of the boards and pic of a pre-hookup of the arduino. Yeah I had a feeling seperate power supplies might be a problem so I was thinking on setting up a 5v to 12v circuit for the arduino such as this > http://www.eeweb.com/blog/circuit_projects/usb-5v-to-12v-dc-dc-step-up-converter-by-lt1618 As I said before each board works individually, the main difference now is I have long leads from a breakout board for each GSCLK, blank, SCLK, GND, V1, V2. and XLat. So if i power on the chips with V1 and everything is ok, then power the arduino and some of the chips instantly fry. So the only difference is GSCLK, blank, SCLK, Xlat are on but that should not cause some of the chips to fry should it? The only other thing is the second ground that is shared being activated which might be solved with the single power supply.

I suppose I also could run everything off a single 12v/9v high current power supply but the excess voltage will be burned up in the FETS, it might not be bad as I have everything heat sinked pretty good.

I'll get another photo or vid tonight of everything hooked up

ledboard.bmp (424 KB)

You probably need buffers on the control lines. You get them as ic's and they are reasonably cheep. The TLC9540 chips seems to produce a lot of high frequency noise and i think it can have really strange effects when it gets onto other lines, like power. When i switch my set-up on my fm radio upstairs has a loud interference sound all over it. I think it might need shielding. I think i might need some kind of de- coupling cap over the power supply but i cant find anything that makes a difference....(Grumpy_Mike??)

That circuit is rather strange, I can't understand what the thinking behind it is. I am OK with constant current drive bit but not the bit that goes into the TLC chip.

Layout is important as well as supply decoupling. You need to look at things with a scope rather than just use a radio. A scope will show went things are going wrong, a radio just detects noise which will happen even when things are right.

I used the power supply from a computer and everything works great. I was even able to ramp it up to the rainbow code that uses the maximum amount of power that I will be using for hour and a half without problems.

One weird thing, some of the tlc chips get really hot while others stay cool even with the same code pumping through them. I noted that the ones that stay cool actually have internal resistance of about 200k between power and ground pins while the others stay open as I would assume it to be.

One weird thing, some of the tlc chips get really hot while others stay cool even with the same code pumping through them

That doesn't sound right. Have you looked on an oscilloscope to see that the waveforms out of the hot chips are not oscillating?

have internal resistance of about 200k between power and ground pins

You can't make meaningful measurements like this with a meter. If that measurement was made in circuit then you should measure the same because all the power and ground should be common so you will not see any difference even if there was one.

Unfortunately I do not have an oscilloscope at my disposal at this time :frowning:

I made the measure of ~200k between power and ground pins when the chip is completely out of circuit. I'm going to buy some heatsinks for the tlc to combat the issue but I don't think it will be too big of problem since I will only run the rainbow code a fraction of the time. I should have a video of the fiber hooked up within a week or two.

So i'm just diving back into this and doing some simple code to test things but something is very weird. The regular knight rider code works fine but when I'm just trying to turn a set of leds on it seems to shift out and makes it look funky.

Here is the video VIDEO0030.mp4 - YouTube

#include "Tlc5940.h"
int green = 32;
 int blue = 0;
 int red = 64;


 void setup()
{

  Tlc.init();
}


void loop()
{
     Tlc.clear();  
    
      Tlc.set(red + 0, 4095);
       Tlc.set(red + 1, 4095);
        Tlc.set(red + 2, 4095);
        
      Tlc.set(green + 0, 4095);
       Tlc.set(green + 1, 4095);
        Tlc.set(green + 2, 4095);
        
 Tlc.update();
        
    delay(350);
}

This should just make the first three leds yellow which it does but it shouldn't change like it does. Any clues? Maybe I"m missing something simple.

I would say that you have a marginal signal somewhere that is on the edge of working. Without a scope to see what sort of signals you have there is not much chance of tracking it down by guessing. See if you can get one on lone.

Ok, thats what I was kinda thinking. I'm going to test the boards indiviually and then rebuild it all but this time soldering all the wires in the permanent box i'm building for it.

Well I soldered everything together and it all worked. I tried to do some color blending but noticed some of the led were not getting full brightness as they should so I re-soldered the anodes to the 12v rail instead of the 5v and they are at full brightness now which is good but it goes back to shifting as my previous problem. I want to add buffers to the signal lines but not sure what to buy or how to hook them up.

Do I hookup the signal line to the buffer chip and then connect the 6 tlc signal lines to that single output (I'm guessing this is the way), or does the signal line split to 6 buffer chips, one for each tlc? Or is there a buffer chip that has a single input to many buffered outputs?

I have a decoupling capacitors on each tlc, I think they are .1nano (say 104 on them). Should I add some larger ones as well to each board? I'm thinking with the higher current it may be outputting more noise than the small caps can filter.

~Krolo~

I think they are .1nano (say 104 on them)

No they are not.
They are 100000 pF or 0.1uF. The way you read it is 10 followed by 4 zeros.

Do I hookup the signal line to the buffer chip

What buffer chip is this?

I guess that was a general question about buffer chips as i'm not sure what to order.

You need a high side switch or current source buffer like the BTS6143 or the 2981.
You should have a pull up resistor from the TLC output to the +ve. Then connect your buffer input to the TLC output.
However then you no longer have a constant current output and need to do something else to limit the current.

I was referring to buffer the signal lines since they are getting split to 6 tlc's, not the outputs.

In that case a 74HC04 will act as an inverting buffer, two in series will give you a non inverting buffer. As you get 6 in a chip this will buffer 3 signals.
Alternatively a 74LS07 is 6 non inverting buffers with an open collector output, that means a pull up resistor is needed.

Sweet, I've ordered a bunch of those. The regular knight rider code works ok but it still has a shifting problem if I try to power on 32 leds or more. I'm hoping that the buffer chips on the signal lines will help but also wanted to add bigger decoupling capacitors to filter out the lower frequencies that the .1uf caps can't filter out. I've got 10uf, 47uf, and 100uf caps, besides putting them on the power line of the tlc chip is there any other place I want to put them?

Keep the 0.1uFs in as well to handle the high frequency stuff. I would put the extra ones at the power at the end of the line, and maybe another in the middle.