Table LED Matrix, Weeks of debugging, column problems

[mekohler]

1) I successfully constructed a programmable LED matrix using this schematic: http://www.injektilo.com/LEDTable/CIRCUIT.jpg

Which worked prefectly.

2) My next step was to make the real version, where each LED in the original version would be replaced by 6 LED's in parallel. All of this was going to be mounted in a table under some stained glass etc.

3) I was told I would need to modify the schematic a little, by adding transistors to each column and row, like this:
http://www.injektilo.com/LEDTable/NEW%20CIRCUIT.JPG

Note: 4.7k's are actually 2.2k

4) The rows work fine and independant, but when ONE column is programmed to be on, all the others go on as well! Video of it in action:
http://www.injektilo.com/LEDTable/Debug%20xvid.avi

I was told this:

"what happen when you have a voltage difference between the emitter 12volts(column transistors) and 5volts when the 74ls line are not active the transistor base sees this as more negative and turns the transisitor on and is on all the time. What we need on the base of the column transistor is to get it up to 12volts the same as the emitter until the 74ls pulls it to ground and then turns the transisor on. The problem is can we put the base at 12volts which puts 12 on the lines of the 74ls and should only be a max of 5volts to hold it high."


So far, I have tried removing the 2.2ks from the columns but that made the 74ls get really really hot really really fast. So I replaced the fried 74ls with a new one and re-added the 2.2k's. Then I was told to try adding a 100k resistor between the Base and emitter of columns but that didn't work either (I think, my soldering job on that may have been shoddy bcs there is little room to work with)

Any advice? This is my last problem as my table has been constructed. I hope I don't have to start from scratch or something. I spent so much time making the LED bundles and circuits. Help!


Edit: someone suggested: "if you add a 8.2V zeners in series with each of the base resistors for the column PNP transistors, and use a 22k between base and emitter to turn the transistor off, you will be good. This way the transistors will be turned properly off when not activated, even if the voltage difference is 7V. The zener voltage is a bit higher than necessary, to cancel any leak problems with the zener. They tend to start leaking a bit before they turn on. You still hav almost 4 volts to drive the transistor.

Without the zener, all transistors is on all the time and will light the complete row you activate with the row transistors."

[Ioannis]

I have three suggestions.

1. Replace all 2222 transistors and their base resistors with a single ULN2004 chip.

2. Replace 74LS138 with a HC138 type.

3. Add an NPN to drive the PNP transistors since there is current flowing from the emitter-base junction of each PNP transistor. I had this problem too. Put res. from +12 to base of PNP. Then res. from PP base to collector of NPN. Emitter of NPN to ground. Base of NPN with res. to driver.

Ioannis

[mekohler]

I don't think using the ULN2004 chip would require recoding. Would the replacement of the 74LS require the coding for the PIC to be different? Should I use an inverter, and if so, can you recommend one? I have been working on this for so long I am losing my mind haha. Here is the sample code I am using

#include <pic.h>


__CONFIG(INTIO & WDTDIS & PWRTEN & MCLRDIS &
UNPROTECT \
	& UNPROTECT & BORDIS & IESODIS & FCMDIS);

int i, j, k, n, Dlay, CurLetter;

const char Letters[] = {
	
	0b0000001, // A
	0b0000001,
	0b0000001,
	0b0000001,
	0b0000001,
	
	0b0000010,
	0b0000010, // B
	0b0000010,
	0b0000010,
	0b0000010};

main()
{
	PORTA = 0;			// All Bits are Low
	PORTC = 0b000001;	// Start With Top Left
	CMCON0 = 7;			// Turn Off Comparitors
	ANSEL = 0;			// Turn Off ADC
	TRISA = 0b101000;	// RA5/RA3 Inputs
	TRISC = 0;
	
	CurLetter = 0;		// Start With "A"
	
	while (1 == 1)		// Loop Forever
	{
		for (Dlay = 0; Dlay < 25; Dlay++)
		 for (i = 0; i < 5; i++)
		 {
			 j = Letters[(CurLetter * 5) + i];
			 k = (j >> 2) & 0b010000;
			 PORTC = j & 0b111111;
			 PORTA = k + i;
			 for  (n = 0; n < 259; n++);	// 4ms Delay
		}

	CurLetter = (CurLetter +1) % 2; //Increment Letter
	}
}

If you have the time, a quick schem would be great. I am new to all of this, this is my first project. thanks!

[Ioannis]

The ULN chip is inverter! Of course you will have to change the data driven to output of PIC. Just reverse 1's and 0's.

Ioannis

[jukeman]

Does the 74LS138 and 74HC138 see data in different ways. Is it truly a drop in replacement?

[mekohler]

Before I go about trying to implement some of the suggestions, can anyone explain to me why my row's, when activated, only go to .98V? Also the ouput to my decoder from the PIC is at a constant 1.98, 1.98, .98 V (3 ouputs controlling 5 columns). Why is this so? I posted the code in a few posts back.