What is the best signal interface for PC ?

[marconi]

Not read yet what you posted recently but this is the transistor and resistor hardware to interpret the 16 bit data word I mentioned earlier. The pnp transistor is a BC557B.

 

[_q12x_]

You can use NPN as well. Depends what you have more.
Also dont forget each resistor to each matrix board pin !!! You should include resistors for each pin, in your circuit. 1k is good enough.
Also very important, make your circuit as logical and simple to follow as possible, as much intuitive you can make it. You really have to put your neurons to work for this part. I know for sure. It will pay up for you in the near future and for others that are reading your circuit.

 

[marconi]

Yes. I missed off the current limiting resistor in each of the pnp transistor collector - line ( row) paths.

 

[_q12x_]

This led matrix pin number assignment is pretty much like a code.
All that I did was to 'decode' it into a human easy to understand and intuitive way of dialing.
If you remember in WW2 - the german made 'Enigma Machine' ? Pretty cool device. This was pretty similar to that. In my mind at least.
You basically have to link 7 positive (+5V) pins that will represent your lines and 5 negative (0V) pins that will represent your columns. All passing through transistors (whatever they are, doesn't matter) to be able to command each pin separately from your future hardware code.
You know what is interesting and relatively fast? to put buttons and press them all at once to get ...well, only 1 line at a time. pretty much like I did in my code at very low fv. Or simple mechanical switches !!! So --12-- in total.

 

[marconi]

You have it. The column Cn and line Ln (my row Rn) bits in the 16 bit data word at the output of the MCP2017 select the column and line transistor switches as on or off. This is done column by column as you see in my earlier video with all Ln set to logic O so all 7 LEDs in a column light up as a test.

 

[marconi]

q12x good afternoon to Romania. I have attached a short video of the eeprom programmer I made today which demonstrates the first five memory locations loaded with different test patterns for the 8 bit Line (my Row) data words - only 7 bits actually used. The top set of dip switches sets the data word to be stored and the lower set selects the address to store the data word. The small black button generates a pulse to tell the eeprom to store the data as detailed on the dip switches. I will build next some sequential logic counters to address the eeprom to retrieve a particular group of 5 L/R data words to drive the 7 x 5 LED display and produce the image I want. In my case the 16 letters a through to n for the earlier Gray Code project so there will be 16 groups of L/R data words.

The test patterns would display the third attachment - note drawn for 8 lines/rows.

 

[Jim_Mc]

Hi there,

Looks like you're using a single ATMEGA328 for each 16 LEDs? This is completely unnecessary. If the only requirement is to drive multiple LEDs in a sequence, then you could do this using a shift register, like in this example:

Shift Registers - 74HC595 & 74HC165 with Arduino

Learn to use the 74HC595 and 74HC156 shift registers to add extra input and output ports to your Arduino, then use them to build a fancy LED light display!

dronebotworkshop.com

You can get multiple shift registers, or shift registers with more parallel output pins to drive the number of LEDs which you need.

I believe you can even chain them to get more output.

This will be a) cheaper and b) more energy efficient than powering multiple 328p's.

 

[_q12x_]

I reinstalled back my arduino Uno and everything is working as before.
But in my latest program that I made, I can obtain a diagonal line on a Led Dot Matrix as a test program, but its refresh rate can not be lower than 1ms as arduino is restricting. MCP chip is minimum 100kHz (actually has higher values specified) which means 10μs per cycle.
- Do anyone know how to lower arduino fv to μs instead of ms ?
In the movie here I present using it's internal fuction
delay(dly);
which is a ms function, I read about it and I get that. I tried also its other function as well, named
delayMicroseconds(dly);
and the result is exactly as the delay(dly); in ms and not under 1ms, to it's promoted μs.
This is a 2min video where I explain the entire problem, please watch it.
Thank you.

 

[marconi]

5x7 DOT Matrix - https://fontstruct.com/fontstructions/show/847768/5x7_dot_matrix

Today, I programmed the eeprom with the 7 x 5 fonts for the capital letters A to P. Now I will look at your latest video.

My initial reaction to the word 'delay' is that you may be using the wrong command. A delay is just that - it does something later than demanded. This may be your problem because the delay is slowing things down such as data transfers.. eg: The data transfers occur every 10 microseconds (say) and then you delay each transfer by 10milli seconds. The delay will dominate. But as I say I am not at all versed in the arduino you are using.

For most people a refresh rate of 50 to 100Hz fools the eye and brain to see a flicker free image as in a conventional TV picture - but modern TV's may do something different these days.

You could average over time the output in software to remove the flicker to make the output virtually constant until there is a change in output. Write a new sub-routine which takes your current sub-routines output as its input and produces an output which is flicker-free.

 

[_q12x_]

For most people a refresh rate of 50 to 100Hz fools the eye and brain to see a flicker free image as in a conventional TV picture - but modern TV's may do something different these days.

100Hz not 100kHz! - VERY Interesting that you mention this! This is helpful. So my 1ms delay in code might not be a real 1ms at all. The speed in reality is probably somewhere at 20Hz now that I think about it. Hmmmm... very intriguing. They definitely have some issues on their I2C pins on that arduino board.
I just google a bit and I find and confirm with: "most humans can't easily distinguish individual frames at 30 hz."
The visible fast flickering is under 30Hz !
So the SCL pin from arduino is simply under 30Hz and not even close to 1ms which is 1000Hz. Hmmm, very interesting.

 

[marconi]

Exactly.
Here is the actual program for 'letter A' that I made and presented in the movie.
You may not understand everything but read the entire program anyway and you will get something out of it. I've also put notes what is line and what is column.
(it was imperative to switch to Input everything after each step, with my custom Reset(); method, otherwise it was liting up the entire next line or column)
Remember a line is HIGH because is +5V and a column is LOW because is 0V. So we lit a single line putting +5V on line01(for ex) and 0v on each desired column.

#include "Arduino.h"                    //core library that contains all of the Arduino’s built-in functions (from 2022)
#include <Adafruit_MCP23X17.h>          //this is Adafruit-MCP23017-Arduino-Library version 2.1.0
 
Adafruit_MCP23X17 mcp1; //Instantiate mcp1 object
 
void setup()
{
  Serial.begin(19200);
  mcp1.begin_I2C(0x20);
  for(int i = 0; i <= 15; i++)
  {
    mcp1.pinMode(i, INPUT);
  }
  for(int i = 0; i <=15; i++)
  {
    mcp1.digitalWrite(i, LOW);
  }


//LED MATRIX DISPLAY - Single LED - one by one
/*  LOGIC:
//Columns----------------------------------------
//  mcp1.digitalWrite(1, LOW); //A1=1=column A
//  mcp1.digitalWrite(10, LOW);//B1=10=column B

//  mcp1.digitalWrite(3, LOW); //A3=3=column C middle
//  mcp1.digitalWrite(11, LOW);//B2=11=column C middle

//  mcp1.digitalWrite(4, LOW); //A4=4=column D
//  mcp1.digitalWrite(13, LOW);//B6=13=column E

//Lines----------------------------------------
//  mcp1.digitalWrite(5, HIGH);//Line1
//  mcp1.digitalWrite(0, HIGH);//Line2
//  mcp1.digitalWrite(6, HIGH);//Line3

//  mcp1.digitalWrite(2, HIGH);//Line4
//  mcp1.digitalWrite(12, HIGH);//Line4

// mcp1.digitalWrite(8, HIGH);//Line5
//  mcp1.digitalWrite(14, HIGH);//Line6
//  mcp1.digitalWrite(9, HIGH);//Line7
*/

}

void Reset()
{
 for(int i = 0; i <=15; i++)
 {
   mcp1.pinMode(i, INPUT);
   mcp1.digitalWrite(i, LOW);
 }
}


int dly = 1;
void chr_A()
{
  //line1
 mcp1.pinMode(5, OUTPUT);   //line
 mcp1.digitalWrite(5, HIGH);
 mcp1.pinMode(10, OUTPUT);   //column
 mcp1.digitalWrite(10, LOW);
 mcp1.pinMode(3, OUTPUT);   //column
 mcp1.digitalWrite(3, LOW);
 mcp1.pinMode(4, OUTPUT);   //column
 mcp1.digitalWrite(4, LOW);
 delay(dly);
 Reset();
 
  //line2
 mcp1.pinMode(0, OUTPUT);   //line
 mcp1.digitalWrite(0, HIGH);
 mcp1.pinMode(1, OUTPUT);   //column
 mcp1.digitalWrite(1, LOW);
 mcp1.pinMode(13, OUTPUT);   //column
 mcp1.digitalWrite(13, LOW);
 delay(dly);
 Reset();

  //line3
 mcp1.pinMode(6, OUTPUT);   //line
 mcp1.digitalWrite(6, HIGH);
 mcp1.pinMode(1, OUTPUT);   //column
 mcp1.digitalWrite(1, LOW);
 mcp1.pinMode(13, OUTPUT);   //column
 mcp1.digitalWrite(13, LOW);
 delay(dly);
 Reset();

  //line4-midle
 mcp1.pinMode(2, OUTPUT);   //line
 mcp1.digitalWrite(2, HIGH);
 mcp1.pinMode(1, OUTPUT);   //column
 mcp1.digitalWrite(1, LOW);
 mcp1.pinMode(10, OUTPUT);   //column
 mcp1.digitalWrite(10, LOW);
 mcp1.pinMode(3, OUTPUT);   //column
 mcp1.digitalWrite(3, LOW);
 mcp1.pinMode(4, OUTPUT);   //column
 mcp1.digitalWrite(4, LOW);
 mcp1.pinMode(13, OUTPUT);   //column
 mcp1.digitalWrite(13, LOW);
 delay(dly);
 Reset();

  //line5
 mcp1.pinMode(8, OUTPUT);   //line
 mcp1.digitalWrite(8, HIGH);
 mcp1.pinMode(1, OUTPUT);   //column
 mcp1.digitalWrite(1, LOW);
 mcp1.pinMode(13, OUTPUT);   //column
 mcp1.digitalWrite(13, LOW);
 delay(dly);
 Reset();

  //line6
 mcp1.pinMode(14, OUTPUT);   //line
 mcp1.digitalWrite(14, HIGH);
 mcp1.pinMode(1, OUTPUT);   //column
 mcp1.digitalWrite(1, LOW);
 mcp1.pinMode(13, OUTPUT);   //column
 mcp1.digitalWrite(13, LOW);
 delay(dly);
 Reset();

  //line7
 mcp1.pinMode(9, OUTPUT);   //line
 mcp1.digitalWrite(9, HIGH);
 mcp1.pinMode(1, OUTPUT);   //column
 mcp1.digitalWrite(1, LOW);
 mcp1.pinMode(13, OUTPUT);   //column
 mcp1.digitalWrite(13, LOW);
 delay(dly);
 Reset();
 
}



void loop()
{
 chr_A();
}

I counted 7 one millisecond delays, one after the other in this sub-routine. I suppose for now that from the start to end of this sub-routine takes of the order 7mS. I cannot quickly work out if 7mS is to generate a single column of 7 LEDs, a single line of 5 LEDs or all 35 LEDs. Please check my analysis because I am pretty sure I have not completely understood your code. You understand how you have written the code so ought to be able to arrive at figure of about 30Hz or lower which you mentioned earlier.

If 7mS is for all 35 LEDs then the refresh rate is 1/0.007 = 140Hz.
If 7mS is for a column of 7 LEDs, and there are 5 columns the refresh rate is 140/5 = 28Hz.
If 7mS is for a line of 5 LEDs and their are 7 lines the refresh rate is 140/7 = 20Hz

Why do have delay command at all in this sub routine?

 

[marconi]

My progress so far. The eeprom addressed by sequential logic to find the five 7 bit data words to form the font for N. Refresh rate is about 200Hz ie A complete N two hundred times every second. Clock is 1kHz and 1000/5 = 200Hz where 5 is number of columns. And letter B too.

 

[_q12x_]

Congratulations mister @marconi
Now put a movie with it and your comments !
I managed to optimize my code and now is less flicker that original. Its still a vibration but really almost not visible at all. I also installed a second mcp23017 library and all last night and until now today, I stay on it, but it has some very-very weird logic inside that I could only partially descypher. So no complete descypher at the moment of this second library that is promising greater potential, since is more close to a PIC program and also closer to our simplified schematic diagram. I literally hammer it last night, im very tired from it. But I learned some interesting stuff doing it, very technical and very in the programming domain. Just putting on the table what I did.

 

[_q12x_]

 

[_q12x_]

Good morning mister @marconi ,
Here is something interesting... Im not sure how much may have been influenced my program testing so far, I will have to make a more extensive research on this particular detail, but in time. For the momment, with the programs I run already, I didnt really observed much of a difference to be honest. But it crossed my mind that this little bugger might be important in some special cases, special programs. Time will tell.
So here it is - full view:

The blue lines represent the Ground or 0V. I marked with a square the 0V marking next to the metalic pad from which the wire is going beneath the cardboard to that long track on which, each negative pin for all the LEDs are connected. So what I have there is a pull down connection through each 12k resistor put in series to each LED to ground.

This problem I realize it yesterday, when I was programming and was meditating to the fact those pins from the IC are not really floating or inAir. I then realized they are actually pulled to ground through each led. And today, I mounted that little switch only between this rail for all the leds and 0V. I tested a couple of normal programs but no visible difference. Ill have to test some in depth programs that I have 'somewhere' that really needs some floating pins. I really hope it will be a difference this time. If not, then, daaaaeeem.

 

[marconi]

https://www.matrixtsl.com/mmforums/viewtopic.php?t=8425

I did some googling and found this. Does it help?

 

[marconi]

q12x. You asked for a video. The letters A to O represent the 16 states of the 4 bit Gray code. The 4 white leds top right are the 4 bit Gray Code. The two 7 segments are numbered representation for the first 12 states of the Gray Code - last 4 states are not decoded because I ran out of space on my home made ROM on the brown veroboard. The eeprom is on the bottom right board on the far right. Enjoy

 

[marconi]

May I suggest a way ahead? It is a seven step approach.

1. Design a way to repetitively scan the 7 x 5 column by column. I showed this in my first video.

2. Now introduce a way to display one pattern of column data say 1010101 which scrolls from left to write on the 7 x 5.

3. Now form a 7 x 5 data table in software.

4. Now write a sub routine to read the data table column by column . Load the data into a 7 bit register.

5. Now add some code to indicate the column number and write this into a register.

6. In hardware make a circuit as I showed you earlier which takes as its input the data in the column and column number at paras 4 and 5.

7. Now load table with chequerboard pattern of bits and check 7 x 5 displays it correctly.

You now have way of displaying anything you write into data table. And this table can change over time if you wish.

 

[marconi]

Good morning mister @marconi ,
Here is something interesting... Im not sure how much may have been influenced my program testing so far, I will have to make a more extensive research on this particular detail, but in time. For the momment, with the programs I run already, I didnt really observed much of a difference to be honest. But it crossed my mind that this little bugger might be important in some special cases, special programs. Time will tell.
So here it is - full view:
View attachment 100791
The blue lines represent the Ground or 0V. I marked with a square the 0V marking next to the metalic pad from which the wire is going beneath the cardboard to that long track on which, each negative pin for all the LEDs are connected. So what I have there is a pull down connection through each 12k resistor put in series to each LED to ground.
View attachment 100792
This problem I realize it yesterday, when I was programming and was meditating to the fact those pins from the IC are not really floating or inAir. I then realized they are actually pulled to ground through each led. And today, I mounted that little switch only between this rail for all the leds and 0V. I tested a couple of normal programs but no visible difference. Ill have to test some in depth programs that I have 'somewhere' that really needs some floating pins. I really hope it will be a difference this time. If not, then, daaaaeeem.

I do not immediately see the connection between the potential of the outputs and its impact on your time problem.

 

[_q12x_]

I do not immediately see the connection between the potential of the outputs and its impact on your time problem.

First off all, congratulation on the movie and showing the full working project. Very fascinating !!!
I still have to try it your way someday, in the future.
-to your question now: - Well, when a pin of the IC is floating , it is not 0V nor 5V. Right? But by having all those leds and resistors to 0V, I efectively put to 0V every IO pin of the chip. THus, not leaving anything really floating. This is important in code, when I am switching from Outputs to Inputs. By switching to Inputs is the programming way of 'floating' the Output. By inversing it. Its a trick learned when I was programming PIC's MCU's. But... having all these pins to 0V ---anyway--- , it doesnt matter if I was switching to Input or to Output in code, because the state was 0V anyway. So the speed was not affected but the entire logic of the code was affected, which is more important and more drastic. I hope is more clear now what I did there.