Improving the LCD.h library for LCD_IO 1 (Driving the SN74HC595 with 1 wire)

[elreypic]

Hello everyone,

This is my first topic in this forum. Hopefully this can help to many people for this Great Compiler.
I have realized that lcd.h library can be improved with the option 1, which one is for driving the SN75HC595 with 1 pin. Let's procede:

1) Inside the library there is 3 subroutines: Zerobit, Onebit and ResetShiftReg. So the first and second ones look like this:
Sub Zerobit ' Used in 1-wire mode; a "zero" bit is 10us low and minimal 20 us High
SET LCD_CD OFF
wait 10 us ' bit time
SET LCD_CD ON
wait 20 us ' recovery RC time
wait 1 ms
end sub

Sub Onebit ' Used in 1-wire mode; a "one" bit is 1us low and minimal 5 us High
SET LCD_CD OFF
wait 1 us ' bit time
SET LCD_CD ON
wait 5 us ' recovery RC time
wait 1 ms
end sub

So, the bolded instructions are slowing down the bits shifting to the SN74HC595, that means with this one it is delaying the next bit a 1 milisecond, which is not necesary at all. To improve these subroutines we should delete these 1ms delay.

So the subroutines should be:
Sub Zerobit ' Used in 1-wire mode; a "zero" bit is 10us low and minimal 20 us High
SET LCD_CD OFF
wait 10 us ' bit time
SET LCD_CD ON
wait 20 us ' recovery RC time
end sub

Sub Onebit ' Used in 1-wire mode; a "one" bit is 1us low and minimal 5 us High
SET LCD_CD OFF
wait 1 us ' bit time
SET LCD_CD ON
wait 5 us ' recovery RC time
end sub

The ResetShiftReg subroutine looks like:
Sub ResetShiftReg ' Used in 1-wire mode; reset is 350 us low and minimal 300 us high
SET LCD_CD OFF
wait 350 us
SET LCD_CD ON
wait 300 us
wait 1 ms
end sub

So the "wait" bolded instructions could be reduced to make the Reset faster. So, I did a analizis with a oscilloscope and I found out that the best time for both "wait" could be only 225 us. And once again the "wait 1ms" is not necesary, because once the LCD_CD line is high the shiftregister is already reset. We can delete that wait also. The subroutine should be:

Sub ResetShiftReg ' Used in 1-wire mode; reset is 225 us low and minimal 225 us high
SET LCD_CD OFF
wait 225 us
SET LCD_CD ON
wait 225 us
end sub

2) The subroutine called LCD2_NIBBLEOUT can be improved also. It contains something like this:

ResetShiftReg 'generate reset signal for shiftreg and activate monoflop
'character delay settings
#IFDEF LCD_Speed FAST 'Char Rate ~20K
wait fast_us us
'#ENDIF

       '#IFDEF LCD_SPEED MEDIUM  'Char Rate ~15K
            'wait medium_us us
      '#ENDIF

       '#IFDEF LCD_SPEED SLOW  ' Char Rate ~10K
            'wait slow_us us
       '#ENDIF

       '#IFNDEF LCD_SPEED ' default to slow
            'wait slow_us us
       '#ENDIF

All above section doesn't make sense at all, becasue every time a data is shifted into the shiftregister a ResetShiftReg is executed and the line LCD_CD is set high for 225us (300us in the actual library) that is enough time to print a character on the LCD. I have done a test case and it is working good.

I have attached the library and an example, so you can test it in deep just to double check and make sure all the functions are working fine.

Regards,

elreypic.

[elreypic]

I tested it with GCB V0.95, so don't overwrite your actual lcd.h library.

[bed]

Nice find!
Do anybody know the source for the original schematics? Elektor, which year, Month? I would like to read the original Description

[Theo]

Please read this thread:
https://sourceforge.net/p/gcbasic/discussion/579126/thread/da302460/

There you can find an answer for the delays, source etc,etc.

don't forget: a library must be general for all kind of hardware (slow and fast).
Sometimes you need some delay in the software, otherwise the hardware can't follow it.

But building a library for your own purpose; go for it.

[elreypic]

Hi bed,
I took all the information from the on-line help from nest link:
http://gcbasic.sourceforge.net/help/_lcd_io_1.html

Hi Theo,
I agree about how to create the library. This is an initial test. My next step is to implement the changes into the newest version of the compiler. I am testing on PIC microcontrollers only, but my next step is to test it on AVRs also.

Regards,

elreypic.

[elreypic]

Hi everyone,

I have tested the modifications on the newest lcd.h library and GCB V 0.98, so they work perfectly (as expected). I did a test using a PIC16F627A running at 4MHz (internal oscillator) and 16MHz. No issues.

So, I just need to test it with an AVR microcontroller.
Once I make the test I will upload the lcd.h library if you want to test it, but the final release needs to be aproved by the administrator and/or compiler creator.

Regards,

elreypic.

[elreypic]

Hi everyone, its me again.

I have tested my modifications to the lcd.h library in a AVR microcontroller, I mean, I did a test using a Arduino UNO board and it works perfectly fine.
I have attached the lcd.h library (the newest + my modifications) and the exmaples that I did.

I would like to ask to the administrator to review it and if he agree, to be aproved and released as official release.

Thanks a lot and enjoy it.

elreypic.

[Anobium]

Thank you.

If someone can regression test these adaptation then I can include in the release

Anobium

[bed]

Just for the records: the english root article should be here http://www.romanblack.com/shift1.htm
especally the explain how it works is worth to read.
Interesting stuff. When I consider a own DEv Board for Pics I'll take this as an option for Displays