Why does making EN = 0 not disable LCD??

[ee_martin]

I have been using a 16x2 LCD display no. MC1602F-SYR for some time and
I think I understand its features. However, I hit a new one today.

I am sending data to the display 4 bits (one nibble) at a time, and I
know I must wait 40us between each transmission. What I am doing is
sending the first nibble, going off to do something else, and then
coming back to send the second nibble. The sequence is:

Set up first nibble
Enable LCD (make EN = 1), wait short time, disable LCD (make EN = 0)
Do something else that takes at least 40us
Set up second nibble
Enable LCD, wait short time, disable LCD

The problem arises when I am doing something else. The LCD is on
a bus so while I am doing something else its inputs may change.
However, so long as EN = 0 I would have thought that would be OK.
But it's not.

I have reduced the problem to this short section of assembler
(PIC18F4550):

        bcf     DB7             ; Set bits DB7-4 to 0000
        bcf     DB6
        bcf     DB5
        bcf     DB4
        bcf     LCD_R_W         ; Make R_W = 0
        bsf     LCD_RS          ; Make R/S = 1

        nop                     ; Short extra wait to ensure data is stable
        bsf     EN_LCD          ; Set enable bit, EN_LCD
        nop                     ; Short extra wait
        nop
        nop
        nop
        nop
        bcf     EN_LCD          ; Clear enable bit, EN_LCD

        call    Delay20us       ; 20us delay.  Half of 40us
        [COLOR="Red"]btg     LCD_R_W         ; Toggle R/W.  EN is 0 at this point (**)[/COLOR]
        nop
        [COLOR="red"]btg     LCD_R_W         ; Toggle R/W again.  EN is still zero (**)[/COLOR]
        nop
        call    Delay20us       ; 20us delay. The other half of 40us
        bcf     DB7             ; Set bits DB7-4 to 0000
        bcf     DB6
        bcf     DB5
        bcf     DB4
        bcf     LCD_R_W         ; Make R/W = 0
        bsf     LCD_RS          ; Make RS = 1

        nop                     ; Short extra wait to ensure data is stable
        bsf     EN_LCD          ; Set enable bit, EN_LCD
        nop                     ; Short extra wait
        nop
        nop
        nop
        nop
        bcf     EN_LCD          ; Clear enable bit, EN_LCD

(**) Problem instruction

The instruction time is 83.33ns and I am sending a special
character I made up in CG RAM address 0.

This fails. The reason is that I change R/W half way through (see (**)).
It also fails if I change RS half way through. I can change DB7-4 without
a problem. If I replace the "btg LCD_R_W" with a "nop" it all works fine.

I would really like to be able to do something else between writing
each of the nibbles because I cannot afford to wait for 40us doing
nothing. Is there any way around this problem?

Martin

Crystalfontz LCD Blog

Looking for additional LCD resources? Check out our LCD blog for the latest developments in LCD technology.

 

[CF Tech]

I do not have a good answer for you. I looked a bit at the 4-bit timing diagram in the HD44780 data sheet:

http://www.crystalfontz.com/backlight/eHD44780U.pdf

They do not show "don't care" periods between the two E pulses (page 33 of above document). This is just an observation, may or may not be meaningful.

Are you meeting the setup and hold times (page 58 of above document)? These controllers were designed long ago, and are not particularly fast.

It could be that the controller designers just never conceived of the lines changing between the two 4-bit pulses.

 

[cosmicvoid]

I have never tried using a display in 4 bit mode on a bus with changes between E pulses, so I am surprised to learn of this. But, then again, since the data sheet shows steady state R/W and RS between E pulses, I am not really surprised that your situation does not work correctly.

The E signal is poorly named, as it does not enable/disable the display. It enables data transfer. If you are "doing other stuff" between E pulses, you will probably have to set a flag when you do the first nibble, so that your "other stuff" can test to see if a LCD write is in progress and not change the control signals. Then reset the flag after the 2nd nibble.