HD44780 compatible,only 8 digit displayed?why?

[blisca]

I'm wasting hours trying to understand why this display based on ks0070
can't go over the 8th of 16 digits

When i worked with 8 bits all was ok,now that i'm using only db4-7
everything looks more difficult

I guess i initialized it in the right why,at least this 8 characters are
correct and some command works.

Normally my problems are very trivial ,i hope even this one would be.

Thanks,
Diego.

 

[BobG]

44780 chip thinks all lcds are 2 lines of 64 chars. Address of 2nd
line starts at ram addr 0x40

 

[Joel Kolstad]

I'm wasting hours trying to understand why this display based on ks0070
can't go over the 8th of 16 digits

People figured out a long time ago that they could save a 2nd column driver IC
(paired with the HD44780) if they built a display that was *electrically* 2x8
even though it was *mechanically* still 1x16.

Assuming you've set up the LCD to automatically increment the cursor's
position, write yourself a loop that just writes "0123456789012345..." to the
display, and have it output, say, 256 characters. You'll probably find
that -- around memory location 40, if I remember correctly -- you'll get a
character at the 9th column on your display (i.e., the 9th column is really
the 1st column of the "second line").

When i worked with 8 bits all was ok,now that i'm using only db4-7
everything looks more difficult

4 bit mode is a little more difficult, but it's definitely worth it for the
pin reduction. Try to get a couple of different documents and study the
initialization in 4 bit mode carefully... what they often don't make clear is
that, even you power up the LCD, it is in 8 bit mode and the fancy
initialization scheme you use to get it into 4 bit mode is really just a
handful of 8 bit commands (where 4 bits are "don't care"), with the last 8 bit
command setting the part into 4 bit mode.

---Joel

 

[mpm]

If it will help you, I can send you 8051 Assembler Code to initialize
and use LCD's in 4-bit mode.
Just let me know....

The only thing I can can think of is make sure you're waiting the
required times between the upfront initialize commands... I don't
think you can check the busy flag while you are switching to the 4-bit
mode.(?) And as others have mentioned, verify the character address
for position #9. It may not be #8+1.

Also, once you are in 4-bit mode, remember that all Commands and Data
are sent high-nibble first. Good luck. We've all been there.... -
mpm

 

[blisca]

Thanks to Bob and Joel

People figured out a long time ago that they could save a 2nd column driver IC
(paired with the HD44780) if they built a display that was *electrically* 2x8
even though it was *mechanically* still 1x16.

do you mean i have to cheat the address ?in wich way?

and why with 8 bits all was ok?


Bob wrote
44780 chip thinks all lcds are 2 lines of 64 chars. Address of 2nd
line starts at ram addr 0x40

Assuming you've set up the LCD to automatically increment the cursor's
position, write yourself a loop that just writes "0123456789012345..." to the
display, and have it output, say, 256 characters. You'll probably find
that -- around memory location 40, if I remember correctly -- you'll get a
character at the 9th column on your display (i.e., the 9th column is really
the 1st column of the "second line").

looptest

movlw '1'
call send_data
movlw '2'
call send_data
movlw '3'
call send_data
movlw '4'
call send_data
movlw '5'
call send_data
movlw '6'
call send_data
movlw '7'
call send_data
goto looptest

strangely this loop gives me 12345671 and nothing else
i hoped that at the 40th send_data a char in the 9th coulmn would be
appeared,is this that you mean?

4 bit mode is a little more difficult, but it's definitely worth it for the
pin reduction. Try to get a couple of different documents and study the
initialization in 4 bit mode carefully... what they often don't make clear is
that, even you power up the LCD, it is in 8 bit mode and the fancy
initialization scheme you use to get it into 4 bit mode is really just a
handful of 8 bit commands (where 4 bits are "don't care"), with the last 8 bit
command setting the part into 4 bit mode

i spent the NIGHT reading documents .. i'm using the 8 bit sequence
suggested by the constructor of the lcd module

i guess i 'm fully in the 4 bit modes because when i want '1' i read the
char '1' on my LCD.Do you think is it enough to be sure?....


still thinking that solution is trivial ......


Thanks again

 

[Joel Kolstad]

do you mean i have to cheat the address ?in wich way?
and why with 8 bits all was ok?

Hmm... if it worked in 8 bit mode, it probably really is electrically 1x16
(and not 2x8). My only guess as to what would cause it not to work, then,
would be if during initialization something went wrong.

looptest
movlw '1'
call send_data ....
goto looptest

strangely this loop gives me 12345671 and nothing else
i hoped that at the 40th send_data a char in the 9th coulmn would be
appeared,is this that you mean?

Yeah, that's what I meant... so as mentioned above, apparently my initial
guess as to the problem was wrong.

i spent the NIGHT reading documents .. i'm using the 8 bit sequence
suggested by the constructor of the lcd module

i guess i 'm fully in the 4 bit modes because when i want '1' i read the
char '1' on my LCD.Do you think is it enough to be sure?....

Yes, I believe that's enough to confirm that you're really in 4 bit mode
(since your characters come out correctly). Hence the only suspect left is
the configuration registers you write to at start-up.

Do you have another module (different brand, but still 44780 compatible)
around to test with? I used a module only than was only "roughly" 44780
compatbile -- it *only* supported 4 bit interface mode (it didn't have pins
for the other 4 bits), it didn't do customer characters, but it did have a
handful of little annuciators that bits that were unused in the 44780
controlled.

---Joel

 

[blisca]

it looks that the problem was due to the checking of the busy flag

the loop (pseudo code)
check db7 -gobackthere if still high

was simply too long,shortening it seems enough to solve the problem

the strange thing is that checking the signal with my logic state
nalyzer( a vintage 8 bit tektronix 308) it looks that once the enable pin is
turned high the DB7 becames low noramlly,but because of the low depth of
the machine i can suppose that in some case it misses it,eventough i used
dummy pulses to trigger it in the zone most likely problematic

Thank you all again,once this Ng is very useful even for a not-so-skilled
guy as i am

i assure you that i post questions only after more than 3 hours of
depression about the problem

 

[Joel Kolstad]

it looks that the problem was due to the checking of the busy flag

Glad you fixed it. I had a product once with a (HD44780 compatible) LCD that
was working just fine when we switched to a (HD44780) Noritake VFD for a nicer
look (this was an obscenely expensive piece of semiconductor inspection
equipment... it had polished stainless steel sides that cost something like
$4k!). It was supposed to be a drop-in replacement, but nooooo... it didn't
work at all. After some debugging, it turned out my LCD code was buggy, and
it was just the fact that the timing of the LCD was different than the VFD's
that had let me get away with it originally. I think it was related to busy
flag checking...

 

[Tom Bruhns]

it looks that the problem was due to the checking of the busy flag

the loop (pseudo code)
check db7 -gobackthere if still high

was simply too long,shortening it seems enough to solve the problem

the strange thing is that checking the signal with my logic state
nalyzer( a vintage 8 bit tektronix 308) it looks that once the enable pin is
turned high the DB7 becames low noramlly,but because of the low depth of
the machine i can suppose that in some case it misses it,eventough i used
dummy pulses to trigger it in the zone most likely problematic

Thank you all again,once this Ng is very useful even for a not-so-skilled
guy as i am

i assure you that i post questions only after more than 3 hours of
depression about the problem

What you see reminds me of a bug I discovered several years ago. I do
not remember exactly all the details now, but I can tell you that if
you are using a PIC processor, and using the code that Microchip
suggested in an ap note about driving LCD numeric displays, it may be
wrong. It went something like, "enable reading status, disable
reading status, read the bit" instead of "enable reading status, read
the bit, disable reading status." Most of the time the first one
worked, because the tri-state bus had enough capacitance to hold the
correct value, but sometimes it failed, and the failure could be
consistent because of other things going on. It could be influenced
by the difference between 8 bit and 4 bit modes. The last time I
checked, quite some time ago, but also quite some time after I
reported the bug to Microchip, it was still in the ap note the same
way. I wonder how many people have used their code and not caught the
error. -- Yes, I think it's still there in AN587, at code lines
419-421. If this isn't the problem you're having, maybe someone else
will see it and be saved some agony. ;-)

Cheers,
Tom

 

[blisca]

What you see reminds me of a bug I discovered several years ago. I do
not remember exactly all the details now, but I can tell you that if
you are using a PIC processor, and using the code that Microchip
suggested in an ap note about driving LCD numeric displays, it may be
wrong. It went something like, "enable reading status, disable
reading status, read the bit" instead of "enable reading status, read
the bit, disable reading status." Most of the time the first one
worked, because the tri-state bus had enough capacitance to hold the
correct value, but sometimes it failed, and the failure could be
consistent because of other things going on. It could be influenced
by the difference between 8 bit and 4 bit modes. The last time I
checked, quite some time ago, but also quite some time after I
reported the bug to Microchip, it was still in the ap note the same
way. I wonder how many people have used their code and not caught the
error. -- Yes, I think it's still there in AN587, at code lines
419-421. If this isn't the problem you're having, maybe someone else
will see it and be saved some agony. ;-)

no i did it by myself,and i did at first the error checking the busy flag
ENA high
read the bit ;error because the data is not yet out on the DB7 of the
LCD
ENA low


instead of

ENA high
ENA low
read the bit


but my last error was only about the loop length