LCD 1602 Keyboard

Hello,

I've been trying to get the LCD1602 keyboard module working on my nanopi2,
but have encountered some challenges. I'm running debian jessie, updated
and upgraded to current. I downloaded the latest matrix.git and ran the
makefile successfully, but when I run the matrix-lcd1602_keypad test
program, I get random letters on the LCD display. I don't get any
initialization errors displayed in the cle when the program runs, but it
appears like the lcd module isn't correctly sync'ed or there is some
communication error. The LCD responds to the keys, but again all I get is
random gibberish on the LCD.

Can someone please help me troubleshoot the problem?

Thanks,

Geoff

Well,

I just ran another package update with apt, and lo and behold the LCD
commenced to working. I'm sorry for the false alarm, the moderator should
feel free to delete this and the above post should they wish.

Geoff

Nevermind, it's stopped working again. For some strange reason, it briefly
worked and then stopped working again after the next reboot. I'm guessing
that the apt-get update was a coincidence and the issue lies elsewhere.

Any idea why this seems to work so intermittently? Faulty LCD module
perhaps?

Thanks,

Geoff

You might get a clue watching the bootup messages.

Try copying the non-functioning bootup messages and then do the update and
copy the new bootup messages and do a diff on them.

Another possibility ... maybe the changes are not being saved to permanent
memory.

I am using matrix on my NanoPi, but have not tried the display.

Thanks for the suggestion Dave. After playing around some more, it appears
that the LCD will work occasionally after running an apt-get call but stop
again the next time the LCD test program is invoked. I don't even reboot
and it ceases working. I'm suspicious that the code doesn't give sufficient
delay after issuing the clear command and characters are being garbled that
way but I don't know what role the apt-get call would play in temporarily
avoiding this. I'm going to try playing with the code a bit and introduce a
short sleep call in the routine. 
Has anyone else seen this issue with the LCD module??

SOLVED:
It turns out that it was a timing issue as suspected. By adding a
usleep(2000) after the LCD clear calls in the test program, the LCD started
displaying consistently correctly. The datasheet shows a 1.52ms duration
for a clear, so something slightly greater than that should work. Reading
other forums and posts, it might make sense to program a check on the busy
flag posted by the LCD as an alternative to the programmed delay, but
that's something I'll investigate later.
Perhaps the test program worked on older CPUs with slower cycles, but it
seems to be challenged on the NanoPi 2. Nonetheless, I'm incredibly
grateful for the matrix library, it's a great starting point.