MIOS 4-bit Display Initialization at Startup

[Hawkeye]

Hi there,

I am a bit out of luck these days, last week or so I burned a 4x20 OLED (looking for a good LCD replacement for the MB6582).

Well, SID happens :-), as I always wanted a VFD in the MB6582 anyways, I have ordered the Noritake CU20045-UW5J. Good news is, it will fit the MB6582 quite nicely without the need for a riser card and also I did not burn it just yet :-)...

I´ve connected and wired it up in 4-bit mode to the MB6582, as usual. When turning it on (with the default mb6582.hex firmware) it shows very scrambled characters, looks like it has not been switched to 4-bit mode by the default display initialization routine.

Consulting the datasheet, the first thing that is being sent to the VFD apparently needs to be the "Function Set to 4-bit mode" command. Here is the relevant section:

7.7.1 Function Set Command

20H to 3FH

This instruction initializes the system, and must be the first instruction executed after power-on.

So, I´ve ported the 4-bit lcd display routines from the MIOS basic driver to modules/app_lcd/clcd/app_lcd.inc (which by default contains an eight-bit driver), changed the initialization sequence to really send 0x02 nibbles at first (4-bit mode), have included the changed app_lcd properly, set the DisplayType to 7 (in main.inc) and have verified that this code really is executed (tested that by playing around with the initialization stuff).

With the changed driver, I do get improved character output, but only some characters look ok, other characters are wrong (the "big D" as in MIDIbox is ok).

So, changing to "my" 4-bit driver improved things a little and that indicates that the display can take a late switch to 4-bit-mode... but now I am utterly confused, if a "D" was written properly to the display, doesn´t that mean I am in 4-bit mode? Maybe I´ve screwed up the driver? Could it be a timing issue, probably with the "LCD_Busy" checking? And is there any other "USER_LCD" 4-bit demo code available, so that I can verify things?

Maybe some MIOS-base-level routine sends data to the display before the user initialization and screws things up?

Hmm, sorry for all the questions and the confusion!

Thanks for any help :sheep:

Bye,

Peter

PS: Thorsten, I´ve also got a 256x64 OLED evaluation display lying around, so if you are into a little bit of display code hacking, just tell me and I´ll come over :-)

Edited June 14, 2011 by Hawkeye

[Hawkeye]

Science breakthrough :-)

It appears the "lcd_busy checking" is the problem, plz don´t ask me why... I have removed the checks and replaced them with yet too long (still need to find minimum delays) wait loops for every character printed/command sent.

As the "driver" is for a specific display with a given timing, I hope this approach is viable.

Hawkeye :frantics: (with a vfd in the mb6582, yay! pics soon!)

Edited June 14, 2011 by Hawkeye

[jojjelito]

凄ã„ã§ã™ã­!

That's orsome! é ‘å¼µã£ã¦ãã ã•ã„! (gambatte kudasai). We wan't some iThingy/Androgyny pixx taken with Hipstamatic and some nasty lens/filter approximation!

:drool:

[Hawkeye]

Thx, dude... the "lcd ready" polling works now, too (fixed delays are not good for the performance).

You can now haz Wilbas riser PCBs if wanted :)

I´ve attached the updated driver for anyone interested.

Ah... and the display is... NICE!!! :laugh:

Picz later, gotta sleep nao :sleep:

Best regards,

Peter

app_lcd.inc.txt

[mitcheaton1]

Hi Hawkeye!

I am having the exact same problem with the same lcd... im not too big of a programmer and was wonder if you could post a complied version of your mb6582.hex?

[Hawkeye]

Hi Hawkeye!

I am having the exact same problem with the same lcd... im not too big of a programmer and was wonder if you could post a complied version of your mb6582.hex?

Hi,

but it is a VFD, not a LCD! Are you sure, you have exactly the same Noritake VFD?

Under these cirumstances, i´d fire up tha compila the evening - as the code is pretty badly hacked together, you should install the modded version only on the first core, where the VFD is connected, if you install it on cores 2-4, they will unfortunately hang, but I did not have the time/nerves to debug it. So you´ll also need to install the unmodded standard .hex of the same version to the other cores and not flash the VFD version to all cores. It is a somewhat complicated process - please report back via forum message :-)

Bye,

Peter

Edited January 30, 2012 by Hawkeye