Display distortion problem (h300)

[S3TUP]

Time to time, at the moments the screen gets updated, i get some distortion on the screen like this: http://img54.imageshack.us/img54/2263/iriver3qs.jpg . Sometimes it affects couple of pixels, sometimes it messes whole screen.
I experience such problem with original and RockBox firmwares, so i think it must be a hardware problem. On the other hand, things gone worse by upgrading from some old build to one with increased display refresh speed by rewriting parts of sources in ASM (a month ago), so i think that something in the current firmware makes it worse than in original/old firmware.

So i opened my player and saw that DC/DC PSU big ceramic caps and Phillips UDA1380 were replaced/resoldered (i've got this player as a replacement to my bricked one from Hong Kong). Other solder spots looked ok. Nothing changed when i touched/pushed chips durning playback + i've checked their soldering, so they are ok too.

The things i see on the screen comes from display's memory (overlay?). It happens every 3-5 sec durning mp3 playback, so i looked at PCM/Codec buffers and i figured out that the codec's buffer fills up by boosting CPU frequency every 3-5 seconds, so that is a reason for artifacts on the screen. Looks like the display goes out of sync durning CPU's frequency spikes... But, one more wild thing happened - when the HD tried to spin up, the display got distorted more than at regular playback without HD, so i think the problem may be in DC/DC PSU and these big caps, because by spinning the HD/ increasing CPU speed we are loading the internal power supply, and it produces unstable voltage, that affects the display...

Anyway, i think lowering CPU speed will solve the problem.

I've got couple of questions:
 - Does anyone experience same/similar problem? On other devices?
 - Is it possible to limit CPU's clock to 90-110 MHz?
 - Where the problem comes from?
 - Any suggestions how to repair it?
 - What have been changed a month ago in display handling routine?

(This player is out of warranty for 2 weeks. Moreover, it works fine on original firmware so i can't RMA it).

[LinusN]

The CPU is boosted when refilling the audio buffer from disk, so that may be why it gets distorted when the disk spins up.

We have set the LCD wait states pretty aggressively, but my logic analyzer measurements have shown that they are within the specs in the data sheets. You could try to increase the wait states in firmware/system.c as a test.

[S3TUP]

Thanks! I increased wait states of LCD in 45 and 125mhz modes to 0x00000980 and 0x00002180, and it helped. Actually, i increased flash wait states to be the same as LCD's... I'll try to lower these timings to match my LCD speed.

Thanks for help! Damn, i love RockBox!

[scretus]

hi,
i've the same problem on my h300 (or, at least, i think it's the same problem).

with new versions of rockbox (from the end of december, maybe), my display often has some distortion problems.
in fact i'm using the rockbox 2005-12-25 build, which works great with my display.
what is the cause of this problem? is there any setting i can do to solve this?

thanks!!

ps: rockbox is FANTASTIC!!!

[salival]

*big bump to not have to start a new thread*

For some time now I have had altered values for the CPU wait states, since I experienced the said problems.

With the values I have set I have almost no artifacts left (only during some plugins, like doom and tetrox). My current settings are 0x00003980, ...0980 and ...0580 (max, normal, default)

Since these values are already significantly higher than default and I don't want them to be higher than needed, I have two questions regarding the values of the wait states:
1. are there any limitations or can the values be any hex number ?(i.e. they don't have to be even or the double of the previous value.)
2. Are there any known downsides on upping these values, like limited framerate or responsiveness or are those differences insignificant?

I hope someone can answer at least my first question.

thanks in advance.

[salival]

well, it seems that if your wait-state is too high the problem re-arises. currently I am at 14 (#...3980) and with #...4180 it gets worse again.

I guess I leave it at these values, since the only places I have the issue still are on scrolling text in the wps and the odd plug-in.

[scretus]

with today version the problem is less evident.
i haven't tried other version from december until today.
only some pixels are bad: i only I get small random pixels of various colours on the screen.

[dantheman10110]

alright, i have the same type of problem. Mine seems a little more severe, since it prevents a lot of navagation. The file navagation screen becomes corrupted so badly, i cannot read it. I helped combat the problem by changing the background color to deep blue, since white backgrounds screw it up more. I read the other parts of this topic, but since i'm very new to rockbox, i don't understand any of it. I understand that I need to slow the refresh on the screen, but i have no idea how. Can somebody send me a patched firmware?
EDIT: I'm using today's (20060626) version
It gets lots worse when it's plugged into the wall. On battery it isn't an issue much. Just at the beginnings of songs and when scrolling quickly through the library

[cainey]

I have the same problem and do not know how to fix it.
please help

[salival]

http://www.misticriver.net/showthread.php?t=36004  Post number 10 has the solution. It looks harder than it might be.

[salival]

Does anyone know where to set the LCD wait states in the newer builds? I can find

Code: [Select]

set_cpu_frequency(long frequency) functions in system.c, but I can't find the variables I'm supposed to change anymore.

edit: nvm., got them already (..\firmware\target\coldfire\iriver\system_iriver.c)

[sleepwalker]

I am also experiencing display distortion problems with my H340 and Rockbox Build 2006-12-22.

However, using the posted values 0x00000980 and ...0580 (normal, default) and various max-settings between ...2180 and ...4180 makes things only slightly better. It does not really lead to a satisfactory result. Since I do not really know the technical impact of these settings, maybe someone can help me with a few questions I came across:

• Is it necessary/useful to also alter the normal/default value?
  
  
• Is there any specific ratio between these values which should be kept?
  
  
• Are there allowed minimal or maximal values for each of these settings?
  
  
• Is it necessary to also adapt other values (e.g. Flash wait state as indicated by S3TUP in his second posting above)?
  
  
• Is there any other file than system_iriver.c which needs to be modified?

Any hint is highly appreciated - Thx in advance!

[scretus]

hello,
since I had this problem since today, I decide to solve it with a small patch.

This patch applies to this file
/firmware/target/coldfire/iriver/system-iriver.c
and removes almost any distortion on my iriver h340.

I hope this could help somebody!