Bad Sound Quality


^ It might be related to slight variance in components and/or PCB, I suppose. Each Pandora is unique.
 
mali said:
^ It might be related to slight variance in components and/or PCB, I suppose. Each Pandora is unique.

True.. "But it works on my computer!" :p

I assume the fix has also been tested on working P's, to be sure it doesn't break anything on them? :)
 
Last edited by a moderator:
Qvazar said:
Nein, der Buffer war fein. Da war ein Bug im Code von TI in der Soundinitialisierung.

Der Fix ist natürlich im nächsten Hotfix (erscheint vermutlich in den nächsten Tagen :D ) fix mit drin :)
So TI had a bug in their sound initialization... why didn't it affect all Pandoras then? :blink:
Oh well, I'm a programmer myself, I know how bugs can be "situational" :p Damn bugs!

Hardware changes :) I'm not really sure if the bug is still in their latest code - they sent us the code back with Rev3 :)
 
Last edited:
The bit clock polarity setting was wrong. Most boards work with the wrong polarity, but as you can see some don't.

That configuration came from old sample u-boot code that MWeston sent along with those blue rev1/mk0 prototypes back in summer 2008. The sample code itself was for Mistral evaluation board and (I assume) was adjusted by MWeston for pandora. Later I converted this to board specific code for Nokia/TI ALSA/ASoC OMAP driver (which is what we use now). So that faulty configuration worked on all prototypes (it is probably correct for the EVM, no need to blame TI), and it was so long ago so it wasn't really obvious what the problem is when we first got those reports. Thankfully it's all resolved now :)

BTW that kernel has some additional improvements in underflow handling (which was broken and sometimes caused repeated noise and high CPU usage), MAME was suffering this the most I think.
 
This is great news notaz .. and what is your opinion about realtime kernel extensions to make Audio (synthesis/generation/DAW/editing) even better? Necessary, or not?
 
Ah, thanks for clearing up notaz :)
I thought it was a bug - as I didn't think that a reversed clock polarity would be correct on a board when the SoC which includes audio stuff is the same :D

But I'll never stop learning. Kudos :D
 
Is there anyway to make this fix permanent myself? So I don't have to wait for the hotfix?

nevermind... I seen I could call it autoboot and do it automatically. That'll work for me.
 
HackModford said:
Is there anyway to make this fix permanent myself? So I don't have to wait for the hotfix?

nevermind... I seen I could call it autoboot and do it automatically. That'll work for me.

HotFix should be ready early next week. Already working on it :D
 
Last edited:
EvilDragon said:
HackModford said:
Is there anyway to make this fix permanent myself? So I don't have to wait for the hotfix?

nevermind... I seen I could call it autoboot and do it automatically. That'll work for me.

HotFix should be ready early next week. Already working on it :D

how about a fresh zaxxon image with the hotfixes built in? :)
id like to reflash mine... accidentally did some nand stuff while i thought i was in sd mode, haha xD
 
Last edited by a moderator:
notaz said:
The bit clock polarity setting was wrong. Most boards work with the wrong polarity, but as you can see some don't.

That configuration came from old sample u-boot code that MWeston sent along with those blue rev1/mk0 prototypes back in summer 2008. The sample code itself was for Mistral evaluation board and (I assume) was adjusted by MWeston for pandora. Later I converted this to board specific code for Nokia/TI ALSA/ASoC OMAP driver (which is what we use now). So that faulty configuration worked on all prototypes (it is probably correct for the EVM, no need to blame TI), and it was so long ago so it wasn't really obvious what the problem is when we first got those reports. Thankfully it's all resolved now :)

BTW that kernel has some additional improvements in underflow handling (which was broken and sometimes caused repeated noise and high CPU usage), MAME was suffering this the most I think.

Who is this wizard! Honor him, for he has the great powers!



(thank you very much notaz, you da man)
 
Last edited by a moderator:
senorgomez said:
notaz said:
The bit clock polarity setting was wrong. Most boards work with the wrong polarity, but as you can see some don't.

........

Who is this wizard! Honor him, for he has the great powers!

(thank you very much notaz, you da man)

Could not agree more, sorted out my 'scratchy' sound problem straight away. Glad I held my tongue as was planning to write a long post explaining how you were barking up the wrong tree looking for a software fix when it was obviously a hardware issue.

Just goes to show that there are Geeks and there are Uber Geeks.
 
Last edited by a moderator:
Just another reply to say, Thank you Notaz! I just got my Pandora this week and the sound had completely turned me off from the console because everything I would play sounded horrible, but it's working perfectly now :)
 
Notaz you are the noted star that you are sir!

I was having sound issues purely in Mame, Picodrive and Fuse and nothing in Video or Music playback that was noticeable.

The fix, corrects Mame (there is a slight background crackle but no longer any intermittent sound and garbled noise, imminently playable even on old romsets v77 and v34), Picodrive (similar to Mame previously, now sounds perfect), Fuse (Still an issue of crackling, perhaps the emu itself? No offence to SteveM if I am correct in assuming he is the maker and porter, as it's amazing to re-see these old games, even if the control mapping needs addressing for many games, has anyone sussed this yet?).

Snes9x does lose sound via the fix though and in my case, no amount of re-installs pre-fix, or with fix booted into, address this. No matter, as Snes simply needs a reboot to have sound unlike the others. The Snes9x sound is by default, already very clear, crisp and loud!

I have tested most of the appstore games (all that do not require original PC files, as mine are locked away) and all work fine with the fix applied at boot time. Just Snes9x that takes a hit. AV files also continue to playback fine in my case.

Hu-go does still sounds atrocious but I have found in options>input settings if you enable joystick 1 option, the sound of the game is finally heard beneath the static and the framerate is considerably higher. Joystick 2 setting, stops the games from working. Sorry to go off topic.


Many, many thanks for your work. If I've electronic funds on tap a donation wings your way :)
 
Here's a question: If this scratchy sound bug was fixed by Notaz end of june, how come I received my Panda yesterday with that exact same bug still in it?

Hasn't the fix been incorporated into the codebase? I'm reluctant to apply Notaz' fix until I know whether I do have that problem or maybe something entirely new.
 
^ I'm just guessing, but maybe it has still the old firmware on it. Install the latest image(see official homepage or Tweaks Thread for the link). Btw, congrats :)
 
mali said:
^ I'm just guessing, but maybe it has still the old firmware on it. Install the latest image(see official homepage or Tweaks Thread for the link). Btw, congrats :)
The boot screen reads "release 1 - Zaxxon", which I took to be hotfix 3, unless the naming coventions are a lot less transparent than I thought.

As the hotfix is quite a download for someone on - mostly - GPRS, I'd welcome a confirmation/denial.

Thanx, the panda came just in time for my long visit to Thailand. This might turn out to be interesting, as I'm planning to essentially use it as my only computer over there. I'll keep y'all posted, allthough the lack of decent power management might turn out to be a bit of a bummer (I'm not used to turning off computers anymore).
 
Last edited by a moderator:
^ I have no clue if the boot screen changes with hotfixes and I don't know how to check the version number. Something I know is that hotfix 3 introduced a new system menu entry for mass storage mode. Just check if you have got that entry in your menu. If not you might be running an outdated firmware.
 
Back
Top