Overclocking Broke My Pandora

[Loon]

This is posted mostly just for info just in case anyone else has a similar experience.

I decided to see what CPU speed my Pandora would run at using the included CPU speed slider utility (i.e. no over volting etc.) I knew it was stable at 850MHz as I tried this on day one and have never had a problem running at this speed.

Tried 860MHz and after running Mupen64 for about 10 seconds the Pandora locked up, no problem I thought, just rebooted and then set the CPU speed to 850MHz, the pandora locked up. Oh dear, rebooted tried 750Mhz, the Pandora locked up. Rebooted, then shut down normally, left the machine switched off for an hour or so. Rebooted and after several more tries found that the highest stable speed I could clock to was 575MHz.

Was just about to do a reflash then thought 'it wouldn't do any harm to take the battery out and try again before the reflash'.

This proved to be correct, took out the battery for 10 seconds or so, after reinserting the battery and rebooting found that normal service had been resumed, my Pandora will now clock to 850MHz without any sign of problems.

So, a quick question for you hardware gurus out there :

Why would taking the battery out solve the problem?

My suspicion is that the battery is still being used somehow when the Pandora is switched off, it is also my suspicion that this is not a good thing.

 

[skeezix]

Bad title, you're going to cause heart attacks, when really it is nothing

I do find on occasin that sometimes you just need to yoink the battery to reset 'something' .. pretty rare, but does happen I think

jeff

 

[WizardStan]

My suspicion is that the battery is still being used somehow when the Pandora is switched off, it is also my suspicion that this is not a good thing.

1) Real time clock
2) You never actually shut it down, did you? It always locked up and then you restarted, yes? Even your desktop PC maintains some information if the only thing you've done is hit the reset button.
I'd've been curious to know what happens if you had reset, done a proper shut down, and then tried again. You're not the first person this has happened to, but regrettably I didn't have my debugger cap on last time. If you can recreate the problem (because I haven't been able to) we should do some further experimentation.
One thing that comes to mind is that when it locks up, maybe the current OPP value is getting corrupted and gets reset to 1 preventing you from overclocking, but that's just a wild guess.

 

[Loon]

2) You never actually shut it down, did you? It always locked up and then you restarted, yes?

I did shut it down :

..... Rebooted, then shut down normally, left the machine switched off for an hour or so. Rebooted and after several more tries found that the highest stable speed I could clock to was 575MHz.
.....

Forgot about the RTC, just checked and my Pandora has the wrong time, however the date is correct, how is the date value retained correctly when the battery is removed?

@ skeezix, agree the title is a little dramatic was hoping to get people to read the post in case they had similar issues.

 

[WizardStan]

I did shut it down :

Huh, how'd I miss that?
eah, I dunno then. The OMAP must still be maintaining some state during shut down beyond the RTC. I suggested before that there could be an issue with some People noticing severe drops in capacity during a power down, but not enough to do any proper analysis on.

 

[notaz]

It's all because of power chip, it keeps it's state as long as some kind of power reaches it. So if something goes wrong (this time CPU voltage was set too low somehow) you have to remove battery for a while so that it forgets it's bad state. You have to wait up to a minute with battery removed so that it discharges backup capacitor, that was designed to retain RTC while battery is changed.

 

[HackModford]

Yep... same thing happened to me. Removed the battery and it fixed the problem... phew....

 

[Exophase]

It's all because of power chip, it keeps it's state as long as some kind of power reaches it. So if something goes wrong (this time CPU voltage was set too low somehow) you have to remove battery for a while so that it forgets it's bad state. You have to wait up to a minute with battery removed so that it discharges backup capacitor, that was designed to retain RTC while battery is changed.

So if you explicitly write OPP to the nominal value is there a chance it'll get fixed?

 

[urjaman]

So if you explicitly write OPP to the nominal value is there a chance it'll get fixed?

I think underclocking to OPP2 region (under 250Mhz), then "normal-clocking" to OPP3 area should fix it.
NOTE to self: add setting of OPP3 into the init routine... (now it assumes that the board comes up in OPP3...)

 

[Exophase]

I think underclocking to OPP2 region (under 250Mhz), then "normal-clocking" to OPP3 area should fix it.
NOTE to self: add setting of OPP3 into the init routine... (now it assumes that the board comes up in OPP3...)

Yeah, sounds like good ideas.