Release SuperZaxxon Final released!

[porg]

It's probably the charging chip (so many problems with it..), since 3.2 does power saving the current flowing through the chip is lower and probably triggers end-of-charge condition much earlier. The current starts to fall when charge is finished though, so it could also be that 2.6 kernel was severely overcharging the battery and battery monitoring chip got severely miscalibrated.

My charging also only goes until 68%. I guess I'd benefit, if the charging conflict between OS and power-chip would be solved. Right?


I will let my battery outside the case for 1 night now, re-insert, and see how charging behaves then.

 

[elwing]

How can I test for a continuous long lasting uninterrupted internet connection?


(I am not sure wether ping qualifies, as this is not a constant stream but single packets, with pauses inbetween. Or I could wget a large file. But a hint for a serious test with proper log output for debugging is be appreciated.)

how about quake3 with the netgraph enabled? "cl_shownet" that seems to be a fun way to do it

 

[porg]

NETWORKING


@elving: Thanks for your hint but for me violent games are not "fun". But this is another topic. Further net test related hints appreciated.


@peelie: I am now web browsing with Jumanji. So far it did not crash, but I already got a "Connection terminated unexpectedly" message while my first login attempt, and PNDstore failed to download the newest Conky. Conclusion -->


WIFI


on my CC Pandora with SuperZaxxon Final is unstable. Please review.


CHARGING


10 % Starting to charge the first time with SuperZaxxon Final. Using USB charge this time as my charger died yday. LED lights accordingly properly.


68% Charging stops. Shutdown. Taking battery out over night. Reinserting. Reboot.


75% The formerly 68% are now shown as 75%. Charging continues. When I return 2h later it is at 98%. Gave it some more minutes.


100% were indeed reached. LED turned off meanwhile.


Seems taking out your battery resets the charging chip properly. This is a preliminary success. Will report back later some time.

 

[notaz]

For wifi, I've spent months tuning that driver over the timespan of pandora's existence and I don't think anything can be done from software side to make it work better. There were a few changes that did not make it from 2.6 to 3.2 before (caused disconnections for some), but they are now in the latest firmware. Either the chip is crap or it doesn't like something on the board. It works for some, does not for others. Rebirth pandoras have some hardware tweaks that help for some, but not everyone.


Charging situation is similar. The chip stops charging whenever it wants, out of software control (there are some threshold settings, but trying to change them did more harm then good). Charging meter is hardware based too, no software control. Charging acts different between old and new firmware because new firmware uses much less power (maybe 20% of what if used to use) when idle, so that makes charging chip act different as it reacts according to amount of current running through it. Companies like Nokia could do proper R&D so N900 has proper charger chip (they could just use the one we use instead; but that is not without problems too, there is no open driver for N900 charger, that part is closed and not documented). In comparison, Pandora's software development budget was (and is) ZERO, we could barely get base driver functionality before release.


About browsing. You are using desktop browsers on a 256M machine, so a swap file is a MUST - there are various threads about how to set it up, I think even a .pnd exists. Try digging out old Pentium3 with 256M RAM, install latest firefox, disable swapfile and see how far you can go. It won't be much further until it crashes or you see a bluescreen, I'm sure.


Also if you haven't recently tried Chromium try it again, it should be more stable on latest firmware.

 

[White Demon]

I think the charging is still an issue somehow. When I turned my Pandora off 2 nights ago after my previous post, the bettery level reported was at 92%. I turned it on again about 90 minutes ago tonight, and the battery level was 77%. I thought that was odd, how can it deplete by 15% when it's turned off for only 2 days?.


After about half an hour of use (I absolutey LOVE Angry Birds) I noted the battery level (74%) and turned it off again, When I got home half an hour later, I took the battery out for a few minutes, then when I put it back in it was reporting the battery at 66%.


Now, the battery does not deplete by 8% in 30 minutes, even when turned on, so there's simply NO WAY it could when turned off. It seems something is very wrong here somewhere.

 

[WizardStan]

Stop turning it off? It doesn't solve whatever may be wrong, but does avoid the problem for the time being. Just putting it to sleep mode gets you over a week of idle time before it needs to be charged.

 

[peelie]

PAL TV-out with composite seems ok here.

tried it on another tv and colours seem okay!

 

[hmc]

Hi peelie,


maybe a Composite / S-Video issue? What mode, cable (wiring) and kind of input did you use?


Daniel

 

[peelie]

Hi peelie,


maybe a Composite / S-Video issue? What mode, cable (wiring) and kind of input did you use?


Daniel

yeah it was to do with wrong mode i was using s-video input with s-video mode initially giving washed out colours when i switched to composite colours mode came back. this was on a lcd tv screen but going via scart adapter.


got same results going via direct video in to tv as well.


on the other tv i tested am sure i had composite pal mode but maybe i did not!


cheers

 

[spiralofhope]

Just had the slowdown again coming out of suspend - no PNDs running, so a proper suspend. Went straight to the CPU-Speed setting, and found it was all the way back at 125MHz. I have no idea why it's doing this, but it's very common - once in every three or four wakeups, it's setting the CPU speed down to minimum.

I confirm this bug.


--


I have the same usage pattern I had when using previous betas. So this seems to be new to the last beta or the final release.


- The Pandora is on.


- Plug the adapter into it.


- Slide the power lever to put it to sleep.


- When I return, sometimes it's already awake and sometimes it's still asleep.


I'm guessing it wakes up and runs off of AC power after it's done charging. This is not what I want.

 

[hmc]

Maybe related, but maybe not new (I am not sure): When the unit sleeps, and I insert an SD card, the unit comes out of sleep mode. Is this intentional?

 

[Linux-SWAT]

The suspend to ram kills the right nub on my non-wifi OP. A reboot is needed.

 

[A2000]

Maybe related, but maybe not new (I am not sure): When the unit sleeps, and I insert an SD card, the unit comes out of sleep mode. Is this intentional?

What did you expect? If I insert an SD card I want the system to recognize it, react accordingly and thus wake up.


Edit: Spelling corrected (hopefully)

 

[hmc]

I want the SD card to be recognized only after wakign up the unit


I often insert the card only so that it is available later when I continue my usage of the Pandora, not to instantly use it there. And often the Pandora is in sleep mode then. I would prefer that the Pandora does not wake up, but that only when I activate it again, it would run the recognition routine for the newly inserted SD card.


Not sure, if this can work at all, as probably this is interrupt / hotplug driven, and also I am not sure, if my usecase is representative, or if most other users would expect the unit to wake up, when a card is inserted.


Not a big deal for me, but maybe worth some optimization, especially if it's only a slight change.


Daniel

 

[porg]

Regarding SD card wakeup: I prefer hmc's desired behavior.

 

[_wb_]

Wouldn't it be more practical if it wakes up, auto-mounts the card, detects the PNDs etc, and then goes back to sleep? So when you wake it up again, you don't have to wait for all that...

 

[hmc]

sure, this would be even better. And I assume this would even be easier to implement, as no interrupt driven behavior would have to be delayed.

 

[porg]

Yes, that would be the most desireable behavior!


Silent system background preparation without any user interaction,


than going back to sleep, and ready to serve the up-to-date content right after wake-up.

 

[notaz]

That's rather problematic to implement, there is no easy way to find the reason why system woke up. With some heuristics it could probably be done, but there is no one to do the work, as usual.

 

[porg]

@notaz:


I'd say the PND/wakeup/standy issue is minor.


Concerning the major issues WiFi and Charging you wrote, that the best possibly is done, not much room for improvement anymore. The hint to take out the battery for chip-recalibration is important, I think this hint should be added in the initial post the SuperZaxxon Final thread.


And Web browsing (as other RAM-hungry apps) can be improved by using swap and zram. I'll give that a try.


@all: I would appreciate if someone could answer my open questions:

• Can I use any qualitative 5V 1A (or more) charger with the Pandora?
  
• How to run a memory test on the Pandora?
  
• How to test for a continuos network connection / stream (with standard UNIX CLI tools) ?