Nub issue


MrConfusion

Very Active Member
Joined
Apr 11, 2013
Messages
331
Hello,

So, the Pandora arrived yesturday. I haven't had time to really check it out completely quite yet, but, as expected: Looks and feels awesome. I'll only get SD cards for it today (maybe today, depending on, once again, available time :-( ), so there hasn't been much chance to use it yet...

However, quick testing shows the right nub definitely does not work.

In Pandora input tester the left nub gives OK results, right one never moves from 0-coordinates. I've tried pushing it to extreme positions slowly, quickly, rotating etc. to get it to calibrate.

But movement doesn't help.

During the initialization phase I tried to use it as a mouse, but it didn't work. At the time I figured it just wasn't configured for mousing yet. Before trying the input tester, I tried to set it up as a mouse and at that time it did give some random response, which, however, just caused the pointer to float to a corner of the screen by itself slowly and then stay there. After that I restarted the Pandora and since then: nothing.

I tried to leave the unit without the battery overnight to see if that would reset something, but that did not help.

So, any suggestions or ideas?

Addition:

Suggestions and ideas can also consist of stuff which involves opening the Pandora and soldering... assuming that is OK without voiding warranty in Pandora's case.

Basically I appreciate any ideas that might save me time looking for the problem and I'm ready to diagnose this myself as far as possible before sending it back, because the right nub probably won't get that much use at least in the beginning.
 
Last edited by a moderator:
Seems to me like you've tried everything logical to get the right nub working.

Not sure what tests are done before shipping Pandoras but it could be worth checking with ED to see if Nubs are tested before despatch.

In the mean time you could use the left nub as a mouse and use ABMouse to use the A&B buttons as mouse buttons.
 
Yep, they are being tested (the tester doesn't even allow to flash the OS when something in the test failed), but it could very well be a bad soldering blod which came triggered when the unit shipped.

Of course, we will fix that for free, but if you like, you can also try to fix it yourself (if you're a bit experienced with soldering)
 
Thanks ED, I hoped as much :).

I'm heating up the soldering station now and in a minute we'll see. There's a "suspect" among the joints when looking through the loop. I just hope the kids have finally gone to sleep now, or this will take considerably longer than "a minute" (the buggers keep disturbing me just when I least need it ;-).

Edit: It wasn't the "cold joint looking"-suspect... I'll try the others too. Any test points (beep) for these?

Edit 2: No luck by simply adding a bit of flux and "tipping" the joints (i.e. not adding solder more than what's on the tip of the iron... just reheating what's already there... hmmm... soldering is one of the things I definitely have trouble expressing myself in english ;-).

Unfortunately I don't have more time for this tonight :-( I need to get some sleep for work tomorrow. Darn.

If there's testpads for these in the motherboard, I could try checking them before resuming - no point in trying to solder things that are already OK. At this time of night my eyes seem incapable of even trying to decipher the mess of tracks that makes a Pandora :).
 
Last edited by a moderator:
It could also be U22 on the backside, but that one is harder to resolder (not impossible if you got proper equipment though).
 
So, U22 is the square Atmel chip? Right... I can try that one over the weekend, when I can concentrate in this for a longer while.

Anything I've build myself I've always started from these annoying surface mounts, so it's been super easy to just make a great big blobby mess and then suck it all up with wick. I can't do that so "effectively" (i.e. messily ;-) here, it will take a while... But I have successfully done this before to one of my laptops.
 
Yep, U22 is the firmware for the right nub, so if this gets lose, then you'll have no movement at all.
 
OK, I've now soldered 3 sides of the chip and because I'm using both a tabletop magnifying glass and a headmount magnifying glass my eyes are basically starting to look like Major Asshole's from the Spaceballs :

https://www.youtube.com/embed/XGdjKvivJA8?feature=oembed

:rolleyes:

I think I'll do the last side of the square tomorrow.

I just tried the Pandora, these three sides have not helped yet: the nub still doesn't respond.

The last side is going to be the toughest one, it's the one closest to the SD slot. Very tight on space. Like I guessed, using wick is a pain in the butt even in the less populated sides, so I try to avoid shorting pins in the first place. And that makes this even slower going...

Anyway, I've located another issue in the PCB, in another thread I started; the charger LED is missing. What specs for the LED?
 
Missing charging LEDs? I'd assume seeing if it charges would be one of the primary things checked for during inspection.. There  are some extra LED pads left  un-populated according to the hardware hacking guide.  

 
 
Last edited by a moderator:
Missing charging LEDs? I'd assume seeing if it charges would be one of the primary things checked for during inspection.. There  are some extra LED pads left  un-populated according to the hardware hacking guide
I know, but this is a CC unit, which I take to mean "if not refurbished, then at the very least, 'salvaged'". If that were not the case I'd have been slightly annoyed by now to say the least :mellow: .

Thank's for the link to the guide, I did not realize something like this exists. The missing LED really is LED3. There is solder on its pads (whereas 1 and 2 are clean and golden) and to my eyes it doesn't look like the LED would have been there and then broken out, but been soldered out (a darker layer of flux is visible). It's weird, but like I said; "salvaged goods"... I didn't expect the PCB to look like it came from the GC product ilne...

Addition: No nub :-(.

I soldered some pins of the last so-far-un-resoldered side of the chip, though this time (starting to lose hope, I guess, plus I now have a headache...) I skipped pins without visible traces (if there are traces leading directly underneath the chip, ED, let me know and I'll still try them too later...). I also skipped the pins I could easily test with a multimeter (i.e. tested by beeping between the pin and that annoying resistor that was on my way on this side of the chip).

I do have to say I now respect both ED (and Askarus? if I understand correctly) even more than I did before this ordeal... my head is really aching from staring through the magnifying glass, I had to steady the iron with both hands to be accurate with it (though, admittedly, I do suffer from shaky hands anyway...) and even if ED were to have a fancy top notch microscope for doing this it must still be horribly fatiguing stuff to do for hours on your "theoretically free evenings", with people asking "where's my Pandora" all the time adding some nice pressure to the experience!-(

Of course, the CC units are a bit of a special case, but that's small a very small relief...
 
Last edited by a moderator:
Hmmm, but all CC units are fully tested here, and it includes an LED test as well. Was SuperZaxxon 1.54 preinstalled? If not, then maybe we put a wrong board into the system.


If yes... maybe it fell off during transport...?


I wonder if that Atmel died, because the Atmel or Nub itself are basically the only two things that can cause a completely nonworking nub...
 
Hmmm, but all CC units are fully tested here, and it includes an LED test as well. Was SuperZaxxon 1.54 preinstalled? If not, then maybe we put a wrong board into the system.


If yes... maybe it fell off during transport...?


I wonder if that Atmel died, because the Atmel or Nub itself are basically the only two things that can cause a completely nonworking nub...
I PM'd you, because I found more weird stuff on the PCB and already figured my unit is a special monday morning edition :rolleyes: .

Discussing this here isn't very good press and obviously other forum members cannot help me fix it without seeing the PCB. So I figured I'll continue with PMs.

But:

Yes, it was 1.55 when I got it. Once you see the PM you'll probably be even more convinced that it is a wrong board. Either that or someone has done you a selfish "mailorder 14 days money back, no questions asked"-return and hasn't bothered to say there's more there than a change of mind behind the return.

If you don't give such returns a "full drill", then things like this can happen (i.e. I've seen exactly this happen at local shops...).

I don't think the LED could have been in the case. I opened the unit very carefully the first time and on white surface, because "extra parts" dropping out is often the reason for nonfunctioning features and if you loose such extra parts then its automatically pretty much game over for any fixes... It isn't entirely trivial to source something like a surface mount LED or resistor... you don't go into a shop and say "I'd like 1 please" :D .

One more question here (answer might be of use to someone in the future):

I wonder if that Atmel died, because the Atmel or Nub itself are basically the only two things that can cause a completely nonworking nub...
I beeped all tracks from the nub to the Atmel and they check fine (i.e. exactly like they do in the other nub). Further work on the Atmel would be far too time consuming for me.

(Yes, I have an oscilloscope, no, I don't want to use half a day trying to connect it reliably... and especially I don't want to boot the Pandora while working on it, that is, so to speak, going too far with fixing a nub :D )

However, is there anything meaningfull I could measure from the nub itself? I noticed there are two pins that do not go into the Atmel, whether they're just ground or NC, no idea... but basically, I don't know what the nub really is, for example, if it's just a fancy looking potentiometer that the Atmel is reading and turning to bits, is there something that can be measured to verify that the nub itself works?

(I apologize for being so interested in the inner workings of the Pandora, I can't help myself ;) . This is almost more fun than playing the emulators  :rolleyes: . )
 
Last edited by a moderator:
I've a somewhat similar issue on the left nub of my Pandora, it's just become a lot more obvious since I've gotten into using the new Mupen2.0. Down-Right on my Pandora is only half as effective as all other direction. On the nub tester that section produces a diagonal rather than a quadrant of a circle. I can still use that direction, but with games like Mario 64 I can no longer run in that direction, only walk slowly. I know I coul go into the config and change the analogue sentivity settings for the emulator, as I think they were much higher on previous version, so this wasn't as noticeable.

I've tried the nub dance numerous times, resetting the nub etc. I've no knowledge of soldering, so I can't take that approach to see if that would fic the problem. Is my only solution trying to get ED to replace that nub?
 
I've tried the nub dance numerous times, resetting the nub etc. I've no knowledge of soldering, so I can't take that approach to see if that would fic the problem. Is my only solution trying to get ED to replace that nub?
If the problem is the nub itself, there's nothing much you can do by soldering either. You cannot open it to clean it or anything, even if you could detach it... at least I think you can't.

It it isn't the nub, then it's the chip reading the nub and converting the values coming from it to bits. And there's totally nothing you can do about that.

So, yes, that is your only solution as far as I can see.
 
Usually, it gets better when you play a few weeks with it. Often you can fix that as well by heavily moving the nub around in circles and up/down/left/right very fast (you may need to reset it afterwards, as that can fully confuse the calibration).
 
Thanks for the info ED, I'll try those suggestions on my one anyway, but I've had this upgraded unit for several months so I would have assumed the nubs would be well bedded in by now! The odd thing is, sometimes it seems to work fine, then others it's really noticeable that I don't have the full range of motion.
 
Did you end up RMA'ing your unit in the end yourself?
Not yet. I did some fixes myself and some of them worked.

I still want to keep tweaking for a while, who knows, maybe I'll manage to fix the remaining things too.

I've been a bit time constrained in my efforts :-(...

The nub2 still does not work, but I still intend to take one last look at this problem later after thinking about this for a while and maybe after asking for a bit more help.

I might add that my nub2 does not respond at all, so it is a different problem from yours.

Thanks for the info ED, I'll try those suggestions on my one anyway, but I've had this upgraded unit for several months so I would have assumed the nubs would be well bedded in by now! The odd thing is, sometimes it seems to work fine, then others it's really noticeable that I don't have the full range of motion.
Whenever it occurs, pause the game and try moving the nub in extremeties and large circles (this is the same thing ED said above, but said differently). Whether that causes anything to move on screen should not matter at all.

That would fix calibration related issues in the chip, because it gets all ranges of voltages from the nub then.

If the nub even sometimes works OK that would imply the nub itself has no "range" or "dirt" problems as such and this implies a "changing" issue, namely calibration.
 
Last edited by a moderator:
Todays update, I've tried the suggestions listed above, I even removed the batter from my Pandora for 30min before use to see if that would reset calibration data. When I restarted my Pandora I seemed to have full range of movement again. I quickly loaded NubNub, as that has a better calibration tester in the pause menu than the system one. for the first few circles it was fine, then quickly returned to the usual reduced range. No amount of frantic waggling and resetting of the nubs from the nub configurator seemed to alleviate this.

Would completely reflashing the Pandora help me in this circumstance? I'm assuming there's no way to reflash the nubs themselves.... :wacko:
 
Todays update, I've tried the suggestions listed above, I even removed the batter from my Pandora for 30min before use to see if that would reset calibration data. When I restarted my Pandora I seemed to have full range of movement again. I quickly loaded NubNub, as that has a better calibration tester in the pause menu than the system one. for the first few circles it was fine, then quickly returned to the usual reduced range. No amount of frantic waggling and resetting of the nubs from the nub configurator seemed to alleviate this.

Would completely reflashing the Pandora help me in this circumstance? I'm assuming there's no way to reflash the nubs themselves.... :wacko:
Are you sure it's not "reduced" range, but actually "improved in some directions, stayed the same in others, thus you end up with an egg-shaped pattern"?

Anyway, I downloaded NubNub too (<rant>After looking for it for longer than it downloaded...grumble... why isn't that storethingy alphabetically sorted and tapping on the title doesn't sort it either... </rant>), but I can't find that calibration option.

Since the input tester gives numbers, I think it is quite nice, actually.

Looking from the input tester, it seems the values given by the nubtester range from -32 to 32 on each axis. Directly horizontally or vertically I get that easily.

Now to get a circle you should get 22,22 from the corners, 29,14 "halfway to a corner".

I get exactly that easily from the bottom half of the circle of nub1, but practically not at all from top half and the left upper half of the circle starts to look more like the edge of a triangle than a quarter of a circle.

It doesn't sound like an easy task to adjust a suitable "1 size fits all" curve for these things, though...
 
Back
Top