Can't Get Nub To Cooperate

[sinoth]

I understand that you have to calibrate the nubs on bootup, and I do... up, center, down, center, left, center, right center... and for a while the mouse seems to do okay. But after some time it always gets worse, and I can't go down, or some other random direction. I really notice it when I try quake2. Within a few minutes I'm always stuck unable to move some direction. The right nub seems to do okay. Do I just have a bum left nub? Is there any way to know for sure, some kind of diagnostic tool or something? Maybe a better app to test with?

 

[paddy]

I noticed this as well ,but to be honest i think it is more to do with quake/hexen games than the nubs in general ,becase if you are only on the desktop they work perfectly all the time but in say quake 3 or hexen 2 they work for a small while then start to get all slow and unresponsive.

 

[SomeGuy99]

I noticed this as well ,but to be honest i think it is more to do with quake/hexen games than the nubs in general ,becase if you are only on the desktop they work perfectly all the time but in say quake 3 or hexen 2 they work for a small while then start to get all slow and unresponsive.

I haven't noticed this in Quake2, but the exact same thing happens when using them for the mouse cursor.

 

[sinoth]

I changed both nubs to "Joystick" and played Quake2 for an hour... didn't run into any issues. I'm beginning to think this is just a driver/software issue. It almost feels like the mouse driver is constantly in calibration mode. For example, try making a bunch of small movements in one corner of the nub then go back to big movements. For me, it screws up the movement until I move at the extremes and it sort of levels out. Seems like it shouldn't go out of whack so quickly, or maybe have a way to turn OFF calibration once it is working fine.

 

[HackModford]

I notice this strange behavior if you move the left nub in a circle a couple of times. But i have had times where all of a sudden I couldn't go down, next I couldn't go up...

 

[sinoth]

I notice this strange behavior if you move the left nub in a circle a couple of times. But i have had times where all of a sudden I couldn't go down, next I couldn't go up...

Yes, exactly! Seems like it confuses the poor driver. I'll have to give quake3 a shot in 'joystick' mode to see if it works better.

 

[paddy]

yes i agree ,moving in a full circle seems to knock it off and then needs to recalibrate ,i think it does seem like they are in constant calibrate mode ,i think the zones need to be set properly and saved but unfortunately i do not think these nubs work like that as i think they work independently on every power on.

Are these nubs not self calibrating via a chip ? could mean a firmware update just for the nubs ,but i could be talking pish now.

 

[sinoth]

Would love to hear from a nub expert as to whether this can be remedied in the future by a firmware flash of the nubs or through drivers.

 

[fusion_power]

I wonder why such "common" problems never mentioned from OPT. Do they have working nubs or simply never did what you did with the nubs? ^^

 

[JayFoxRox]

We already talked about the poor nub firmware on IRC before, at least bobbens, zhasha and me want to fix it because we have worked with ATMegas before. I plan to buy some nubs from ED or whoever can sell them directly for testing and development of a new driver if I run into any problems with the current firmware.
The nubs themself should be flashable using additional hardware, however, this would obviously be a bad solution to fix already existing pandoras.

//Edit: some facts I remember:

• Nub calibration happens in the nub controller which is an ATMega8 (according to pics labelled by the community)
• The CPU can not download or upload the nub calibration, it can only reset it for re-calibration
• The kernel driver receives data from the nub controller and therefore gets the already calibrated signal
  • This problem can't be fixed with a normal kernel upgrade but would require a special flasher for the nubs (if this is even possible hw wise without additional hw - which I doubt because we heard of the nubs being flashed by a special device, besides that the chip might be locked, making it harder / impossible to write new data to it)

Afaik no-one was able to tell us wether they can also be flashed from the OMAP directly. We should wait 'til MWeston arrives here


fusion_power: Common, game controls in the teams vocabulary always seemed to be: "4 Action buttons and Dpad"
We rarely heard anything about the nubs and shoulder buttons before, only "we are making custom nubs" but that's about it. The shoulder buttons were also broken on cebit so I just have the feeling this is something which just didn't interest them too much personally (maybe the nubs did, but no-one used them for a longer period).

 

[fusion_power]

We already talked about the poor nub firmware on IRC before, at least bobbens, zhasha and me want to fix it because we have worked with ATMegas before.

News for me. Never heared of "poor nub firmware" before.
From OPT came the signals that nubs are ready, set and done, final firmware on it and everything works, case closed.
Now we notice, that the nubs are actualy quite useless for serious actions because of an calibration problem, and because these thingies have their own firmware it can't be fixed from the users via a general Pandora OS update? This makes me sad. :unsure:

I hope, at least I will get an Pandora with unpainted Shoulder Buttons that fit better than these ones seen at the current Pics. It looks like only the R Button make problems.
I really wonder, if this weird Case Factory ever did something like the Pandora before. ^^"

 

[WizardStan]

Afaik no-one was able to tell us wether they can also be flashed from the OMAP directly. We should wait 'til MWeston arrives here

I asked that same question a few days ago. MWeston responded that the calibration code was a binary blob and the stuff he wrapped around it was just enough to read the data and send it out. There's no way to reflash the nubs through I2C.

 

[renwils]

[*] The CPU can not download or upload the nub calibration, it can only reset it for re-calibration

Then could this be a kernel issue? I wonder why the nub would recalibrate itself without any request.