First keys not working

[SchattengestaIt 559]

After using my used Pro1 for over half a year I am getting my first hardware issues. The first keys start to not work when pressed normally, I have to press them quite hard to get the key. It started with the 'c' but I now have four or five keys that often do not get registered when pressed. The 'f' and 'c' being the most obvious keys.

Is there any kind of replacement or fix for hardware keyboard issues? Also, am I the first one encountering this problem? I haven't found any post in the forum about this (or I failed at searching).

Edited August 26, 2022 by SchattengestaIt

[EskeRahn 5,460]

I have used min four three years now, and a few keys are becoming 'unstable', that is 0, 1 or 2 characters when pressed.

As the device is going to be retired as a test/backup device, it is not a huge issue for me, hopping to get my Pro1X soon.

If we can get it as a spare part it should not be that hard to swap, with the help of some heat.

Try taking contact with support.

[raymo 177]

Same problem here, but I wonder me if it's 100% hardware ?
When it happens (trying aaa eee bbb aaa eee  bbb sequence with normal press) dmesg says (when missing a letter) :
"aw9523b_irq_handler: enter" two or three times and doesn't go with "key press", which is related (if I understand correctly) to i2c GPIO expander

But yes when I press very hard it doesn't happen, either when I write slooooowly, wich could be the same here...
Not sure at this point but really annoying

Edited August 26, 2022 by raymo
clarification

[SchattengestaIt 559]

@raymo That's good to know. I had many keyboard issues with double inputs in the past and ghost writing (i.e. thi1es instead of this) in the past. Since pressing hard also means writing slowly, it could be an software issue too.

I'll try to uninstall some apps and free up some ram, maybe it does help. If not, the system will get reinstalled in a few weeks anyways, I'll tell you if it helped.

[raymo 177]

Sadly I've flashed LOS16, 17, 18 and 19 and be back in LOS18 at this moment. With full wipe data at every flash, and the keyboard has always this sort of strange behaviour (but not every time of course...)

- I wonder me if it could be a debouncing problem, also digging a little in internet I found this  :
"The keyboard works with a mixture of interrupts (the keyboard notifies the CPU about a keypress) and polling (the CPU repeatedly asks the keyboard for it's state). Most of the time interrupts are being used, but for some situations (I think they involve multiple keys being pressed at the same time) polling is still necessary."
In the microcontroller world, I would try to solder one more pullup resistor in the i2c bus, or add a .1uF condensator somewhere 😅

- It should also be the possibility that a software tricks make the job ?

-I'm quite sure the hardware isn't in cause here, or maybe for a lack of insulation. Have you ever try typing with the phone on a flare metallic surface grounded ? It looks like the problem disappear...static electricity ? Not 100% sure at this moment, but I suspect a possible explanation....

[SchattengestaIt 559]

When I had the software keyboard issues (see thread below), it mostly happened when the cpu was under load. This time I don't feel like there's a direct connection between cpu load and keyboard issues but maybe the system getting junked up.

I also think that software should be able to handle this, maybe by prioritising keyboard inputs over any other program. Sadly, I'm not experienced enough to write such a script/program. It was easier to control the keyboard from the Motorola Milestones/Photon Q...

 

 

[raymo 177]

12 minutes ago, SchattengestaIt said:

This time I don't feel like there's a direct connection between cpu load and keyboard issues but maybe the system getting junked up.

Agree with this, I tryied this possibility but even with heavy load or quite no load it works with same behaviour. It could be relative long time without issue, and sometimes it's quite useless. AFAIK a hard reboot don't really affect this.
strange...

[daniel.schaaaf 177]

My Pro1 has problems with the "a" and "d" buttons. It doesn't feel quite like a hardware issue. More like the software sometimes swallow a character. Whenever I see an "a" or "d" missing, pressing either of these two buttons very lightly works the second time. I don't see anything suspicious in the logcat either.

This might be the last straw for me and I am thinking about retiring my Pro1 ... with a sledge hammer. Yes, I am that disappointed! With the hardware keyboard having become unreliable, the phone is a nuisance to use. I hate the narrow screen with its curved edges.

[claude0001 1,341]

5 hours ago, daniel.schaaaf said:

My Pro1 has problems with the "a" and "d" buttons. It doesn't feel quite like a hardware issue. More like the software sometimes swallow a character. Whenever I see an "a" or "d" missing, pressing either of these two buttons very lightly works the second time. I don't see anything suspicious in the logcat either.

I've had similar issues with my Pro1 since day 1, though I cannot restrict the phenomenon to a certain subset of keys. In the beginning, me trying to type fast always resulted in many errors, either from missing characters or from the software registering additional (e.g. "addd&nitional") keys, I certainly had not pressed. 

Over time, I improved in hitting the keys "right", i.e., quite firmly and in a well-defined sequence. However, I cannot say that this kind of typing is much fun. I type slowly, and cannot practically use more than two fingers, even when putting the Pro1 down on a table. Overall, even after two years, using the Pro1's keyboard feels unnatural and tiring to me. I do not think it is a problem of "wear" which has worsened over time. I rather believe the keyboard is simply not very well designed with regard to mechanical key action and software readout. My impression is that the keyboard behaves worse during phases of heavy system load -- probably because of the mixed interrupt/polling mode of registering the key presses. The "duplicated hit suppression" workaround, that has been implemented in the LineageOS keyboard driver, improved things somewhat, but did not solve all problems for me. I also sometimes feel that disabling "fast poll" mode (afaik also a Lineage-only feature) gives me better results, but I am not sure about that ...

I recently fired-up my old N900, and was immediately back at ease thumb-typing on its tiny, slide-out keyboard. Some things must have been better thought-out there. On the other hand, I must admit that the Pro1's larger number of keys and special characters does makes it more useful in many of my applications (like e.g Linux terminal work). Combined with the advantage of having the entire screen free for content display, that easy access to the special chars is often the only reason I use the HW keyboard at all. When typing simple text messages, I, honestly, have always been more efficient using the Android on-screen keyboard.

[raymo 177]

I dig a little into this (not sure that a solution will be found, but as understanding can bring a deeper insight about reality....)
But first....
Could anyone pointing a silly noob the correct syntax to insert code please as I can share my results here ?🙈
[code] [/code] doesn't match, ...incapable...
Sorry, thanks !

Edited August 28, 2022 by raymo

[pistikem 85]

me too, having keyboard issues with my 2y-old FxTec Pro1, mainly with letters 'a', 'd', sometimes 's' - characters are either swallowed or doubled for the last month or so...

[claude0001 1,341]

2 hours ago, raymo said:

Could anyone pointing a silly noob the correct syntax to insert code

Just use the "code" icon (labelled "<>") in the text formatting bar.

[raymo 177]

aw9523b: key press: key_nr=38 keycode=0012 ff=0000 gpm=0000 glm=0000                       
aw9523b_irq_work: enter polling mode: p0_mask=ff
aw9523b: key release: key_nr=38 keycode=0012 gpm=0000 glm=0000
aw9523b_poll_work: enter irq mode
aw9523b_irq_handler: enter                                                                 

If I understand correctly, the process with the aw9523 chip is so :
1-key press,
2-enter polling mod, (the CPU repeatedly asks the keyboard for it's state)
3-key release
4-enter interrupt mod after polling
5-send an interrupt to the CPU via i2c

When a key is ignored, I found this :

 

aw9523b: key press: key_nr=25 keycode=001e ff=0000 gpm=0000 glm=0000
aw9523b_irq_work: enter polling mode: p0_mask=ff
aw9523b: key release: key_nr=25 keycode=001e gpm=0000 glm=0000                            
aw9523b_poll_work: enter irq mode
aw9523b_irq_handler: enter                                                                 

// ---- >>> HERE WE ARE, the key 25 was ignored, and you can see that he send the interrupt three times (if I understand correctly ?)

aw9523b_irq_handler: enter
aw9523b_irq_handler: enter

// ---- Then it works as normal, but one press on 'a' key was ignored

aw9523b: key press: key_nr=25 keycode=001e ff=0000 gpm=0000 glm=0000                       ...etc...

Maybe I misunderstand, but I wonder me if the i2c through aw9523 chip could be the source of the problem.
In normal way aw9523b_irq_handler appear only one time and then it goes to the next key event.

[raymo 177]

12 minutes ago, claude0001 said:

Just use the "code" icon (labelled "<>") in the text formatting bar

Thanks, this option disappear when using mozilla with the phone. Had to use "computer version" button

Edited August 28, 2022 by raymo

[SchattengestaIt 559]

Very interesting to see progress on this topic, especially on system level.

A quick feedback on my progress so far: uninstalling a couple of apps (although rarely used) and killing background processes improved the problem. Next step is limiting background processes to 4. I don't totally know what this does, but maybe it reduces background activities.

[raymo 177]

On my side, it's really strange. The keyboard works really (quite) perfect....since I completly disabled the auto-rotate screen in LOS18.1 as I never use it, as 'clacky keyboard on' automatic set the pro1 in landscape mode.

I think it's the first time I write here without a keyboard bug. Too good to be true, so I will test a few days and report here.
(...I write still a few words because of murphy's law and fatality of life to be sure....incredible...)

[raymo 177]

So, after one week of use, it seemed not related with auto-rotate settings.
It works, but sometimes felt one or two characters, then it works perfect, then it doesn't...
Also the next things I would try is to dismount the keyboard, but I have to wait for a possible Pro1x replacement, wich doesn't happen yet as I have a great allergy with android and google's stuffs and LOS is not ready (...yet ? I hope...) also SailfishOS looks really interesting, either droidian...sure I'll have to try at some point.

Sure I definitely don't give up, I agree with @claude0001this keyboard is probably and simply not well-designed. But it works well, this is proved, not everytime, that's my problem.