Closed Thread
Thread Tools
Posts: 339 | Thanked: 1,623 times | Joined on Oct 2013 @ France
#41
Hello !

Why not using a chip like the ADP5588, "designed for QWERTY type phones that require a large keypad matrix" : http://www.analog.com/en/power-manag...s/product.html ?

It is one of the keyboard chips that already have a driver in the linux kernel, so using it should mean near plug and play capability, as the code is already present in the kernel, and would be far more efficient than python.

The problem I can see is that Jolla didn't say if the interrupt line will be availlable. If not, then we will need to poll the keyboard, which is a shame... I2C is already quite limited, so not having interrupt lines will make it even weakier.

Also, one thing to not forget :
Add a contact to detect if keyboard is out or not. The virtual keyboard should be deactivated if the hard keyboard is out, and orientation should be locked in landscape.

My 2 cents. Keep on, I will try to follow your work.
 

The Following 3 Users Say Thank You to Zeta For This Useful Post:
Posts: 1,104 | Thanked: 5,652 times | Joined on Feb 2010 @ Holland
#42
Originally Posted by Zeta View Post
Hello !

Why not using a chip like the ADP5588, "designed for QWERTY type phones that require a large keypad matrix" : http://www.analog.com/en/power-manag...s/product.html ?

Also, one thing to not forget :
Add a contact to detect if keyboard is out or not. The virtual keyboard should be deactivated if the hard keyboard is out, and orientation should be locked in landscape.

My 2 cents. Keep on, I will try to follow your work.
Looks nice! But runs on max 3 volts... (and is ordered in multiples of 1500).

Originally Posted by dirkvl View Post
At this moment the prototype will feature the N900 keyboard (3x13 keys), a i2c i/o expander, some resistors for pull-up logic, external connection for i2c connection to RPi and an on/off switch on the board (testing for recognition whether keyboard is connected).
I wil put a simple switch on the prototype board to simulate sliding out!
 

The Following User Says Thank You to dirkvl For This Useful Post:
Posts: 1,104 | Thanked: 5,652 times | Joined on Feb 2010 @ Holland
#43
For a potential version 2 prototype, this might be a potential candidate! Runs up to 5.5 volts, 1mm thick and lots and lots of functionality!

However, I have zero experience with programming these kind op microprocessors. Also, putting this thing on a pcb is a bit out of my reach. Please help if you have serious skills!

 

The Following 4 Users Say Thank You to dirkvl For This Useful Post:
Posts: 59 | Thanked: 66 times | Joined on May 2007
#44
I've been playing a bit with ST's STM32x chips and started dream about having a qwerty built around it when I bumped into this thread.

For PCB designs I really recommend Kicad. It's open source, commonly used and works on Linux and Windows.

It's not that hard to design a PCB with Kicad and order it from Itead. And Itead's prices are very suitable even for individual hobbyists.

Last edited by kulve; 2013-10-08 at 10:42.
 

The Following 7 Users Say Thank You to kulve For This Useful Post:
Posts: 59 | Thanked: 66 times | Joined on May 2007
#45
How about the mechanical side. What kind of buttons would/could there be connecting the pads on the PCB? I tried quickly googling for general solutions but couldn't find anything useful.
 
Posts: 59 | Thanked: 66 times | Joined on May 2007
#46
Originally Posted by Zeta View Post
Why not using a chip like the ADP5588, "designed for QWERTY type phones that require a large keypad matrix" : http://www.analog.com/en/power-manag...s/product.html ?

It is one of the keyboard chips that already have a driver in the linux kernel, so using it should mean near plug and play capability,
That might be an easy solution serving at least the basic functionality.

After some initial though I think I would like to see e.g. a mode button for special characters, maybe leds to indicate which mode is on, etc. For that it might already make sense to have a separate mcu. While at it, it could notify the OS about landscape etc. over I2C.

We still would definitely need the extra interrupt line because polling isn't an option due to power consumption.
 
Posts: 1,104 | Thanked: 5,652 times | Joined on Feb 2010 @ Holland
#47
Originally Posted by kulve View Post
While at it, it could notify the OS about landscape etc. over I2C.

We still would definitely need the extra interrupt line because polling isn't an option due to power consumption.
I imagine a simple i2c i/o explander will have to use polling, can a microcontroller send out a signal when connected to power that it is in use?
 
Posts: 59 | Thanked: 66 times | Joined on May 2007
#48
Originally Posted by dirkvl View Post
I imagine a simple i2c i/o explander will have to use polling, can a microcontroller send out a signal when connected to power that it is in use?
If there's only I2C, then no.

In I2C only master (i.e. the phone in this case) can initiate communication. Hence if there is not separate interrupt line that the other half can use to notify the main cpu that it has something to tell, the main cpu needs to constantly poll the other half for status.
 

The Following User Says Thank You to kulve For This Useful Post:
Posts: 1,104 | Thanked: 5,652 times | Joined on Feb 2010 @ Holland
#49
Okay.. So what will work?

This one? Seems to have a lot of funtions, but I do not know what to look for in the manual
 
Posts: 59 | Thanked: 66 times | Joined on May 2007
#50
Originally Posted by dirkvl View Post
Okay.. So what will work?
Any modern mcu with enough legs will do. I would probably go with STM32L as I'm already familiar with that.

The interrupt line is something that Jolla must provide in addition to I2C to the Other Half. It's just one extra "wire" while I2C needs power, ground, clock and data.
 

The Following 3 Users Say Thank You to kulve For This Useful Post:
Closed Thread

Tags
keyboard, limited-edition, otherhalf, qwerty


 
Forum Jump


All times are GMT. The time now is 03:31.