Caveat: The problem is still there. When the averaging kicks in, there is a window of time where keypresses won't register
This fix only delays the window a *really* long time, so the likelihood of hitting the delay is very low.
Each keypress resets the countdown.
- This requires that no strobes are floating.
On Rev. 0.4, pins D1 and D0 of the teensy must be manually connected to GND
On Rev. 0.2, when not using the 18 pin connector, pin C6 and C7 must be manually connected to GND
- Added a problem keys cli command, this is very useful to see which keys were disabled at startup because they looked like test points
(the fastest way to get the keys to re-enable is to take off a keycap and jiggle the spring)
- Removed 'Zombie' code
- Still a bit of iffiness with the averaging algorithm (should be looked at again)
- Added more CLI debugging (averaging and press/debounce)
- Tried to add auto cmake reset...so far unsuccessful
- Ported MBC-55X (was used to test compilers)
- Disabled in-source builds, to prevent CMake from shooting itself in the foot.
- Code should be working, but much is untested
- All of the old modules will need to update and use the new DefaultMap keymap
- There might still be some naming conflicts with some Scan Modules
- Includes serial putchar and getchar cleanup (overall)
- Moved avr-capsense to DPH (renaming)
- Basic cleanup for including CLI on the avr architecture
- Any keys that exceed the high average are temporarily disabled
- If the sense value goes below the initial average sense, it is re-enabled
- Even works if *many* keys are pressed initially
- Still needs a bit of detection of consistent high keys that sometimes sneak below the threshold
(May be able to have an elastic threshold that could deal with this)
- Removed dead code (and redundant calculations)
- Added debounce mechanism, currently looks for 5 consecutive scans
- Temporarily disabled the test point strobe on the kishsaver (spams debug messages, and not useful anymore; needs to be addressed before strobe detection is implemented though)
- Reorganized code to cleanup the main scan loop function
- Basic strobe mapper added (only kishsaver for now)
- Removed more (less obvious) dead code
- Moved keypress detection closer to detection
- Added more informative keypress debug messages
- Added a print macro for colourful convenience
- Removed the usb_keymap variable as it is no longer needed
- Changed usb_dirty to keymap_change (more accurate description)
- Removed the dumpkeys function and now detect key changes much sooner as well as displaying error messages more often
- Added a warming up information message and removed its error status (as it's not an error)
- Adding "template" keymap
- Removed "stray" functions, variables
- Cleaned up warnings
- Now builds
- Added buffered macro integration (rather than dealing with USB directly)
- Updated the print messages to use the Kiibohd print header
TODO
- Add generic matrix integration (this will require some changes to the matrix code)
- Add more comments...lots more
- Clean up dead code
- I won't do any more, as I can't make the converter fully functional on a modern computer
- Basic typing support is working
- Some of the function modes move around the ASCII values, and this takes a while to decipher and figure out the pattern; it's not hard to do, and I don't really feel like doing it
- The hard part is already done :P
- All modules have a default avr compatibilty right now (since that is what they were developed on)
- As modules get tested and added, arm support has to be turned on in the respective setup.cmake file
- CDC Output seems to be working
- USB Keyboard output has not been tested, but is "ready"
- UART and Timers have not been tested, or fully utilized
- Issues using Timer 0
- Initial template for MBC-55X Scan module (only module currently compatible with the arm build)
- Updated the interface to the AVR usb module for symmetry with the ARM usb module
- Much gutting was done to the Teensy 3 usb keyboard module, though not in an ideal state yet
- Abstracted USB Module
- Abstracted compiler specific includes for Scan, Macro, Debug, and USB modules
- Updated CMake build files to support changes
- Added abstractions necessary to main.c as well as a compiler specific include file
- Required some small API changes, shouldn't affect any other Scan modules (addition of a parameter)
- Still have the latching annoyances for some of the modifiers
- Swapped Esc and Tilde for convenience
- Removed the old Esc key from the latch list (no one needs 13 latches anyways...)
- Added the control cluster mappings (Insert, Delete, Home, End, PgUp, PgDn)
- All other misc keys are currently un-assigned if a keycap legend doesn't map to anything known
(e.g. blank keys, 000 and 00 are mapped, but your OS probably doesn't know what they mean :P)
Unfortunately, the protocol annoying, with only 3 keys having Press/Release signals.
Some of the keys were only meant to be pressed once, and do not have repeat functionality.
The current implementation is a bit slow, so some keys may get lost during USB sending.
This is probably fixable, not sure if I want to spend the time on it though.
There is a keyboard input protocol, but I'm not sure of what information can be transmitted to the keyboard.
The terminal does send some information back and forth between the keyboard, but I'm unclear of the purpose, as I can never get any response out of the keyboard.
- All spare cycles are used to get a more accurate data line sample (waaaay more than actually needed, but this helps against noise)
- Switched to a pre/release (add/remove) rather than an add->clear mechanism (or a buffer+debounce/add->clear)
- Note that a pull-down resistor is NEEDED on the data line
- Not fully reading for usage, but 90% of the way there for typing.
- Some soldering is required for 2 keys to work properly (Shift and Shift Lock)
- Even when complete, be careful when doing multiple key combos, as the key buffer is only cleared when all general keys are released (all except Shift, Shift Lock, and Code)
- Better macros are still needed for the clicker
- Current macros should be separated as a sub-module
- No LEDs implement yet
- Caps Lock LED seems to be burnt out...
- Typing and modifiers fully working
- Still some features left to implement
* Stop input
* Diagnostic test
* Setting interval before starting repeat rate
* More comments
* Handling LED stats
- LEDs are working
- Reset command is working
- Basic typing is working (USB side drops some characters)
- Changing repeat rate is working
- Querying modifiers and LEDs is working
- Initial layouts are complete
- Modifiers are not fully working yet
- Keyboard output fully working
- Not all of the features are implemented yet (input, speaker, or power
detection)
- There is no plan for mouse support
- Full support (NKRO, provided the USB module can handle it)
- Full LED support (available to the macro modules per input signals)
- Reset/Hold line support
- Handles repeat rate issues (keyboard does not send key updates as most
keyboards do...)
- Uses USART instead of polling
- Supports two way communication (only 3 commands to the keyboard
though)
- Kaypro sends data as ASCII, which severely limits the handling of
modifiers
- No release signal is sent, so even faking modifiers is next to
impossible outside of Shift and Ctrl
- Includes default QWERTY and Colemak layouts
- Developed a new buffer macro, which will become the basic macro module
once all the other modules have been ported (much more efficient, as
it keeps serial scanning schemes serial, and parallel scanning schemes
such as matrix, that already serialized into a sort of buffer for the
debouncing evaluation)
- This module is quite efficient, and would be able to handle very
excessive macro processing in the future.
- If more of the keypboard communication protocol is desired, I have a
rom dump of the microcontroller (M5L8049)
- Does not output to usb yet
- USART is functionally getting characters from the keyboard and sending
beep codes and ACK request code (Baud rate properly calculated)
- Uses hard-coded keys for the PC to Keyboard commands (not desireable)
- Keyboard sends ASCII data, which makes doing things like holding
multiple keys almost impossible to detect (the board isn't NKRO
anyways, though I've found 8+ key combos that work)
- Parent module no longer allows for a stand-alone compile
- Added HeathZenith matrix (pinout still needs to be done)
- Updated the HeathZenith keymaps
- Changed the original example matrix.h to matrix_template.h
- Not tested yet
- Added bug fixes for the scanCol and scanRow methods
- Merged in the powered versions of scanCol and scanRow
- Added the scanDual, but still needs heavy testing
- The scanDual uses twice the memory per key, so it can only scan half
as much compared to the other scan modes
- Includes new "table pinning" setup (much easier to understand)
- Matrix layouts will be split into their own dependent modules later
- Changed the HID Raw view id (too many teensy devices on this
computer...)
- Moved the macro processing outside of the usb timer signal
(this will be slower overall, but will result in more consistant
behaviour)
- Commenting cleaned up
- usb_keyboard_debug is not finished yet in terms of cleanup (needs DEBUB_PRINT #ifdef's)
- Added LoadFile script generation
- Formalized KeyIndex and USBKeys data containers
- Split parts of scan_loop into Macro, USB, and main.c
- Added interface to USB modules
- Added interface to Macro modules
- Added proper Debug module enable/disable
- Include option is currently "hacked" and needs to be fixed
- Builds on Linux, but Mac and Windows needs to be tested
- Loader script generation isn't complete
