Firmware for the ErgoDox keyboard
The official website is [ergodox.org] (http://www.ergodox.org).
Also see the [geekhack] (http://geekhack.org/showthread.php?22780-Interest-Check-Custom-split-ergo-keyboard) and [deskthority] (http://deskthority.net/workshop-f7/split-ergonomic-keyboard-project-t1753.html) discussion threads.
Kits are periodically for sale [on Massdrop] (https://www.massdrop.com/buy/ergodox), who also hosts [build instructions] (https://www.massdrop.com/ext/ergodox/assembly.php) and a [web UI] (https://www.massdrop.com/ext/ergodox) for configuring layouts.
A rewrite is currently underway (see the 'partial-rewrite' branch) and so this branch isn't quite receiving the attention it used to. Things that have changed since this document was last properly updated:
- Ryan Prince ([judascleric] (https://github.com/judascleric)) implemented sticky keys, and added a Colemak layout using them.
- Revision 2 of the firmware (on the 'partial-rewrite' branch) is working! Not ready for merging or a general release though, and I'm not ready to declare anything final, so you may not want to switch yet; but developers may want to think about checking it out. :)
- Ryan Prince ([judascleric] (https://github.com/judascleric)) implemented 3 media keys (play/pause, previous track, and next track) in Revision 1, using a bit of code from [hasu's tmk firmware] (https://github.com/tmk/tmk_keyboard). Revision 2 probably won't have this functionality for a while.
- Gyuri Horak implemented NKRO and mouse keys in [his branch] (https://github.com/dyuri/ergodox-firmware) of rev-2. Not merged, but I plan to implement the features later, if I can. See [Pull Request #28] (https://github.com/benblazak/ergodox-firmware/pull/28).
- Oleg Kostyuk (cub-uanic) [ported the TMK firmware] (https://github.com/cub-uanic/tmk_keyboard) (written by "hasu") to the ErgoDox!
- Jacob McIntosh (nacitar) implemented 4 more media keys (stop, mute, vol_up, and vol_down) in rev-1, fixed a media key bug, and implemented workman-p.
- Jacob McIntosh (nacitar) added
Table of Contents
generated with DocToc
- About This File
- About This Project
- About This Project (more technical)
- Dependencies (for building from source)
About This File
If you're viewing this on github, please note that directory links will only work if you're viewing this from the directory, and file links will only work if you're viewing this as a file. This is true for all the '.md' files here. The limitation is due to the way github addresses directories and files, and the fact that Markdown doesn't have any way (that I know of) to rewrite the URLs as would be required.
About This Project
This project is meant to be a usable firmware for the ErgoDox keyboard. I started writing it because, at the time, it was the easiest way for me to get I²C working, and I very much wanted to avoid having the designers put a 16-pin connector (or something awful like that) between the two halves :) . It was also my first project in pure C (though I'm planning to rewrite it in C++), and my first attempt at anything remotely this close to hardware. I've done my best to put decent documentation everywhere, in the hope that my perspective as a n00b to this area might be useful for others.
The 'master' branch is meant to be stable - the one you want if you're going to update your keyboard firmware, and usually the one you want if you're going to fork. This is a small project, so there may be times when other branches have a bunch going on, and 'master' doesn't, and things sort of fall out of sync... but I'll try to avoid that. See the other branches for what's going on there.
Also, lots of other documentation can be found intermixed with the source (especially in the accompanying '.md' files), and [references.md] (references.md) contains lots of good links, along with descriptions.
If you're just looking for binaries, they can be downloaded [here] dropbox-download-page.
If you're just trying to compile, jump to the How To: [Compile the Source Code] (#compile-the-source-code) section.
Issues and Feature Requests
Open issues, feature requests, and such are tracked [on github] (/benblazak/ergodox-firmware/issues).
Features (on the ErgoDox)
- 6KRO (conforms to the USB boot specification)
- Teensy 2.0, MCP23018 I/O expander
- ~167 Hz scan rate (last time I measured it) (most of which is spent communicating via I²C)
- firmware level layers
About This Project (more technical)
If you're looking to hack on the source, or just feel like reading it:
- The [makefile] (./makefile) and [build-scripts] (./build-scripts) folder in the toplevel directory are for building a collection of files for easy distribution. They are not guaranteed to work on non-Unix systems, and may be (read: are) more hackish than the stuff in [src] (./src). They help me out though.
- [src/lib] (src/lib) is for generally useful stuff relating to the firmware. [src/lib-other] (src/lib-other) is for generally useful stuff that I didn't write myself. The TWI and USB libraries are in there, along with the files containing key press and release functions.
- [src/keyboard] (src/keyboard) is for keyboard specific stuff. All the chip initialization code is there, along with the layout files, the software matrix to hardware matrix mapping, and hardware specific documentation.
- [src/main.c] (src/main.c) ties it all together, and provides a few higher level functions that are useful in the key press and release functions.
A few concepts that might be different:
The layer stack
When activated, layers are pushed onto the top of a stack. When deactivated, layers are popped out from wherever they are in the stack. Layers may be active in the stack more than once. When a keypress occures, the top layer is used to determine what actions to take.
Keypresses are functions
Each time a key is pressed, the "press" function assigned to that key on the current layer is called. When the key is released, the "release" function (from the same layer the key was on when it was pressed) is called. These functions may do pretty much anything - from sending multiple different keypresses to the host, to changing the firmware state in some way, to activating a new layer. They may also be "transparent", i.e. execute the function assigned to the key on the layer one down from the top layer (allowing for layers that are effectively "masks" over whatever layer was active before them).
Dependencies (for building from source)
See the PJRC [Getting Started] (http://pjrc.com/teensy/first_use.html) page for instructions on how to set up an AVR programming environment (be sure to click through all the subsections (in the navigation bar on the left), as there is essential information in each one). This project uses C (not Arduino), and Make. I'm compiling with GNU tools under OS X, but other environments (especially Linux, appropriately set up, or [WinAVR] (http://winavr.sourceforge.net/) under Windows) should work too.
I also assume that you are using [git] (http://git-scm.com/) (for
Most of these instructions (or notes, rather) are meant for people who's googling skills and patience are directly proportional to the amount of C programming they don't already know :) . I've done my best to organize and comment things though, so I hope that just about anyone who manages to find their way all the way here will be able to figure out a good deal from context.
Load Firmware onto the Teensy
Before beginning, make sure:
Your Teensy is plugged into your computer via USB
You have the appropriate version of [the Teensy loader application] (http://www.pjrc.com/teensy/loader.html) from PJRC installed.
You know what the "reset button" (a.k.a. the "tiny pushbutton") on the Teensy is. See the "HalfKay Bootloader Mode" section of the [Teensy First Use] (http://www.pjrc.com/teensy/first_use.html) page on the PJRC website.
After getting set up:
Run the Teensy loader program.
- This will give you a window, as shown on the website.
Click the "Auto" button on the upper right hand side of the window.
- The button will light up brighter green.
- This tells the loader program to load its current file whenever the Teensy is ready for it.
Locate your '.eep' and '.hex' files.
- If you don't have any, you can grab the latest ones [here] dropbox-download-page (in a '.zip' file). Choose the newest file who's name contains the name of the keymap you want (look for "qwerty" if you're not sure).
Drag and drop the '.eep' file onto the Teensy loader window.
- The information bar at the bottom of the window should now read "firmware.eep" followed by the percentage of the Teensy memory that will be used by this file.
Press and release the Teensy reset button (a.k.a. the "tiny pushbutton").
- The Teensy loader should inform you that it is loading the file. Wait until it's done: it shouldn't take long.
Drag and drop the '.hex' file onto the Teensy loader window.
Press and release the Teensy reset button (a.k.a. the "tiny pushbutton").
Your firmware is now loaded! Press a few buttons on your keyboard to make sure everything worked out all right, and enjoy :)
It may not be necessary to load the '.eep' file (if the file is 0 bytes, and the Teensy doesn't have anything loaded into its EEPROM already, it doesn't make any difference), but it's good to do so anyway, just to be safe. It won't hurt anything either way.
Now that your firmware is loaded, there should be a keyboard shortcut you can press instead of the Teensy reset button. See the documentation on your layout for more information.
Change the Direction of the Diodes
That is, how to change whether the rows or the columns are being driven. This can be done for each side of the keyboard independently.
See [src/keyboard/ergodox/options.h] (src/keyboard/ergodox/options.h)
After this, you'll need to recompile. See the [About This Project] (#about-this-project) section above.
Compile the Source Code
Note: This relates to compiling in the [src] (src) directory. The toplevel build process (for generating the ui-info file and such) isn't really intended to be portable; but you could probably get it working without too much trouble, if you're familiar with programming in a Unix environment. I'd suggest looking through the toplevel Makefile, as a staring point, if that's your goal.
Read the [Dependencies] (#dependencies-for-building-from-source) section.
Take a quickish glance at the [About This Project (more technical)] (#about-this-project-more-technical) section.
Navigate to the [src] (src) directory (not the toplevel directory) in a terminal, and type
If everything worked, the '.hex' and '.eep' files will be in the [src] (src) directory (where you currently are).
Create a New Keymap
Files to reference:
- Keycode macros: see the file
- See [keyboard.h] (src/lib/usb/usage-page/keyboard.h) for the actual values, with references to the specification.
- Keyboard functions: see all files in the folder [src/lib/key-functions/public] (src/lib/key-functions/public).
- Template layout files: see the QWERTY keymap source files in the folder
- Currently [qwerty-kinesis-mod.c] (src/keyboard/ergodox/layout/qwerty-kinesis-mod.c) and [qwerty-kinesis-mod.h] (src/keyboard/ergodox/layout/qwerty-kinesis-mod.h)).
- You'll probably want to make a copy of each to use as a template.
- Keycode macros: see the file [src/lib/usb/usage-page/keyboard--short-names.h] (src/lib/usb/usage-page/keyboard--short-names.h).
You will need to set the
LAYOUTvariable in [src/makefile-options] (src/makefile-options) to the base name of your new layout files before you recompile. ('.h' files may be called what you wish, but '.c' files must all have the same prefix (i.e. "base name") or they won't be compiled).
Among other things, the '.h' layout file defines the macros that control the meaning of each of the LEDs on the keyboard (capslock, etc.). They may be changed freely (or removed, to disable that LED).
The '.c' layout file defines the values (keycode|value, press function, release function) assigned to each key, for each layer.
- If a "press" function is set to NULL for a given layer, nothing will be
called when the key is pressed on that layer. Likewise for "release"
functions. If both are set to
NULL, nothing will happen when the key is pressed in either event, so it doesn't matter what the keycode|value is for that layer - but you should probably pick something like
0and stick to it, just for clarity.
- The default number of layers is 10 (defined in [default--matrix-control.h] (src/keyboard/ergodox/layout/default--matrix-control.h) - you can override it in the layout's '.h' file, if you like). You don't have to define all of them in the '.c' layout file, just the ones you want (C sets the uninitialized portions of the matrices to 0).
- Make sure that in layer-0 nothing is transparent (see [About this Project (more technical)] (#about-this-project-more-technical). Behavior is undefined if this occurs (off the top of my head, it'll probably cause an infinite loop - and you'll have to reset your keyboard (unplug it and plug it in again)).
- Be careful how you assign things. Pay close attention to the possible combinations of keypresses that could occur. It's perfectly possible, for example, to make a layout that can shift to layer 2 (or some layer that doesn't even exist) and can't shift back, or that fails to tell the host when keys are released. It's fairly unlikely that anything bad will happen if you mess up (though you could theoretically generate a sequence of keypresses that accidentally do very bad things on your machine), but it's important to have the possibility in mind. Please consider yourself warned :) .
- If a "press" function is set to NULL for a given layer, nothing will be called when the key is pressed on that layer. Likewise for "release" functions. If both are set to
Add Code for a Different Type of Keyboard
All the function prototypes and macro definitions required by the rest of the code should be in the following files (using the ergodox code as an example):
- [controller.h] (src/keyboard/ergodox/controller.h)
- [layout.h] (src/keyboard/ergodox/layout.h) (which in the ergodox code only exists to include [layout/default--led-control.h] (src/keyboard/ergodox/layout/default--led-control.h) and [layout/default--matrix-control.h] (src/keyboard/ergodox/layout/default--matrix-control.h))
- [matrix.h] (src/keyboard/ergodox/matrix.h)
(In broad terms, you'll need to define functions that initialize and update the matrix, LED control macros, macros defining matrix dimensions, and keycode | key-lookup macros (or functions).)
You will need to set the
KEYBOARDvariable in [src/makefile-options] (src/makefile-options) to the name of your new keyboard folder before you recompile. You may also wish to change some of the other options in that file.
The '.h' files listed above must exist, with those names, in the toplevel of your keyboard's directory (e.g. in
src/keyboard/new-keyboard-name). They may include other '.h' files if you wish to put various definitions or prototypes in other places. They will be included (with the help of some handy macros) in the corresponding files in [src/keyboard] (src/keyboard), which are in turn the files included by "main" and the keyboard functions.
If you change the way things are included, be careful for circular includes. "main", the keyboard functions, and the keyboard code all need various parts of each other.
Make sure to keep your eye on the
SRCvariable in [src/makefile] (src/makefile), to make sure all your '.c' files are getting compiled.