TMK keyboard firmware collection

[snacksthecat]

Unfortunately disabling bootmagic didn't work. I'll keep playing around with different things.

The pcb is an Alps plate spring pcb that I designed (using your kicad libraries btw). It's really not at all comparable to GH60. I just used that project as a starting point.

[hasu]

Ah, then post schematic of your pcb so that we can check your matrix scan code.

And, what is your exact MCU part number or source?
Did you fiddle with MCU fuse setting with ISP programmer, perhaps?

[Muirium]

So, I've had a Caps Lock macro on my shoulder today. Having gotten it working again on Xwhatsit earlier, I decided to give my HHKB a go. Sure, Fn + Tab = Caps Lock, I know, but I really like Shift + Shift = Caps Lock the best. I hadn't tried this sooner because I don't really understand macros in TMK, and I also know Hasu uses Shift + Shift as the magic key combo for entering bootloader mode etc. But, hey, I've always got that reset button on the back! So here's my attempt today in Unimap.

Screenshot 2019-09-27 at 5.51.38 pm.png (281.74 KiB) Viewed 24540 times

I'm using the same trick Hasu showed me for my Command + Esc = Command + Tilde "macro". Namely: borrow another layer. In layer 5, both shifts are remapped to Caps Lock, like I did on my Kishsaver and AT in Xwhatsit's GUI. No timed macro cunningness here, but it works anyway. Well, with one downside.

To enter the HHKB's bootloader, you can either pull off the DIP switch port cover and press the red reset button, *or* press Shift + Shift + Fn + P (namely Shift + Shift + Pause/Break) to do it entirely on the keyboard. This trick has lost me that capability. I tried mapping P to Pause/Break in Layer 5 but that doesn't trigger it, and loses me the rather more useful ability to type capital P while holding Shift!

I wonder, is there a better way to achieve this? I don't want to have to mess in code, and all too likely break compatibility with Unimap. But I do like playing with my HHKB's Hasu-powered programmability often enough that the Magic key combo is plenty handy in its own right.

[Muirium]

Oh, actually, I think I've got it! Check this out:

Screenshot 2019-09-27 at 6.10.42 pm.png (286.9 KiB) Viewed 24532 times

I realised the problem I made for myself was I'd disabled Hasu's "Magic key" by swapping it out to Caps Lock. But! The only time I use his Magic key is when I'm pressing Pause/Break to enter the bootloader. And! Break is on Layer 1. So how about I just made sure those two Shifts were real Shift keys on Layer 1?

See, I read enough of TMK's documentation to know how "transparent" keys work. Any given layer overwrites the one below. My special layer shifting (and Shift destroying) Shift keys are down on Layer 0. If I overwrite them with normal Shifts up on Layer 1, all subsequent layers inherit those, until Layer 5, which is where Caps Lock looms, far away until it's called for.

So, as long as I press Fn + Shift + Shift + P in that order, Fn first!, I can enter the bootloader all posh style, without having to pull the DIP switch cover off. Very nice! Got to love this programmability. Powerful enough you can make yourself new problems and fix those too!

[John Doe]

Hi, guys, need your help here.

When I tried to compile for .hex file from the source code for a vintage kbd, error info was showed like this:

Code: Select all

In file included from keymap_mario.c:1:
keymap_common.h:50:37: error: 'KC_12' undeclared here (not in a function); did you mean 'KC_F12'?
   50 |     { KC_NO,    KC_##K11, KC_##K12, KC_##K13, KC_##K14, KC_##K15, KC_##K16, KC_##K17 }, \
      |                                     ^~~
keymap_common.h:79:3: note: in expansion of macro 'KEYMAP'
   79 | ) KEYMAP( \
      |   ^~~~~~
keymap_mario.c:23:5: note: in expansion of macro 'KEYMAP_N'
   23 |     KEYMAP_N( \
      |     ^~~~~~~~

And keymap_common.h file was defined like this:

Code: Select all

#ifndef KEYMAP_COMMON_H
#define KEYMAP_COMMON_H

#include <stdint.h>
#include <stdbool.h>
#include "keycode.h"
#include "action.h"
#include "action_macro.h"
#include "report.h"
#include "host.h"
#include "print.h"
#include "debug.h"
#include "keymap.h"


#define KEYMAP( \
    K00, K01, K02, K03, K04, K05, K06, K07, \
    K08, K09, K0A, K0B, K0C, K0D, K0E, K0F, \
         K11, K12, K13, K14, K15, K16, K17, \
    K18, K19, K1A, K1B, K1C, K1D, K1E, K1F, \
    K20, K21, K22, K23, K24, K25, K26, K27, \
    K28, K29, K2A, K2B, K2C, K2D, K2E, K2F, \
    K30,                          K36, K37, \
    K38, K39, K3A, K3B, K3C, K3D, K3E, K3F, \
    K40, K41, K42, K43, K44, K45, K46, K47, \
    K48, K49, K4A, K4B, K4C, K4D, K4E, K4F, \
    K60, K61, K62, K63, K64, K65, K66, K67, \
    K68, K69, K6A, K6B, K6C, K6D, K6E, K6F, \
    K70, K71, K72, K73, K74, K75, K76, K77, \
    K78, K79, K7A, K7B,           K7E, K7F  \
) { \
    { KC_##K00, KC_##K01, KC_##K02, KC_##K03, KC_##K04, KC_##K05, KC_##K06, KC_##K07 }, \
    { KC_##K08, KC_##K09, KC_##K0A, KC_##K0B, KC_##K0C, KC_##K0D, KC_##K0E, KC_##K0F }, \
    { KC_NO,    KC_##K11, KC_##K12, KC_##K13, KC_##K14, KC_##K15, KC_##K16, KC_##K17 }, \
    { KC_##K18, KC_##K19, KC_##K1A, KC_##K1B, KC_##K1C, KC_##K1D, KC_##K1E, KC_##K1F }, \
    { KC_##K20, KC_##K21, KC_##K22, KC_##K23, KC_##K24, KC_##K25, KC_##K26, KC_##K27 }, \
    { KC_##K28, KC_##K29, KC_##K2A, KC_##K2B, KC_##K2C, KC_##K2D, KC_##K2E, KC_##K2F }, \
    { KC_##K30, KC_NO,    KC_NO,    KC_NO,    KC_NO,    KC_NO,    KC_##K36, KC_##K37 }, \
    { KC_##K38, KC_##K39, KC_##K3A, KC_##K3B, KC_##K3C, KC_##K3D, KC_##K3E, KC_##K3F }, \
    { KC_##K40, KC_##K41, KC_##K42, KC_##K43, KC_##K44, KC_##K45, KC_##K46, KC_##K47 }, \
    { KC_##K48, KC_##K49, KC_##K4A, KC_##K4B, KC_##K4C, KC_##K4D, KC_##K4E, KC_##K4F }, \
    { KC_##K60, KC_##K61, KC_##K62, KC_##K63, KC_##K64, KC_##K65, KC_##K66, KC_##K67 }, \
    { KC_##K68, KC_##K69, KC_##K6A, KC_##K6B, KC_##K6C, KC_##K6D, KC_##K6E, KC_##K6F }, \
    { KC_##K70, KC_##K71, KC_##K72, KC_##K73, KC_##K74, KC_##K75, KC_##K76, KC_##K77 }, \
    { KC_##K78, KC_##K79, KC_##K7A, KC_##K7B, KC_NO,    KC_NO,    KC_##K7E, KC_##K7F }  \
}

#define KEYMAP_N( \
    K00, K01, K02, K03, K04, K05, K06, K07, \
    K08, K09, K0A, K0B, K0C, K0D, K0E, K0F, \
         K11, K12, K13,      K15, K16, K17, \
    K18, K19, K1A, K1B, K1C, K1D, K1E, K1F, \
    K20, K21, K22, K23, K24, K25, K26, K27, \
    K28, K29, K2A, K2B, K2C, K2D, K2E, K2F, \
    K30,                          K36, K37, \
    K38, K39, K3A, K3B, K3C, K3D, K3E, K3F, \
    K40, K41, K42, K43, K44, K45, K46, K47, \
    K48, K49, K4A, K4B, K4C, K4D, K4E, K4F, \
    K60, K61, K62, K63, K64, K65, K66, K67, \
    K68, K69, K6A, K6B, K6C, K6D, K6E, K6F, \
    K70, K71, K72, K73, K74, K75, K76, K77, \
    K78, K79, K7A, K7B,           K7E, K7F  \
) KEYMAP( \
    K00, K01, K02, K03, K04, K05, K06, K07, \
    K08, K09, K0A, K0B, K0C, K0D, K0E, K0F, \
         K11, K12, K13, NO,  K15, K16, K17, \
    K18, K19, K1A, K1B, K1C, K1D, K1E, K1F, \
    K20, K21, K22, K23, K24, K25, K26, K27, \
    K28, K29, K2A, K2B, K2C, K2D, K2E, K2F, \
    K30,                          K36, K37, \
    K38, K39, K3A, K3B, K3C, K3D, K3E, K3F, \
    K40, K41, K42, K43, K44, K45, K46, K47, \
    K48, K49, K4A, K4B, K4C, K4D, K4E, K4F, \
    K60, K61, K62, K63, K64, K65, K66, K67, \
    K68, K69, K6A, K6B, K6C, K6D, K6E, K6F, \
    K70, K71, K72, K73, K74, K75, K76, K77, \
    K78, K79, K7A, K7B,           K7E, K7F  \
)

#endif

The keymap_mario.c file was here, and the specific keys will be changed then if compiling without error, just for test now.

Code: Select all

#include "keymap_common.h"

const uint8_t PROGMEM keymaps[][MATRIX_ROWS][MATRIX_COLS] = {
    KEYMAP_N( \
        1,   1,  1,  1,  1,  1,  1,  1,   \
        9,   2,  3,  4,  5,  6,  7,  8,   \
             1,  9,  12,     1,  1,  1,   \
        8,   1,  2,  3,  4,  5,  6,  7,    \
        1,   9,  0,  1,  1,  1,  1,  1,   \
        1,   1,  1,  2,  3,  1,  1,  1,  \
        1,                       1,  1,   \
        I,   Q,  W,  E,  R,  T,  Y,  U,   \
        1,   O,  P,  1,  1,  1,  2,   2,   \
        K,   A,  S,  D,  F,  G,  H,  J,     \
        3,   L,  2,  3,  3,  2,  1,   1,  \
        1,   7,  4,  5,  6,  1,  1,   1,  \
        2,   Z,  X,  C,  V,  B,  N,  M,     \
        1,   2,  1,  2,          2,   3 \
)const action_t PROGMEM fn_actions[] = {};

Using Macos Catalina10.15.2 system environment, Xcode for coding, maybe no business with the error tho.

THANKS in advance.

[John Doe]

Oh, I made a big mistake for the keymap_mario.c. No way for the row 3 and col 4 with a 12.
Can't believe my stupidness.
Solved now.

Edit:
another problem appeared:
When flashing the .hex file to the teensy2.0 with teensy loader, it seems failed. Then I tried original blink_fast_Teensy2.hex as a test, no issue appeared and the LED start blinking normally. Then I flashed soarer's hex file lot of times, some of them flashed successfully but more failed. I guess a possible reason may has connection with the .hex size (blink_fast_Teensy2.hex 7kb, soarer.hex 40, and my .hex file 62kb, all under MacOS)? Or what other potential causes?

Any help will be appreciated.

[John Doe]

Need to throw more money to get teensy2.0++ as more test?

[John Doe]

Sorry for fuss, I got a teensy2.0++ and make it worked finally. But a new phenomenon appeared, thats when I press ctrl + alphabet characteristic, like I press a-z while holding the alt key, it shows like this:

Code: Select all

å∫ç∂´ƒ©˙ˆ∆˚¬µ˜øπœ®ß†¨√∑≈¥Ω

still MacOS system, anybody could give a help? Thanks.

[Findecanor]

Isn't that what is supposed to happen on Mac when you hold Alt/Option and press an alphabetic key?

[John Doe]

Isn't that what is supposed to happen on Mac when you hold Alt/Option and press an alphabetic key?

Sorry, I just change to MacOS from windows for few days, not familiar with it yet. Is there a way to switch them to like alt+a = select all status?Thanks.

[John Doe]

hey, guys, confused about features of tmk, so better post it here.

I have had experience in using tmk to make 3rd-party controllers for vintage keyboards, and begin to realize a question, emm, or perhaps couldn't be called question at all. It's like I have to wire the pins of the board, like teensy, to every col&row of the matrix respectively, which means there'll have to be lots of leads. Further, keyboard PCB normally has 1/2/3 decoders to reduce wires connected with original controller, replacement of teensy controllers ignored them sadly. So what i concerned is if there're ways/features in the whole project to reuse the original decoders as much as possible and reduce the leads just like the original controller does?

Soarer's firmware uses muxstrobe to achieve it, so maybe it's good trick to grab it in the open source project.

[hasu]

You can use the decorders of course, it would be easy to write code to use them if you can track traces of the pcb and find datasheet of the decorders.

But I can't say about how to implement universal feature supports various decorders used on every keyobard but it would be certainly helpful to other users if you can work on this instead of write code just for your specific decoders. I can help you.

[Muirium]

Now I know macros are a thing in TMK, if you build from source. But, anyone got a crafty idea how to do them using Unimap alone? I’m not talking about the general case, but rather specifically: typing grave accented vowels using the Control key on the Mac and iPad. I’m after this:

Control + E >> Alt + backtick, then E = è
Shift + Control + E >> Alt + backtick, then Shift + E = È

(I might use some other trigger besides bare Control, on the off chance I wind up needing it in Terminal. But that’s the pattern I’m looking for, for grave accent vowels.)

Kbdfr may be smirking, but I’m looking for a little mòr on the vowel front from my HHKB. Just grave accents, as they’re for Gaelic, which uses them plentifully and exclusively: no other accents, just strac gu mòr. The dead key dance is a little tedious when you need it more than once or twice a line.

[John Doe]

Heya, hasu and guys. Have to ask for help here.

I wanna to add one P00 key on my pingmaster so that I expect to get result like: I press the key, then output double 0, easy, right?
But when I really tried this, it outputs nothing. I dont know if my understanding of the P00 key is wrong or somewhere I set up incorrectly.

Any help will be appreciated, thanks lot.

keymap_p00.png (48.33 KiB) Viewed 21891 times

[hasu]

what is OS you are using? Did you try other OS if you have any?
And firstly you want to check if the OS supports the 'KP_00' key(USB HID usage 0xB0, or 176 in decimal).


And debug log on 'hid_listen' would be helpful in most cases.
Press both shifts and K to see keyboard report the converter sends to system.

https://github.com/tmk/tmk_keyboard/wiki#debug

[raoulduke-esq]

I had to implement a macro in code to get 000 on mine because macOS does not do 000 natively.

[John Doe]

thanks, guys.

I tried win64 ultimate, Ubuntu16 and macos catalina(10.15.6), nothing happened. I tried to set the key to A and it did output A so the pcb is okay.

And the hid_listen reported this, could help?

(UX9BC$YOQV}4`G`SEX25EC.png (15.14 KiB) Viewed 21860 times

[hasu]

I checked source code and confirmed that these key codes(0xB0-DD) including KP_00 and KP_000 are not supported at this time.
https://github.com/tmk/tmk_keyboard/blo ... #L357-L402

You can't send the KP_00 with current firmware in the end. I may change this in future release but not sure.

I think KP_00 is not supported and not useful on Windows probably. My Ubuntu setup recognizes the key itself but doesn't register any character by default.

With this patch you can send the key to test if you can build firmware yourself.

Code: Select all

diff --git a/tmk_core/common/keycode.h b/tmk_core/common/keycode.h
index 7ed711a4..e5773ac4 100644
--- a/tmk_core/common/keycode.h
+++ b/tmk_core/common/keycode.h
@@ -27,6 +27,7 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #define IS_ANY(code)             (KC_A         <= (code) && (code) <= 0xFF)
 
 // Use original HID usages on B0-DD instead of TMK specific codes
+#define USE_ORIGINAL_HID_USAGE
 #ifndef USE_ORIGINAL_HID_USAGE
 #define IS_KEY(code)             (KC_A         <= (code) && (code) <= KC_EXSEL)
 #else

[John Doe]

thanks, hasu. You helped alot, I'll check how it'll go.

[John Doe]

Hey, me again.

I have added the P00 thru using the Macro feature.

Code: Select all

const macro_t *action_get_macro(keyrecord_t *record, uint8_t id, uint8_t opt)
{
    switch (id) {
        case P00:
            return (record->event.pressed ?
                    MACRO(T(P0), T(P0), END ) :
                    MACRO_NONE );
    }
    return MACRO_NONE;
}

But now I face another question: my expectation is the key could output consistently when holding the key and stop that until key released. But seems the key of current setting is it only output once when you press the key down without releasing, and output again until you press again.

Within each macro, the following commands can be used:

I() change interval of stroke.
D() press key
U() release key
T() type key(press and release)
W() wait
SM() store modifier state
RM() restore modifier state
CM() clear modifier state

I didn't find other functions that could help yet, did I miss sth or it's just how the macro keys work? Noticed in the Key Action section at least some modifiers could be "hold" like I mentioned above. So I'm a little bit confused.

Thanks in advance.

[hasu]

You can't run a MACRO repeatedly in current implementation at least.

This will be what you can do for similar behaviour.

Code: Select all

            return (record->event.pressed ?
                    MACRO(T(P0), D(P0), END ) :
                    MACRO(U(P0), END));

[John Doe]

You can't run a MACRO repeatedly in current implementation at least.

This will be what you can do for similar behaviour.

Code: Select all

            return (record->event.pressed ?
                    MACRO(T(P0), D(P0), END ) :
                    MACRO(U(P0), END));

I have replaced code block with this one, not exact what I want but pretty enough. Appreciated alot Hasu.

[doomsday_device]

recently acquired a tatung branded sun 5 (?) board, successfully restored and converted with hasus tmk sun firmware ( https://github.com/tmk/tmk_keyboard/tre ... er/sun_usb ).
it has a big beeper and screams everytime i connect. so my guess would be it is working like the other sun keyboards and is potentionally able to output beeps/clicks with each keypress.

to use this feature i would need to use the tmk "magic" key combo, which in this case would be both shift, both alt or both meta keys + a single extra key. it seems the matrix isnt able to do that.

my question here is, is there any way to remap "magic" key combos, or even to map a key to a single hexcode?
beside wiring up converters and putting the *.hex from the online editor on to the chip my knowledge is very limited.

pic of board attached for beauty bonus

IMG_20220515_202217.jpg (3.26 MiB) Viewed 19867 times

also off topic: if anyone has any infos on the board or the system it came with it would be much appreciated. couldnt find any infos or pictures at all. its a forward fdw-107a

[hasu]

to use this feature i would need to use the tmk "magic" key combo, which in this case would be both shift, both alt or both meta keys + a single extra key. it seems the matrix isnt able to do that.

my question here is, is there any way to remap "magic" key combos, or even to map a key to a single hexcode?
beside wiring up converters and putting the *.hex from the online editor on to the chip my knowledge is very limited.

See this post.
https://geekhack.org/index.php?topic=10 ... msg3125687

[Muirium]

A finicky layer-logic question here!

So, I've been trying something new with my HHKB-style arrow keys: I want to toggle between arrow keys and mouse cursor control, on the same diamond of keys. I put the mode toggle on Fn+Esc. The idea is that when I press Fn+Esc the HHKB toggles between Fn+; = Left and Fn+; = Mouse Left, etc. Whenever I'm not holding Fn, the keyboard behaves the same. Ideally!

The way I've gone about this is using layers, of course. Take a look. Here's what all the layers look like:

µMouse Keys toggle.jpg (434.01 KiB) Viewed 19338 times

I made a second base layer (2) for my mouse key layer (3) to fall down onto when I'm not holding Fn. That works quite well. I'm almost pleased with it. The trouble is the precise behaviour of that Fn+Esc toggle.

See, what I want is to be able to hit Fn+Esc and toggle immediately. In practice, that means without having to release the Fn key first. Trouble is: what actually happens is the toggle is not instant. It still toggles, but only after I have released the Fn key. So you get strange situations where you're pressing arrow keys, hit the toggle, the arrow keys are still active, but then switch over to mouse keys after you've let go of Fn then hold it again. Wrong!

Strange (stateful) behaviour like this kinda ruins the simplicity I'm aiming for with that toggle. It really needs to be reliable and simple for it to feel right and for my muscle memory to rely on it.

So what I've got just now is a working toggle—yay—but one that needs double de-clutched first—boo! Using it like this feels like needing to "switch into neutral first" every time I touch it. Not the smooth toggle I'm looking for.

Is there some better way I should do this?

(What I'm really up to is implementing this on all my keyboards. So QMK is in the picture, too. I have HHKB arrow keys everywhere, way beyond my literal HHKBs, so mouse keys with a reliable toggle is my big new idea. If I can make it stick everywhere, job done! In my experimenting so far, QMK's layer toggling logic is different to TMK's.)

[BuGless]

And the keyboard you now use TMK on cannot run QMK instead? As far as I understood it, the keyboards supported by QMK should be close to a superset of the ones being supported by TMK?

[Muirium]

Good point. The keyboard I’m using is my HHKB Pro 2 running Hasu’s own HHKB controller. I did just find this:

https://github.com/qmk/qmk_firmware/tre ... /hhkb/ansi

Don’t know if it supports my Bluetooth controller however. The readme doesn’t mention it and only links to Hasu’s USB controller.

Maybe this?

https://github.com/qmk/qmk_firmware/tre ... /hhkb/rn42

Looks a bit outdated compared to the /ansi (USB) folder. And my QMK experience is limited to Pandrew’s Xwhatsit boards. Does the standard QMK configurator know these HHKB controllers exist?

[Muirium]

Digging a little deeper, I tried a couple of modifications to the Fn key itself. First I tried changing it to TRNS in Layers 1 and 3 (same system as the chart above) just in case L1 is somehow overriding the T2 layer toggle while it's held down. That didn't have any effect, fair enough. So then I tried this:

Screenshot 2022-07-17 at 12.28.21 pm.jpg (89.77 KiB) Viewed 19243 times

I made the all-important bottom corner Fn key into NO key at all on layers 1 and 3. It surely can't override the toggle if it's no key at all, right? I fully expected this to disable my HHKB's function layers entirely, too; it's just an experiment to probe the layer logic. But that's not what wound up happening.

So the result? The layers are still exactly the same, even after double checking I flashed the right hex file! Holding Fn still gives you Layer 1, even though it's absent there. And I still get the same unintuitive behaviour when pressing Fn + Esc until I release Fn.

Here's the gory details, it's actually different in each direction of transition:

• When in Layer 1—normal arrow key function—pressing the toggle turns the keys into their unmodified Layer 0 selves: holding Fn and press ; you get ; The keys only transform after you release Fn then press it down again.
• When in Layer 3—mouse key function—pressing the toggle appears to do nothing: they remain mouse keys: holding Fn and press ; you get Mouse Left. The keys only transform after you release Fn then press it down again.

Perplexing! I'm scratching my head here. Is there a way to toggle my layers instantly at all?

[Muirium]

Now this is getting good! I just tried the same logic on QMK on my Beamspring. Here's my keymap:

QMK on Beamspring.jpg (206.08 KiB) Viewed 19228 times

Guess what happens?

Almost exactly the same thing. Almost. The transition is still wrong, but subtly differently wrong now on both sides:

• When in Layer 1—normal arrow keys—press Fn+Esc then release Esc, Fn+; gets you ; until you release Fn then hold it again. Now it's Mouse Left as desired.
• When in Layer 3—mouse keys—press Fn+Esc then release Esc, Fn+; gets you ; again until you release Fn then hold it again. Now it's Left Arrow as desired.

In short: QMK is not the silver bullet.

Anyone know the TMK/QMK layer logic well enough to know if I can get the lazy Fn behaviour I desire? Having to release Fn just to "make it stick" doesn't feel right.

[BuGless]

Anyone know the TMK/QMK layer logic well enough to know if I can get the lazy Fn behaviour I desire? Having to release Fn just to "make it stick" doesn't feel right.

I'd say, the current behaviour is a bug, and should be fixed in the generic case to switch layers as soon as the Fn key is pressed. I can't imagine any benefit to delay the layerswitch until the toggle key is being released.