XT/AT/PS2/Terminal to USB Converter with NKRO

[Retrete]

I didn't do any ground connection  , also I have no idea who to ground it, If I remember well you mentioned something about american tape, like this ? http://www.amesasuministros.com/productos/1533971379.jpg

[Soarer]

I know it doesn't look like much, but I think that gives me something to go and investigate! It's an error that's preventing the config from being activated on the Teensy. (So maybe leave the resistor pack for now, but I think it will have to be removed eventually).

Making a ground connection can just be with another wire - from one of the GND pins on the Teensy to a ground pin on the keyboard. To find a suitable pin, you'll need to find a pin that is joined to the large area I marked on a previous picture - each of the chips will have at least one, on one of the corner pins of the chip. I think the one with a double circle, most likely...

[Soarer]

OK, this .hex should fix that problem. Let's see what the next one is  )

[Retrete]

Now it seems to recognize the keys, but not all.

(cross ones not working, maybe a soldering issue)

[Soarer]

The two shifts and the top-right one won't work until ground is connected.

I'd guess that the other 8 are all on the same strobe line, so yes, quite possibly a soldering/conenction issue there, or maybe a config issue (but I think there's something in the config for every pin you've connected). The first question is, which strobe is it? Looking at the first digit of the codes being printed, which one of 0,1,2,...,8,9,A is missing?

[Retrete]

This weekend I will resold the whole teensy, It doesn't deserve his actual look 

[Soarer]

Be very careful if you do that! It's easy to lift a pad from the Teensy PCB when desoldering

The missing code will tell us exactly which wire it is, so really only one needs a little attention. It could be the solder, or maybe the socket at the other end.

But anyway, it's a promising start

[Retrete]

It was the socket at the end, now all the keys except shifts are working.
I solder the ground pin and tried with the 4 pins marked, no one works. But not only those 3 keys doesn't work, all the rest also fail 

How I must edit the rawhid.dll?

Thanks!

[Soarer]

Fail only when the wire is connected, or still failed with it removed?

Sorry, I meant that you should test with a multimeter to see where is ground! Or if you can see that there is a connection from that ground area to any pin (perhaps not one of the ones I marked).

rawhid.dll?

[Retrete]

Fail only with the wire connected. I'll test it with the multimeter.

If that all looks good, you can start editing a new config file, replacing all the UNASSIGNED entries with useful HID codes (such as A or SPACE, see the Codes page of the docs). So you press Q, see what the code is, plug Q into the config at the right place. Rinse and repeat!

Thats the reason why I mentioned the rawhid.dll, I don't know which file is  )

One thing I dont understand, this what the hid_listen capture;


Each key must have an individual code? 

[Soarer]

Oh, the Soarer Config file... is kevex.sc. It's just a text file.

I think that resistor pack I mentioned needs to come out. It seems to be effectively shorting all the sense lines together, because you should only get one \ and one / per key.

[Retrete]

Resistor pack = ?

Only remove the cilindric ones? Should I cut them?

[Soarer]

Nono... only the greeny-white one at the end. It looks like it's in a socket, so remove it like the chips.

[Retrete]

Done, now it only show 2 digits  Now Im trying to find the ground pin

[Soarer]

Done, now it only show 2 digits  Now Im trying to find the ground pin

Cool! The ground might still turn out to be where I thought, since that resistor pack was getting in the way.

Now that the matrix is working, you could add a line to the matrix section of the .sc file to activate the blocking function, because the matrix doesn't have diodes. It can go anywhere, but near the beginning is logical...

Code: [Select]

matrix
    scanrate 1
    debounce 5
    blocking

    sense           PB0         PB1     PB2         PB3     PB4     PB5         PB6         PB7


[Retrete]

With the pin connected to the double circle on(top right), the hid listen detect the keys , but shifts and reset still unnoticeable.
I've tested it with the multimeter and the double circle pin with the shift goes to ~0.

[Soarer]

Hmm, 'unstrobed' isn't working for me either - I must've broken something in the code

[Soarer]

And indeed I had - simply hadn't joined up two bits of code...

[Retrete]

YES! Shifts and reset keys now working!!! 
Later I'll try to map the keys, hope it works

Thanks again! W/o you this wouldn't be possible

[Soarer]

Excellent! I'm not sure if you're lucky or unlucky to be the first... anyone else can use all these posts as a guide

[Retrete]

Hmm, I don't understand pretty well how to config the kevex.sc, for e.g I press the "a" key, hid_listen shows;
\01
/01

Now I go to the kevex.sc, open with the notepad, and see this;

# Matrix config for kevex ray 7000CM

matrix
   scanrate 1
   debounce 5

   sense         PB0         PB1      PB2         PB3      PB4      PB5         PB6         PB7
   strobe   PD7      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PE0      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PE1      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PC0      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PC1      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PC2      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PC3      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PC4      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PC5      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PC6      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED
   strobe   PC7      UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED   UNASSIGNED

   unstrobed -PD0   UNASSIGNED
   unstrobed -PD1   UNASSIGNED
   unstrobed -PD2   UNASSIGNED
end

[/spoiler]

Strobes and PB define the matrix(?) Where I have to put the 01? in which form? 01 A , 0x01 A, 0x04 = 0x01 ?
For e.g;

 

       unstrobed -PD0   0x058 Reset
   unstrobed -PD1   0x059 LShift
   unstrobed -PD2   0x05A Rshift

?

[Soarer]

The thing to enter is a 'Name' from the Codes page of the docs, so for the shifts...

Code: [Select]

    unstrobed -PD1 LSHIFT
    unstrobed -PD2 RSHIFT

The matrix position is worked out from the code such that 00 is top-left, increasing along the first row until 07. Then the second row is from 08 to 0F. Then the 3rd and fourth rows are codes beginning with 1, and so on.

Again, just put a name in each position of the matrix. Many are obvious, so you won't even need to look them up in the Codes list, e.g. A, B, C,... 1, 2, 3, ... F1, F2, F3, ...

So if you found that 00=A, 01=B, 02=C, ..., 07=H, then you would enter them in like this:

Code: [Select]

   sense         PB0         PB1      PB2         PB3      PB4      PB5         PB6         PB7
   strobe   PD7  A           B        C           D        E        F           G           H

To give another example, this is the config I made for my mini keyboard:

Code: [Select]

led caps -PC6
led scroll -PC7

matrix
    scanrate 1
    debounce 5

    sense           PB0         PB1 PB2         PB3 PD0 PD1         PD2         PD3
    strobe  PF0     ESC         2   4           5   7   9           MINUS       LCTRL
    strobe  PF1     1           3   UNASSIGNED  6   8   0           BACKSPACE   LGUI
    strobe  PF4     TAB         W   R           T   U   O           LEFT_BRACE  LALT
    strobe  PF5     Q           E   BACKSPACE   Y   I   P           RIGHT_BRACE UNASSIGNED
    strobe  PF6     CAPS_LOCK   S   F           G   J   L           ENTER       RALT
    strobe  PF7     A           D   SPACE       H   K   SEMICOLON   UNASSIGNED  MENU
    strobe  PB6     LSHIFT      Z   C           V   N   COMMA       FN2         RCTRL
    strobe  PB7     FN1         X   UNASSIGNED  B   M   PERIOD      RSHIFT      UNASSIGNED
end


[Retrete]

YESSSSSSS! I'll post the good video when finished  .

[Soarer]

Haha, that's fantastic! You're welcome!

[poxeclipse]

That's great. I watched it from the beginning, and looks very nice. We can try those Chinese/Japanese IBM keyboards now. Thanks to Soarer & Retrete.

[Retrete]

GOOD video 

feature=youtu.be

[Soarer]

I didn't realise you were going to do that with a screen, and keeping the whole casing... it's very cool indeed! 

[Soarer]

We can try those Chinese/Japanese IBM keyboards now.

Indeed, I should be taking mine apart to see what's what soon.

I'm currently mulling over what options to make available for the first proper controller release - a tricky balance between putting in as much as possible, but nothing that I might want to change later (legacy support is painful)!

[Retrete]

Soarer, If I press some keys together(general more than 3) some unpressed keys activate, is this normal ?

[Soarer]

Soarer, If I press some keys together(general more than 3) some unpressed keys activate, is this normal ?

Did you forget to add this line to the config file? 

Now that the matrix is working, you could add a line to the matrix section of the .sc file to activate the blocking function, because the matrix doesn't have diodes. It can go anywhere, but near the beginning is logical...

Code: [Select]

matrix
    scanrate 1
    debounce 5
    blocking

    sense           PB0         PB1     PB2         PB3     PB4     PB5         PB6         PB7


Or did I break it?   I did test it once, but probably before the beta3 version.

[Retrete]

If I put the "blocking" ;
error at line 6, invalid arguments

Do the "blocking"need to be together something? Or it just go like you wrote it?

[Soarer]

Oops, sorry, it needs a 1 after it... blocking 1

[Retrete]

Now compiles w/o problems, but still the same "issue"

[tufty]

Soarer - I've not been watching this closely, but if I gather correctly, this is "hardwiring a keyboard matrix to soarer's code", right?

If that's the case, how many keys do you reckon I could handle on a Teensy 2.0 (not a teensy++)?  I'm thinking the ADB circuitry in the M0116 might just go the way of the dodo

[Soarer]

Now compiles w/o problems, but still the same "issue"

Found the problem, it was in scas...

[Soarer]

Soarer - I've not been watching this closely, but if I gather correctly, this is "hardwiring a keyboard matrix to soarer's code", right?

If that's the case, how many keys do you reckon I could handle on a Teensy 2.0 (not a teensy++)?  I'm thinking the ADB circuitry in the M0116 might just go the way of the dodo

That's right, this will be 'SoarTroller' or whatever 

Number of keys depends on the size of the matrix - especially for a non-NKRO matrix it's quite common to have a few gaps, e.g. so that modifiers don't ghost with anything else. Teensy 2.0 might be enough for the M0116, but perhaps only just - a quick count of lines here gives maybe 24 (8x16?), which is pushing it and leaves no room for LEDs.

[Fawksie]

Another satisfied customer, thanks Soarer

This is a 1986 Model M 122-Key Typewriter Keyboard. A few months older than me.



Wiring isn't particularly well thought out. I could have made it much neater looking with some more effort and willpower. No-one can see it in a box anyway! Spare pins on the 10 pin connector are for lock LEDs.



More photos

[Retrete]

Now compiles w/o problems, but still the same "issue"

Found the problem, it was in scas...

Now it detect even less keys  , muuuuuch slower when I press more than 2, but I think Im making some part of the process wrong, steps I did;

- Open teensy.exe , load hex and reboot
- Drop kex.sc (already mapped config) to new scas.exe
- kex.sc.scb created , load into scwr.exe
- Open keyboard tester

The problem now is the first part, I need to press the reset button and opening again the case its like a nightmare, is there a way to avoid pressing the key?
Dropping kex.sc now result in the 1/10 sec screen and no kex.sc.scb created.

Does the scaswr.bat depends of the actual scas.exe and scwr.exe on the folder?
Also observed,  If I remove the blocking 1, while having the same issue Im able to press much more keys.

[Soarer]

Now compiles w/o problems, but still the same "issue"

Found the problem, it was in scas...

Now it detect even less keys  , muuuuuch slower when I press more than 2, but I think Im making some part of the process wrong, steps I did;

- Open teensy.exe , load hex and reboot
- Drop kex.sc (already mapped config) to new scas.exe
- kex.sc.scb created , load into scwr.exe
- Open keyboard tester

The problem now is the first part, I need to press the reset button and opening again the case its like a nightmare, is there a way to avoid pressing the key?
Dropping kex.sc now result in the 1/10 sec screen and no kex.sc.scb created.

Does the scaswr.bat depends of the actual scas.exe and scwr.exe on the folder?
Also observed,  If I remove the blocking 1, while having the same issue Im able to press much more keys.

This will annoy you... you didn't need to reset the Teensy
It only needs to be reset to load a new firmware (.hex).
And double-clicking scboot.exe will reset it, if my firmware is running. (It can sometimes take a while for the Teensy loader to find the Teensy after that, but only some seconds).

scaswr.bat simply runs scas.exe and then scwr.exe - those exes need to be in the same folder as it.

Droppping a .sc file onto scas.exe will just print an error message and disappear in a split second.

Dropping a .scb onto scwr.exe should work, but again it will disappear in a split second so you won't see if it has worked.

Both of those are command line utilities, so use scaswr.bat if you want to just drop the .sc file.

I admit that the blocking function is a bit severe - perhaps it could pass a few more keys in certain circumstances. I'll probably work on that once I've hooked up a non-NKRO matrix myself.

The slowness sounds like a problem - what sort of slow are we talking about?

[Soarer]

Another satisfied customer, thanks Soarer

Very nice build! The wiring looks pretty neat to me

[Retrete]

Blocking 1 doesn't activate non pressed keys,but If I try pressing near keys of the already pressed it doesn't work.
Blocking 1: no pression error and works well with non near keys
No blocking: pression error but all keys working.

[Soarer]

Unfortunately, those are the options when the matrix doesn't use diodes to prevent the erroneous key presses

I might be able to improve it, but only slightly - are there combinations that you want to use that are prevented with blocking 1, but work fine without it and with no erroneous key presses?

[Retrete]

No no, It doesn't matter, it works fine now. I don't press more than 2 or 3 keys at the same time, and the normal 3 or 4 keys combos on shooters works fine.
No more fail clicks 
Thanks a lot for your help!

[Soarer]

OK, cool! Any combination of two keys should work, but there will be cases where a third key gets blocked. The shift keys are never blocked because they have their own connections, but whatever you've mapped for controls and alts count towards blocking, since they are part of the matrix. Worst case, something like ctrl-alt-del might not work!

[tufty]

Soarer - I've not been watching this closely, but if I gather correctly, this is "hardwiring a keyboard matrix to soarer's code", right?

If that's the case, how many keys do you reckon I could handle on a Teensy 2.0 (not a teensy++)?  I'm thinking the ADB circuitry in the M0116 might just go the way of the dodo

That's right, this will be 'SoarTroller' or whatever 

Number of keys depends on the size of the matrix - especially for a non-NKRO matrix it's quite common to have a few gaps, e.g. so that modifiers don't ghost with anything else. Teensy 2.0 might be enough for the M0116, but perhaps only just - a quick count of lines here gives maybe 24 (8x16?), which is pushing it and leaves no room for LEDs.

Well, LEDs I can live without.  There's no LEDs on the board right now, after all

And if I were to bin the circuit board and hardwire my own NKRO matrix with diodes and all the works? Yes, that would really make it a custom keyboard in a retro case.  I think I could probably get away with doing so; the switches are plate-mounted, so the PCB isn't doing much structurally, right?

Alternatively, I suppose I could get back to hacking on Hasu's code to improve the ADB side of things.

[Soarer]

Some plates are pretty flimsy, and/or don't hold the switches firmly, so you'd need to check. You could still use the PCB as structure, just cut lots of tracks - you might be able to leave all the column (or row) tracks as they are, and just isolate the other side of each switch so you can add in a diode and rewire just the rows (or columns).

[Retrete]

I have accidentaly broke one of the pin sockets of the teensy, this one;

Is it possible to reconect it to another free-socket of the teensy? maybe D6?

[Soarer]

Yep, just rewire it to another pin. But not PD6, since it has the onboard LED attached to it already. PD5 would be fine, for example.

Then reconnect the USB, change the config file to match the new pin, drop it on scaswr... and it should work!

The only reason I picked out the current set of pins was to keep the wiring simple

[Retrete]

PD5,PD4 or PD3 also doesn't work, but PD7 (already asigned to a column) works, could it be the hex loaded onto the teensy?

[Soarer]

Did you edit your .sc to change the relevant strobe line, and load it back on?

I used those 3 pins for LEDs on the converter, so maybe there is a bug and they're not fully disabled. Any of the PF pins wouldn't be affected by such a bug, so you could try one of those. Otherwise I'm pretty sure the .hex is OK.