AEK II (Swedish/ISO) matrix

[intealls]

Figured this might be useful for some. Was pretty messy to map out, it had to be hooked up to read the columns from the decoder pins for some reason.

Serial number is AM0103JJM3501S. It's a 9x24 matrix, with dedicated pins for the modifiers+capslock (columns) on the controller. The rest of the columns are wired to a decoder and then the main controller. The ALT keys are joined together. I couldn't figure out a way to use the columns from the decoder pins (I'm not an EE), so I just took tapped in to them from the switch pads.

Try to ignore the heinous soldering and torn off pads, it's a testing board.

aek_decoder_cols.jpg (55.77 KiB) Viewed 3571 times

aek_rows_and_extra_cols.jpg (151.06 KiB) Viewed 3571 times

Just select all and copy/paste into notepad or something to get proper formatting.

Code: Select all

        1       2       3       4       5       6       7       8       9       10      11      12      13      14      15      16
r1      n9      --      NUMLK   SCRLK   PRTSC   F12     F10     F9      F2      F3      ESC     F1      F6      F7      F4      F5
r2      n/      n8      --      PAUSE   PGUP    BSP     F11     a0      a2      a3      <       a1      a7      F8      a4      a6
r3      n*      n7      --      --      PGDN    '       ´       O       --      --      §       --      a8      a9      a5      T
r4      n-      n6      --      --      END     HOME    Å       a+      --      --      --      --      --      --      G       Y
r5      n3      n5      RIGHT   --      DEL     INS     ~       P       W       E       TAB     Q       I       K       R       --
r6      nENTER  n4      --      --      UP      RETURN  Ä       Ö       --      D       --      --      --      L       --      U
r7      --      n+      --      LEFT    --      .       a-      ,       --      --      A       S       H       J       F       B
r8      --      --      n,      DOWN    n2      n1      n0      M       X       C       --      Z       --      N       V       SPACE

        *1      *2      *3      *4      *5      *6      *7      *8
r9      LALT    LGUI    LSHIFT  LCTRL   CPLK    RSHIFT  RCTRL   RWIN

[sixty]

What a weird matrix! 9 cols is quite rare. I'll compare this tonight to the German one, though I'm almost certain it will be identical.

Thanks for the info!

[sixty]

Btw, when I first converted my AEK to work on a PC I cut the matrix for capslock and soldered a bridge from Capslock to Shift. This way you can use that cool "lock" switch that the AEK uses in Capslock!

[intealls]

Btw, when I first converted my AEK to work on a PC I cut the matrix for capslock and soldered a bridge from Capslock to Shift. This way you can use that cool "lock" switch that the AEK uses in Capslock!

Cool idea! That lock key is great. Since the ALTs are joined, I decided to use them as the FN-key with the new controller (Teensy + dmw HH firmware). I put the lock key on the rightmost one, enabling arrow-key usage etc without having to hold down a key all the time (I cut the board MiniGuru-style). It's not finished though, gotta build a case for it.

9 cols is quite rare.

Corrected the original post to 9x24, got them backwards.

[Muirium]

Wee bit of a necro here, but a relevant one. I can confirm this is the same matrix on the Apple Extended Keyboard I as well as the II. I’ve got the ANSI-US model (typing on it now with Hasu’s ADB converter). Here’s a good writeup on how it works, including the ANSI version of the matrix:

https://ttic.uchicago.edu/~cotter/projects/aek2/

The matrix is really 16 by 8, with additional direct pin assignments for the mods, rather than a 9th row. That’s how Apple got around their cheapskate lack of diodes! It’s actually quite convenient for me, as the isolated power switch could be most handy.

The AEK I and II’s internals are similar but a bit distinct. I’ve got an AEK II in ANSI as well and compared them side by side. The original gangster has the same ICs, but helpfully socketed to make them easy to remove! There’s a lot more room on the older PCB in general, yet the matrix is one and the same. Saved me a lot of time mapping it all out the hard way for myself, because I have something more modern than ADB in mind for this big boy.

[Findecanor]

BTW, the Amiga's non-NKRO keyboards also had its modifiers on dedicated pins. They also purposely all had the same matrix and the matrix published in a manual so that software could be designed to reliably avoid ghost conditions. Most of the numeric keypad was on a separate row, which allowed that to be omitted on the tenkeyless A600.

I wondered if Apple similarly had used a subset of the large Enhanced matrix for other keyboards within the same era. So I checked a few columns on my M0118 with a multimeter ... and they did not match the AEK. But its modifier keys are on dedicated pins.

The ALT keys are joined together.

This sounded weird, so I checked this too on my Swedish AEKII. It is actually the Command keys that are joined together. The other modifier pairs are separate. My right Option key even has "Alt Gr" as a sublegend, which indicates that it was supposed to be used with PC software somehow... (emulator?)
Intealls must be using it with a Windows PC (or something), where the convention is to have Alt and GUI keys in the opposite order compared to Mac.

[Muirium]

Yup. IBM played similar matrix design tricks in the 2KRO Model M membrane. I’m sure everyone was at it, even on $200+ gear like this!

I was actually checking against the version in the link I gave, comparing Apples with Apples, ANSI with ANSI. There, the Commands are correctly described:

both command keys are connected to the same pin, and I wanted my software to be able to distinguish between them, so I cut the trace for the left command (GUI) key, and wired it directly to the teensy

Not sure I want to do anything so serious to my AEK. It’s nice to see them independently but not vital for me. I’d rather tell the Shifts apart, so I can use them chorded to toggle Caps Lock. TMK can’t see them, but they are in fact independent, internally. Here’s the skinny, from Apple’s vintage ADB documentation as linked by Hasu:

Some devices support more than one data protocol. An example is the extended keyboard, which can be asked to send separate key codes for the left and right shift keys. This change is accomplished by changing its handler ID to 3; the new handler ID reflects the new data protocol.

Here’s the ANSI matrix in all its ghostly glory, in case the Chicago U link ever dies:

Code: Select all

 	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	 
0	KP 9	numlk	 	scrlk	prtscr	F12	F10	F9	F7	F6	F5	F4	F3	F2	F1	esc	0
1	KP /	KP =	KP 8	pause	pgup	bksp	F11	0)	F8	7&	6^	4$	3#	2@	1!	`~	1
2	KP *	 	KP 7	 	pgdn	\|	=+	O	9(	8*	T	5%	 	 	 	 	2
3	KP -	 	KP 6	 	end	home	[{	-_	 	 	Y	G	 	 	 	 	3
4	KP 3	right	KP 5	 	del	ins	]}	P	K	I	 	R	E	W	Q	tab	4
5	KP enter 	KP 4	 	up	enter	'"	;:	L	 	U	 	D	 	 	 	5
6	 	 	KP +	left	 	.>	/?	,<	J	H	B	F	 	 	S	A	6
7	 	KP .	 	down	KP 2	KP 1	KP 0	M	N	 	space	V	C	X	Z	 	7
 	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	 

Something I don’t understand is where to actually hook up to the columns on the AEK I. I couldnt find a pin for any regular switch I tried. The way I checked the matrix was from the switches themselves, which I could patch to with my replacement controller if necessary. But it’s more elegant to use the IC sockets. I’m just failing to understand how that decoder works. Think I get the concept, I just can’t find the pins!

Another link I found helpful was this AEK II disassembly guide. Mine is good and tight and I wasn’t sure which catches to coerce once I had the lid off. Correct answer: the back ones.