[Coco] AT306 Mouse and Keyboard

[Paul Fitch]

> 
> The layout diagram in the AT306 manual shows the MDAT and 
> MCLK lines coming off the SuperIO chip, but the lines just 
> end.  No terminaion or grounds, or anything.
> 
> I pulled the data sheet off the web for the PC87303 IAT VUL 
> that is on the motherboard.  The KBDAT is connected to pins 
> 95 and 141 of the SuperIO chip.
> The KBCLK is connected to pins 96 and 4.  The MDAT is 
> connected to pins 94 and 142, and MCLK is connected to pins 
> 93 and 3.  The RTC is on pins 8-10.
> I don't have a big enough magnifying glass (the eyes are 
> going) to see any damage to the traces coming off the chip in 
> the area of pins 3-10.
> 
> Still, after further play, I'm sure the problem is somewhere 
> along in there for the following reasons:
> 
> 1) the Kernal error said it couldn't start the RTC, not that 
> the time was wrong.  
> 
> 2) The keyboards (AT with full size connector and the PS/2 to 
> AT converter) both showed the numberlock light on bootup.  
> That tells me that the keyboard is getting its + and - power. 
>  The only two lines remaining are the KBDAT and KBCLK. 
> 
> Question for all:  Would the RTC fail to keep any time at all 
> if the battery was dead?  And if theres no internal clock, 
> would that affect the KBCLK signal?
> 
> I made a mistake earlier, the CR2025 is a 3v lithium not 5v 
> like I originally thought (please refer to eyes going bad 
> statement above).  Anyone know if a CR2032 (which is the same 
> form factor) could be used in its place.
> I don't want to do any damage I can avoid, but I'm thinking 
> my problem with the keyboard could be traced to a dead RTC battery?
> 

Progress of a sort.  I put the CR2032 in the RTC battery holder and
attempted to boot the machine.  It still claims that it failed to start the
clock.  Then I get the logon prompt, and again, no keyboard response.  I'm
tapping keys and just getting frustrated.

So then my muse speaks to me, and I take a small screw driver and drag it
across the the three pins labeled mouse (remember this header is located
just south of the SuperIO chip) with me calling PIN 1 the closest to the
chip, and PIN 3 the farthest.  When the screw driver connects PINS 1 and 2,
a bunch of characters shows up on the line I was trying to type on.  Holding
the screw driver across PINS 1 and 2, I can type.  Take it off, and I see
nothing, put it back on, and the characters I just typed suddenly appear.

I then took a smallalligator clip and placed it across PINS 1 and 2, and
proceeded to logon to the machine as SUPER.  I'm now logged on.

So my question is, what are the odds of me doing permenant damage to this
machine if I place a jumper block across PINS 1 and 2 of the MOUSE header?
As stated above, the SuperIO chips lines for the mouse don't appear to be
terminated, and there are only two of them, so it looks like I'm shorting
MCLK across MDAT. 

Why would that apparently unclog the keyboard buffer?  Because if those keys
weren't being buffered, I wouldn't see them when the two pins were shorted.

I want to put a jumper block across those two pins and reboot and see if
that allows the RTC to work as well.