IBM 5160 oddball fault

[tony duell]

> 
> That was what i suggested.  it looks like the video memory, and infact
> most all of the memory operations are working or basic would not come
> up, or the bios codes for memory errors would occur.

Remember it does give the beeps for a video failure.

> 
> So the mechanism which sets up the 6845 is broken, with the skew of the
> location of the first message being significant, if it is a multiple of
> perhaps one or more bits stuck when the 6845 registers are written.

The 6845 and the logic on the CGA card must be good, it works in another
machine.

The 6845 must be initiallised essentially correctly, the thing gives a sensible 
video output. So it must be possible to write to the 6845.

The next issue is that block of garbage on the start of the screen. AFAIK all the memory 
on the CGA card is initialised by the ROM BIOS. So even if the 6845 was being mis-set, it
would have spaces, not garbage, there.


> I suspect the setup of the addressing of where the display memory is not
> relevant, so once you mess up where in the memory to fetch from, all the
> memory accesses are skewed but still reference the data the processor is
> writing to that memory (which appears to be correct).

I think the reverse. The CGA card is set up correctly but the CPU can't write
to the correct location in the video RAM due to an addressing problem. 
Possibly a faield buffer on the address bus somewhere.

since there is a lot of crap at the top which is not set up correctly,
perhaps there is a skew factor to be guessed at from that number of
lines or characters.

Isn't it exactly 1024 characters? This suggests a CPU address fault, not
a 6845 one to me.

-tony