Conference 7.286::atarist

    The HALT-state is a 'normal' state of the 68000 when there is something
    badly wrong.

    The CPU goes into this state when it detects a severe bus problem. It
    also disconnects itself from the bus. This is to protect itself from
    being damaged by faulty chips.

    This once happend to me after I replaced the 'LS373 driver chips in a
    MegaST. The system did not boot anymore and the cpu halted immediately.
    This can be seen with an oscilloscope. The halt line is high
    (=inactiv) for a short moment when you press reset. In my case I had a
    tiny little solder bridge between two address pins.

    I do not know what happened to your system. Did you do some thing
    'under the hood'? You might be able to track this with an ohm-meter.
    Disconnect the Atari from the mains and check all the address and data
    lines against each other. You should always have at least some tens of
    kilo ohm resistance. If you find a very low reading of your meter or
    even a short then that might give you a clue

    You might want to give some more info about how it died so that we can
    gie you more help...


    ---markus---

    Hello Markus
    
    I have installed some home made hardware in this machine: chip holders
    for eprom, home designed and home made 2 mb memory extention (I etched
    the PCB myself). All those extentions have been working some times
    (about 2 month) with the machine closed.
    
    It all happened when I did switch on the machine: a serie of vertical
    stripes across the screen and no access to the floppy. I thought to
    myself: same as during memory installation, must be some data line
    shorted to ground. I open the machine, switch on again, still no
    access to floppy but also no more vertical lines, only a black square on
    white background (color monitor). Mono monitor does not display
    anything...
    
    Halt line follows reset line and stays low long enought for a few (how
    nany I do not know) access to eprom. Halt line is active and seems to
    switch on at the same time as a berr activation. There is no short on
    data/address lines.
    
    I guess I have now a dead chip somewhere on the bus. I will replace the
    cpu chip as it is not really expensive and that will permits me to
    have it on chip holder. Then if that does not fix it I guess I will
    bring it to a dealer (I do not want to buy all atari custom chips).
    
    Thank anyway.
    
    	Louis Richard

I thought I wrote this message already, but I must have gone home
without hitting "return" enough times.

The symptoms you describe are exactly what happens when there is bad
memory in the beginning of memory.  When the 68000 goes to process and
address error (very likely when memory is bad) and it get's an address
error while following the interrupt vector, it halts.

The fix for this is to replace the bad memory.  In your case, you may
also be looking at shorts, etc, in your own board.  The way I identify
the bad RAM is to:

	- Attach a color monitor.

	- For each of the 16 RAM, ground and release pins in the
	following order:

	- GND the DI or DO pin (they are bussed, it doesn't matter which)
 	A black line will appear on the screen at the pixel positions
	driven from this RAM.

	- GND the R/W pin.  The rest of the screen will go white.

	- Relase the R/W pin
	The black line shoudl stay.  If it doesn't, this RAM is bad.

    Thank jeff your procedure did give me some more thoughts. I did try it
    and got:
    
    	I see a black square with size about the size of the usefull part
    of the screen. Background is white
    
>	- GND the DI or DO pin (they are bussed, it doesn't matter which)
> 	A black line will appear on the screen at the pixel positions
>	driven from this RAM.

    I see a WHITE line
    
>	- GND the R/W pin.  The rest of the screen will go white.
    
    No change

>	- Relase the R/W pin
>	The black line shoudl stay.  If it doesn't, this RAM is bad.
    
    same reaction on any chip...
    
    Pull my scope again and look at memory chip signals:
    RAS0 and RAS1 goes periodicaly low
    CAS0L, CAS0H, CAS1L and CAS1H are hight all the time... So that, if I
    understand well, a constant value of zero is returned to anybody triing
    to access memory. Could that explain the cpu going to halt? 
    Could the MMU be driven to that state or is something broken inside
    MMU? (Wishfull thinking: I have the failing FRU!!!)
    
    	Thanks again for your help
    

I may have had black and white transposed.

CAS should be pulled by the dispaly controller occasionally.  The RAS
you see means that the refersh generator seems to be working.  If the
processor is halted, the only CAS you will see will be the video
subsystem.

You seem to be on the right track.  I'd suggest swapping MMU and DMA
controller, one at a time, from a working system.  (Better yet, swap
yours into the working system and see if the problem moves with it.)  Do
you have any friends with more loyalty than brain, that would allow you
to disassemble their computer?