MITS 8800B CPU Board
cctalk at randy482.com
Wed Jun 15 15:22:46 CDT 2005
From: "Tom Jennings" <tomj at wps.com>
Sent: Wednesday, June 15, 2005 2:51 PM
> On Tue, 14 Jun 2005, Randy McLaughlin wrote:
>> Please note that by using current RAM chips that are >= 64K address
>> decoding is extremely simplified. To carve out space from ROM not only
>> butchers the address map but adds complexity.
>> As I already said either an I/O instruction or an address line can
>> accomplish the same task.
> I think you missed the point, plus I left out one detail:
> * It specifically doens't require carving the address space up AT
> ALL. The machine simply has 64K contiguous RAM, from 0000h to
> * The EPROM board used PHANTOM to disable RAM when addressed -- I
> left that little detail out!
I understood, I was referring to the trigger for phantom. It does not
matter how the flip flop gets set as long as it does not interfere with the
initialization of the monitor.
I was agreeing that it is best to use phantom to access ROM rather than
address decoding that leaves a ROM window. In the early days of S100
systems mass storage was not the norm, so having a permanent ROM monitor
(some complete OS's with assemblers, etc) made sense. Another point is
there were no 64K RAM chips so address decoding had a different standard to
what anyone would build today. Today I prefer 512K RAM's so that address
decoding is replaced with MMU.
The methods of releasing phantom that I'm most familiar with are either a
bit in an I/O port so the ROM can be brought back in the address space if
needed or a particular addressing bit that shows the ROM is no longer
needed. Tarbell used address bit 5 to reset phantom, the boot ROM just
falls through to disable the boot ROM. Tarbell's boot ROM reads a sector
into RAM and when execution hits it phantom is disabled.
There are lots of ways to reset phantom depending on needs.
With Flash ROM's I prefer I/O port access so the ROM can be re-programmed
More information about the cctech