Emailing: CP-M Z80 home brew computer circuit board.htm - test-drb@ccmp.vtda.org

17 Jun 2008

...

Subject: Re: Emailing: CP-M Z80 home brew computer circuit board.htm
   From: "bfranchuk at jetnet.ab.ca" <bfranchuk at jetnet.ab.ca>
   Date: Tue, 17 Jun 2008 10:09:33 -0600
     To: General Discussion: On-Topic and Off-Topic Posts <cctalk at
classiccmp.org>

Allison wrote:

 Have you ever used 8"SSSD to do anything that required space?  There 
 isn't
 enough space to run a disassembled version of the BDOS through ASM 
 unless you have at least two drives and don't mind doing cleanup.
 I never could run the CP/M system I was using for more than 30 minutes 
with out
the floppy stepper sticking. That was fun working with it.  
BTDT.. :(

...
 Still how do you transfer stuff from a floppy ( or file
online ) to a  
CP/M system with no external I/O devices? 
Explain no external IO?  

A serial line works fine, or a parallel bidirectional port.  Since every 
cp/m system has a console no rule saus it can transfer files on that line
either.

...
   Yes you can
roll your own data sep it only needs three ttl packages.  
 With all the other hardware needed for the 765 case you end up with at 
 least 10 chips
 though If you willing to miss a few features it's been done in 7 plus 
 the FDC
 and that doesn't include the bus side of the FDC interface.

That I would like to see. 
Goo look at David Dunfield's CUBIX system.  CPU is 6809 but the FDC is 
765 with minimal interface parts count.  I'd say it was a good example
of keep ing the chip count to a minimum.

...

 Kicad for linux there are other like cadstd for winders.
 Of course the last 2901 design I'd done in the early 80s was
 with paper and pen!  It's doable that way. I still am using paper & pen. 
Mind you I have to buy JAPANESE pens
since the quality of the USA stuff is all cheap Chinese imports.
Now that I have a serious cad program - DIPTRACE on order
I think I will put the design in hardware. 
I still use paper and pen too.  I do everything engineering 
notebook style.  That is black ink, no erasures and numbered 
pages.

...
   FYI using
250nS eproms will make it terminally slow unless
 you do two things, use a wide microword 64bits or more and
 pipeline the address and decode so you can work right to the
 eprom Tacc minimum limit. I got that covered if  I read the 2901 data sheets
correctly.
I will be using a 6809 style memory cycle optimized for
D-RAM access and clock the 2901 in 4th phase of
the memory cycle. This is retro computing project for the feel
of the late 1970's not using a 8 bit CPU. I am aming for
a 800ns memory cycle and the slow access of micro-code
is not a problem. The only thing pipelined is the next memory
cycle - Read/Write/Refesh and the default is refresh while
doing the the 1st cycle of op-code decoding. The front
panel does RUN/STOP, Single Instruction,Address Load,
Deposit,Read. 
Using Dram is one way to make life hard.  you can use larger 
static devices and save pain but retain the same cycle style.

Last 2901 design was 16bit, horizontal microcode (56bits), 
one microword to a cycle. I was able to get the pipeline 
for Eprom address latch to overlap so that the 2732s used 
would run at 320ns with good timing. The macro cpu executed 
instructions in that time cycle in a four tick cycle.  It 
was fairly fast for a single address machine. However, I 
didn't do the microaddress or micro program counter in the 
2901 I did that with TTL.  No front pannel either the 
system booted a rom minimal debugger.   When I got it 
working I junked it and started thinking about a 32bit machine.

Allison