Optron #33 20 ноября 1999

Железо

Iron - The problem of addressing ports ZX Spectrum.

          On the question of addressing

{} S. Filimonov, 1999


  These short notes are inspired by the article
"On the connection of external devices to the ZX
SPECTRUM ", published in" Polesie "| 13.

  Indeed, writes SSSR, author
this article - when a new
external device (ED) to the computer
Users sometimes find a big bummer.
The reason for this, the author sees is that every device I / O 
in a computer at enhancing the signal-IORQ is committed to 
shove on the data bus every gag.


  I should note that this statement is not
quite correct, because the processor, referring to any "input / 
output devices, not only activates the signal-IORQ, that, in 
fact, evidence of the with the clause, and, even, not only the 
address of the device itself, but also one of the signals -RD 
or-WR, talking about reading from device, or write to it, 
respectively. 

  Based on the foregoing, we can say that these conflicts, in 
theory, occur and should not be, but (we know, as we all do), 
saving on any building circuits, our

"Nice" developers often make
so rough decoding ports that
Adding to this computer every VU
involve alterations, which could be substantial, most of the 
computer. 

  And now - continues SSSR, - a victory in
the conflict between peripherals depends on their priority, due 
to, in turn, Circuit Implementation of specific models

computers and UW.

  But how do you want to understand this expression
- "Priority"? What, Sound Drive - more
the importance of the device than, say, ZX Lprint?
Or, perhaps, meant that "higher priority" has a clause, whose 
output is a more powerful chip? Or maybe something that has the 
most "curve" decoding address sampling? .. 

  In one of the computers on the factory (!)
production I've seen so rough decoding Kempston Joystck'a that 
thereof computer in general would not have to work. But it 
worked! How? It's very simple - data port on the data bus went 
through resistors, "extinguishing" conflicts. (Just did an 
analogy with have got as a something to me in the repair of a 
Chinese player, which goromkost controlled variable resistor, 
stood consistently with batteries). 

  Another example of this approach is associated with
upopinaniem Article circuit controller
kempston-mouse, published in "Optocouplers" | 13 (for which 
SSSR'u, of course, thank you). After all, what SSSR was forced 
to "borrow" from our scheme of the two input flip-flop, just, 
and was caused by bad faith "branded" Developers! 

  The whole root of all evil - it is in such a devil-may-care 
attitude "firmachey" to their offspring.


  As a way to address SSSR
offers for a sample of one of the UW to block all the others, 
so they do not "get underfoot." Benefit from this, he sees 
that, unlike in the correct decoding clause, it does not 
require him to sculpt a dozen or two or more diodes buildings 
shallow logic, and besides - conduct a wide loop, overloading 
Address bus.


  Oznachennuyu blocking I / O ports is proposed to implement 
signal IORQGE.


  Well, in some particular case, it
maybe will help. But! If the signal
IORQGE taken from each connected to
UW computer, then the conflict
can not be avoided: a rough address decoding
This signal will be activated simultaneously by
several peripherals.

  Therefore, in my opinion, absolutely not
have nothing to block! Each VU has
your address, which should be properly selected.

  Only exceptionally is correct,
Detailed decoding addresses may decide
this problem.