Iron
(C) Andy Nelson, 1998
Following the example Twinsoft and taking calls rasp, I, Andy
Nelson (as in passport Smolsky Andrew), also decided to describe
its PowerZX, it maxi-perverted "gadgets" and then I sunk to
this. We first give the parameters:
Motherboard ......." Lviv option "
(Serves as
frame, as
spend more than
tracks on the board and
some chips
are three stories:))
CPU ............... Z80H 3.5 / 7 MHz
71680/114688
cycles per frame.
Memory field ....... common and nezatenyaemoe
(As in Pentagon'e)
RAM ............... 512 kB
ROM ............... 64 kB (firmware changes
Nena: after the last
"Sewing" out there: BASIC-48
TR-DOS 5.05, STS
5.11A, RDS 3.1, Jemmi
ni Commander 4.1)
Virtual ROM ....... 64 kB (used
of RAM)
Number of lines
Vertical ...... 320
Cycles per line .. 224 (for 3.5MGts)
INT ..............." Pentagon "with the author
ically bounded
Niemi its length
Graphics mode .... STANDART - 256 * 192, 1916
colors, attributes, 8 * 8
MULTICOLOR - 256 * 192
16 colors, attributes
8 * 1
HIRES - 512 * 192, 2
color
Number of videos
page 3 ...........
Sound sources ... BEEPER, AY-3-8912, CO
VOX (9 bit)
Drives ......... 2x5.25 "Teac FD-55 (about
Dean of them was
1.2MB)
Joystick .......... Kempston (used
Standard Dendy-jo
ystick)
Mouse ............. Kempston Mouse
Printer port ..... ZX-Lprint
Story about what it is, why it
need and how I did it, I would like to start with a little
history of a primary alteration of my computer. Because the
study best for each other's mistakes, I hope,
underwritten information you ever
useful.
As I used Motherboard
charge of the "Lviv option", as in
later regretted, as this version
Spectrum, despite the great similarity with
brand has a lot of hardware defects. But since the other was
not given, then all my experiments were conducted on this
board. Frankly speaking, I "fuck up" two boards. First, after
the terrible experiments still "gone" for the parts, the second
"survived" and on that computer, I type the text.
First, make 48K 128K + AY was not difficult (six chip logic,
socket by AY, actually AY, a small shawl and
about 2 meters of wire). Craving for modernization came after
more demos "retard" or "not passed" very
(Especially Enlight'96). Also wanted more memory (up disks with
one drive was a very "fun" activity).
We first performed experiments on
Connection 565RU7. Computer, in principle,
earned, but on the screen to create "raving".
The fact that these RU'shki normally worked in "Scorpio" has
pushed me to think that they "do not like the" method of
reading information video controller. I removed "Inhibition" of
memory, resulting in image to appear, but a problem of
regeneration in nezatenyaemoy of RAM which, for unknown
reasons, could not solve. In addition, the VRAM For some
reason, there appeared a "point", and immediately after the
four of them in any place, with a fixed distance between them.
The intensity of their occurrence depends on the voltage
supply. The reason for this was not possible to solve for six
months.
Five months later, I got imported
analogues RU7 (i21256). They earned
"Shaded" mode to another board, but
traction to make a "netormozyaschy" computer took over. And now
after such modifications have a problem with "points" again.
The scheme, which, in theory, should have been working
properly, persistently betrayed a "point" in memory. But most
interesting was that the RU5-Tide worked fine.
After much testing schemes "ATM-turbo",
"Pentagon128" and "Scorpion", it was concluded that the
revamped "Lviv" option different one: as the address counter
circuits used 561 series. Then, after gating all addresses
RU-NIS signal-CAS (through IR23)
This problem has disappeared forever. Cause
same appearance of "points" was that the vertical counters
increment on signals obtained from the vertical sync. Due to
inadequate performance of 561-series, in the end, it led to the
fact that during the cycle of reading from RAM to change the
value on the address bus. RU5-Tide is "tolerated", but RU7-mym
it "did not like."
There was also a problem of regeneration in the nezatenyaemoy
"area of RAM. At first glance, everything seems to be there,
but the memory still fails. Get rid of this failed, decided
problem "in the root." Was converted addressing RU-NIS: all 16
chips now have a common address bus and common signals
And-RAS-CAS, as an attempt to combine pairwise data outputs,
ie, "make" operate two banks of RAM on a single register,
with separate signals-CAS, as is done in the "Pentagon'e" nor
to no good led. Chip did not get along "between
themselves, even if the signal is fed only
-RAS. I had to put another one on IR23
output of the second bank. In general, this option has some
advantages address: RAM properly regenerated, the signal-CAS is
connected to the RAM directly, which increases the noise
immunity of the computer, and even if the "burn" the memory,
the computer can work with one Bank of RAM (256K).
The following description is made on the basis of
the following sources: schemes
"Lviv option", "Pentagon",
"ATMturbo" or "Turbo-2 +, log
"Spectrofon 20" and the "ZX-Format 5. In addition, you should
be patient, considerate, beer ... :) And now let's start order.
Removal of inhibition of memory
Schema modifications result in 'Optocouplers "
problematic for two reasons: firstly,
it is difficult to draw a text editor, and secondly, a single
and final option scheme I do not, because
pattern during set-up adjustments, which are sometimes "forgot"
to record. But despite this, the basic "underwater
Stones I try to describe.
Of the three options "nezatenyaemyh" computers (Pentagon, ATM
and Scorpion) I chose Pentagon, since it is relatively easy to
implementation and subsequent installation
Turbo mode gives the best result.
Implemented almost entirely on the "original" scheme. In
addition, it is necessary to replace the "half" counter 561IE10
(D22.1) on the same counter 555'toy Series and ensure that it
is zero at the moment the arrival of a reset signal (k.7 D22.1)
(suitable 555IE7). It is also necessary to skip all other
address signals through 555IR23 which should be clocked
signal-CAS. This excludes the possibility of entering danyh of
video RAM through KP11 (D32, D33) on the data bus of the
processor.
Distribution of pages within the address
space of the processor, I did two
555KP12. On the address inputs of multiplexers
A14 and A15 are fed and given configuration
pages, the output is the address bus. This method allows
implementation of such a contraption as the Virtual
ROM. In a nutshell, the service mechanism
Memory is easily implemented on 555ID4 so
as is done in ATMturbo. When commissioning
often helps to "obveshivanie" 50-200 pF capacitor signal-RAS
and-CAS. Must also "hang" blocking capacitors on power supply
of RAM (preferably not less 0.15mkF on each chip).
Turbo mode
Some initial recommendations.
It is best to install the processor socket (only good to you
in Subsequently, it was not a problem). Should
also take into account the fact that in turbo mode do not work
for all processors. Single copies like it is working, but
"bent" connecting the periphery, as data bus and address of the
"Lviv version" is not buffered. The best option - Z80H, mozhno
try Z80B. Z80A and domestic counterparts do not work all.
Sometimes do not work in turbo mode ROM-old shki
registration (before 1984).
Turbirovanie scheme is the same as for
Pentagon'a (3 circuits), except for one:
No need to install an element of TM2
(D107.2), which excludes the possibility of
dual circulation of the processor to the memory of
reading time of one video controller
familiarity (4-cycle read). In addition,
should submit the switch "Turbo" with
controller drive signal-VG93.CS,
well-DOSEN. When working TR-DOS turbo mode
will be disconnected, but it does not matter
as in turbo mode boot faster
still no go.
Does not work correctly in turbo mode
AUTOchecking drive many boot'ah (k
Unfortunately, it is eliminated only with the program).
Virtual ROM
It is a pseudo-ROM. The meaning of this 'feature' in the
following: a bit of a special port off the main ROM, but instead
included one of the last pages of RAM
(# C5 - TR-DOS, # C6 - BASIC128, # C7 - BASIC48). Entry into a
pseudo-ROM addresses 0 - # 3FFF not performed. Record any data
in the pseudo-ROM would be to include a page at the following
addresses # C000-# FFFF. Port which includes the pseudo-ROM has
a soft "latch" (as the port # 7FFD), need relief which the
signal-RES set switch. The result is an interesting thing: you
can download any firmware BASIC48, BASIC128 and TR-DOS,
Engage the port configuration to translate
the computer to reset and disable 128K
port latch configuration for signal
-RES. We get a full Speccy-128
loaded firmware change that
software is already impossible. Realized this
"Gizmo" in the presence of two KP12, which specify the page
number for each address "quarters", is relatively simple. On
configuration inputs for the zero-KP12 quarter of address space
served signals ROM.A14 and DOSEN. On the other input is
selected numbers of the last pages of RAM implement port #
EFF7, mutes, and repeatedly turning the main ROM on a certain
signal, we realize the system RAM and a write lock in memory
addresses 0 - # 3FFF. Assigning individual bits of the port
# EFF7 will be given below.
Increase "frame length"
This has, historically, that all
Spekrum-like machines contain a raster
312 lines, and Pentagon - 320 lines. Because
these eight lines, some demos, written in the Pentagon, the
brake on your computer, which have only 312 lines (in
particular, and Lviv option "). Solve the problem is simple: to
"the brakes" vertical counter to eight rows. Install an
additional count of 8 and design the system locks the vertical
counter. Recommend the "stretch" the bottom border, if you do
not want to have Problems with the arrival time INT'a.
The problem of signal INT
More precisely, two of them: the arrival time and
its length. On how to make a "firm"
and "Pentagon" INT spoken and written much (even the demo
special was) and I will not dwell on it separately. If you want
to know how to make a right INT for "Lviv
form ", I advise you to read to help Ram
Doctor-in Tyatina Sergei. Problem length
INT'a solved by selecting the desired rezistora or capacitor.
But this is true only if you do not have Turbo mode. If it
exists, an interesting problem: to pick up
length for 3.5MGts, switched on 7 - INT
became a "longish" picked up for 7MHz - at
3.5MGts keyboard works through time.
The first "threw" the idea of how to deal with it, Valentine
(Twinsoft). He suggested use the fact that the Z80 during
processing INT'a simultaneously establishes
"0" signals-IORQ and-M1. Thus,
put a trigger lock on the signal INT and
doing so he placed the signals-IORQ and-M1, and reset any
another signal, coming at a time
when INT is inactive (eg, signal BORDER). In general, the
scheme is the following:
INT X to 16N. Z80
D1.1
1
+5 V oR
Q
GND D
-BORDER / C
D1.2
oS
-IORQ 1
-M1 D2.1 D1 - 555LL1
D2 - 555TM2
After such alteration problems INT'om,
in Turbo Mode, disappear forever.
>>
