Iron - circuit and INT corrections.

ALK'96                                    
                                          
           HOW TO FIX INT?            
                                          
══════════════ ════════════════════════════
Today in the section "IRON" we will touch on
one painful topic concerning (again-
after all!) imperfections of our domestic
ny schemesSPECTRUMa.                       
 I will briefly describe the essence of the problem: every user
caller"Leningradov", "Pentagon", "Eng-
Stremov",etc., being in good health
and a clear mind, I have repeatedly noticed some
Some "glitches" in programs such as VENOM
STRIKES BACK, MONTE CARLO CASINO,   SHOCK
MEGADEMO (MULTICOLOR, GREETINGS), CHASE
H.Q.128...  - the list goes on
a bunch of trainers for various games.       
                                          
In some programs you can observe periodic
wild disappearance or "trembling" of the moving
xy sprites, in others they turn out to be displaced
various stripes on BORDERe facing up,
although                                      
logically, they need to be
a little lower.                              
...All
this 
like  
go- 

external 
niy  
effect
                                          
                                          
                                          
                                          
 Now I'll tryexplain why
"this" is happening.  All glitchy effects
“they” (i.e. abroad) are written as
For example, on branded SPECTRUMax.             
 Such effects are performed synchronously with
vertical scanning of a television signal
- this technique is necessary for implementation
"smoothness" and "uniformity"  movements
sprites, tickers, etc. Synchronization-
tion of programs for redrawing objects from frame-
howl scanning in SPECTRUMax is achieved
by placing these programs in a handler
maskable interrupts, or by using them
execution in a loop after a microprogram command
processorHALT (which, in fact, does not change
essence).  Maskable interrupts occur
synchronously with vertical scanning at a frequency
50 hertz - to the input INT microprocessor
Short pulses with this frequency become dull.  
 The following is known: maskable interrupt-
development in PENTAGONe occurs simultaneously
with a personnel impulse, i.e. processing of pre-
tearing begins simultaneously with construction
introduction of a new frame.  In proprietary  SPECCY
the interrupt signal is "delayed" relative to
regarding the personnel pulse for a time, approx.
but equal to the construction of sixteen raster
lines.                                    
                                          
                                          
 Programs developed using proprietary
"computers", "configured" to "shifted" relative to
relative to frames INT, therefore, being launched
puppies on non-standard SPECTRUMax,  will
"show" the observer a picture while
moments of time at which the change occurs
drawing a sprite (meaning program
Many redrawings that do not match the appa-
military redrawing ofTV-frame).  And this is from-
fermented in the form of a partially "disappearing"
or “collapsed” image on one line
niya.                                      
                                          
                                          
 If the program provides all sorts of
BORDER-type effects MONTE-CARLO, then on
"bad" computers can still tolerate this
- just think, the stripes are at the top, but it works,
but if MULTICOLORs ?                  
--------------------------
 Lyrical digression...
  For those who don't know,
what  is  MULTICOLOR ,
Let me explain: this is a technique to obtain-
multiple colors -
more than two simultaneously in
one acquaintance. This ef-
fect is achieved   by
FULL synchronization of com-
puter with lowercase size-
twist of the TV signal, and this is
is achieved in turn
synchronization with frame
by signals INT interruptions
- another method in  SPEC-
TRUMesimply not.  
 Naturally, if your computer
non-standard phase INT signal, then you will see
instead of stunning splendor, what
-wide color stripes-
skis, flickering almost randomly-
tea way.  By and large
account,  normal work
MULTICOLOR-effects  depends
not only from the correct fa-
size of INT signal, but also from
schemes for forming raster lines in a computer
tere - I’ll come back to this problem in a moment
later...                                
                                          
------------------------------------------
                                          
When I first met BORDER-effect
fect in the program VENOM STRIKES BACK,
which, as you guessed, did not work correctly -
ok, I came to terms with it, but when I saw
(or rather, I didn’t see it normally!) SHOCK MEGA-
DEMO, namely MULTICOLOR in the form
it is perceived by abstract artists
nists, andGREETINGS, which I do not
I could see on the screen that my patience
It burst... 







 A month of searching for the optimal solution resulted in
in a small diagram given a little
below.                                     
 The essence of the operation of this device is
in the delay of requests for masked interrupts
signals coming from the computer circuit to
input INT of the microprocessor to a strictly specified
new value. Time delay value
can be adjusted. The scheme includes-
in the gap between the input INT microprocess
litter and the rest of the computer circuit.

               Chips:               
DD1 - K5SSTM2,                            
DD2 - K5SSЛН1 or К5SSЛН2,                
DDЗ - K555LAZ,                            
DDCH, DDS - K5SSUE7,                        
DDb - K5SSLA2, you can use both 155- and
1533 series of microcircuits.                     
                 Diodes:                  
VD1,VD2 - KDS22B.                         
              Capacitors:              
C1 - 10n, (may change during us-
triples),                                  
C2 - 1n,                                  
C3 - 10n.                                 
               Resistors:                
R1, R2 - 180 Ohm,                           
R3,R4 - 1 Com.                            
     Supply voltage layout:       
+5 volts to 14output DD1,DD2,DDЗ,DDb,k 16 
 output DDЧ,DDS,                          
"body" to pin 7 DD1,DD2,DDЗ,DDb,k 8  
 output DDЧ, DDS.                          
------------------------------------------
 Circuit setup.              
 You can manage to concoct a circuit using
plate of one-sided textolite sized
rum 25x75 mm. It’s not at all necessary to
conduct printed wiring - all  connections
can be carried out with stranded wireMGTF.
 For ease of setup, connect the outputs
counters DDCh, DDS with m/s DDb long pro-
vodka, to unused (according to the scheme) outlets
ladies counters also need to be soldered
wires - you may need them. Its
baud (and used too) m/s inputs
DDb must be connected to +5V via re-
resistors 1Kom. All DDb inputs are also connected
solder the wires. All these preparations
do not touch the connection of the 3rd pin of the DDCH with
1st DDb is a connection  can be done
immediately without extension wires - in pro-
You won’t have to tear it apart during the setup process.
                                          
 It is advisable to install capacitor C1 on
some distance from the chip housings,
to reduce the influence of temperature changes
air tours, at the same time there is no need
difficulty in removing it too far away
yanie - you will have to look for a reasonable company
promise  After setting up the circuit, the entire board
must be shielded with some kind of meta-
llical housing, connect the housing to
negative (case) terminal of the power supply
nia. However, the board can not be shielded
carry out - but this will affect the stability
delay time.                   
 Having assembled the circuit, it is connected   (observing
polarity!) to the computer power buses.
Contact pad leading topin 16
Z-80 (INT) is cut into the resulting
the gap includes the circuit. You can proceed
else - if your computerZ-80installed
linen on the panel, then carefully removing it
from there, bend back the same 16th pin, and 
put the processor onlocation, connection diagram
in this case we involve soldering
the input of the circuit to the contact pad, and the output
yes to the free 16th processor leg. If
you (happy?)  owner   PENTAGONa
or "Leningrad"with musical co-process
rum, which was “increased” on top of the main one
fees, then consider yourself very lucky
- you don’t need to “mock” yourself at all
processor and computer board, just
take out the coprocessor board and bend the 16th
expansion connector leg, solder to it
the output of the circuit, the input is connected, as in the previous
in the previous case, then install the board
coprocessor in place.                    
------------------------------------------
  Scheme settings.             
 After the “implementation” of the circuit in the
computer is complete, make sure everything is connected
dines are carried out in accordance with the diagram
my, that switch SA1 is located in
top (according to the diagram) position.  For the first
launchit is advisable to connect the outputs of the counter
kov DDCh, DDS with inputs DDb as before -
is shown in the diagram.                            
 Before starting the computer, stock up on dis-
kets with "suspicious" programs, in
which, in your opinion, are observed
"glitches".  Pick up at least five of these
programs and feel free to turn on your computer.    
 If after turning on the computer responds
on your button presses, then you can
we can confidently say that the functional scheme
works fine. If, on the contrary, the keyboard
If you don't listen, then try clicking
toggle switch SA1, if after that the keyboard
"comes to life", then you will have to look for errors in
connections of the assembled circuit, otherwise - you
"burned"  something   inside  computer -
Please accept my condolences in this case. 
 So, you download some program
mu, it is desirable that it be used
someBORDERnyeffect.  Click
toggle switch and watch BORDERom,  if
after turning the toggle switch to the “upper” position,
If the stripes move downwards, then you can
it is even more certain that the scheme
Ma not only functions normally, but also
copes with its direct responsibilities
mi.                                       
 Adjusting the delay time comes down to
setting the optimal amount of re-
bridges between the outputs of the counters DDЧ, DDS (3,
2,6,7) with inputs of the element 8I-NOT(DDb - 2,
3,4,5,6,11,12). Each installed re-
the sliver has its own “weight coefficient”
time delays, so for each below-
standing jumper (according to the diagram) delay weight
increases approximately twice as compared
disagreement with superiors.                      
 After "playing" with the jumpers, you will notice
those that add each new jumper
increases the delay (shifts BORDER-effect
fect down) and vice versa, and the “lower” pe-
straps have a greater influence on the charging time
holders                                   
 If the result isexperiments with jumpers
you just can’t set the exact value
due to the delay, it makes sense to “play”
capacitor C1, changing its capacitance to pre-
cases from 8 to 20 nanofarads - in this case
You can set the delay very precisely.     
                                          
                   ...                    
                                          
 Here we come to the most important thing.
Having configured the circuit for the first program, and
Having put another, you notice with surprise,
that BORDER effect in the new program
chickpeas This is especially noticeable if
you configured the scheme using the Academy program
(Tau Ceti II), and then loaded VENOM
STRIKES BACK (who hasn’t seen these programs
(???) - in them BORDER-effects are in
opposite places on BORDERE - on top
and below).  Looks very good
this is on PENTAGONe, which, as they say,
Xia, “there are no brakes at all.” Here we are
back to the lowercase problemscans...
__________________________________________
 Horizontal problem...      
 It turns out that our homegrown SPECTRUMs
have another implicit "glitch".      
So, during the construction of one line, the
tra hardware circuit of the video controller (in
original ULA) number of microprocessor cycles
processor for "Soviet" and "branded"SPEC-
TRUMob various. So, PENTAGONs have time
perform more operations than SPECCY,a
SCORPIONs, for example, vice versa.           
 This is influenced by factors such as real
high clock frequency of the microprocessor, so
and hardware circuit solution vi-
deocontroller.  Real clock speed
depends on the degree of "WAIT"-nesscom-
pewter, i.e. from hardware delays
you are the processor.  Such delays, in general
undesirable, but paradoxically,in
present SPECTRUMe these process delays
Sora exist. They arise when
tea, when the processor and ULA try to simultaneously
to directly access video memory - process
garbage in accordance with the action program,
ULA to read the video memory byte for
output to the screen. Preference in such a system
situation is given ULA, processor "slow down-
lives." The same thing happens if the program
mma is in the screen area - its speed
The interaction is reduced on average by 25%
compared to the “regular” program.  Pov-
I'm careful, this refers to the standardSPEC-
TRUMy and to some domestic models 
("Leningrad","SCORPION")... Rejoice, happy
cool (!) owners of PENTAGONob !!!  All
these lines said above do not apply to you
sit down. "Pentagon" is devoid of these "brakes".
Programs on "Pentagon" work the same way -
in all memory, and WAIT-leg process
litter is not involved at all.  However, for
everything has to be paid... at least
because "branded" MULTICOLORs even with
the scheme published here will work
only in one part of the screen (about 12
acquaintance).  "Leningraders", corresponding
surely, the picture will look better for
countWAIT-processor stops, but also not
will be perfect.                          
------------------------------------------
 Continuation of settings.          
 If you understand what I was trying to explain
in the last "insert", then you probably...
realized that it was impossible to film by any means
press "Pentagon" work as it is
currentSPECTRUM.  You can only bring the phase closer
INT-signal so that at least
partial"imported" MULTICOLORs and BORDER
-SCROLLERS. Therefore, it is better to configure the scheme
in the middle of the screen (MONTE CARLO) or by
the first line of the screen, i.e. on the first line in
screen area (SHOCK MEGADEMO, MULTICO-
LOR). There is another “criterion” for setting
INT-phases - by the absence of sprite "jitter"
com in CHASE H.Q.128 (main menu), or by
"presence" of the ticker in SHOCK MEGA-
DEMO, part GREETINGS.                    
 So, having configured the scheme to your liking, you can
you need to solder  short jumpers instead
debug, unused wires are unsoldered.
  Inputs to m/s DDb, which will be connected
only to +5V through resistors, you can
combine with each other and connect to those
the same +5V through one resistor.              
------------------------------------------
 Schema improvements.             
 The operation of the delay circuit can be stabilized
wat, if you put it instead of a capacitor
C1 quartz resonator at frequency 100-200
kilohertz But in this case, “typing” for-
Holding is possible only by “playing” with jumpers. 
  Accuracy and delay dosage range
can be increased if, in addition to jumpers, you use
use additional inverters (LN1, LN2)
including them between the outputs of the counters and the inputs
dami 8I-NOT DDb.                           
------------------------------------------
 ...                    
------------------------------------------
 Syndrome of domestic MEGADEMO construction. 
 Everything said below is in no way offensive
"our" coderob, take everything as a fact -
tion of the fact...                              
 It’s no secret to anyone in our country,
that  quantity branded,   real
SPECTRUMob for domesticusers
literally drowning in an ocean of mutants, collected
deep underground, stamped in
serial execution at various production facilities
national associations.  Palm
in the number of different “perversions” occupied
no (still!) "Pentagon".  Exists
a lot of domestic MEGADEM,which are PRIN-
CYPIALbNO can only work correctly
on PENTAGONe. Let's drop the theme of "mythical"
#FD-port is a separate global pro-
problem. Examples of this are darkness: ZY-ZY,INSULT,
SATISFACTION,PSG-WINS,MENTAL MASTURBATION,
ECSTASY,...                               
What can you do,PENTAGONhas become peculiar
standard for "binding" MULTICOLORob.    
 Some may find it strange, why is this here?
the author began to talk about our MEGADE-
Machwhen the article is not at all from that opera.
And I’m leading to the fact that it’s not without reason that the scheme
meINT-delay is a small toggle switch
SA1,which disconnects the circuit from the processor.
  This is done just so that  
it was possible to watch normally MULTICOLOR,
let's say in SATISFACTION,and inSHOCKe.      
 So, actually, we have come to the end.  
══════════════ ════════════════════════════
                                          
                   * * *                  
                                          
__________________________________________
                                          
          Instead of postscript:           
 For those who are not tired of reading this nonsense, I can
explain in detail how the described one works
scheme. You should stock up on a copy of this
diagrams on a piece of paper and turn over
page...                               
------------------------------------------
 How "it" works...           
 The entire scheme can be divided into four
re large blocks:                         
1. Trigger block (DD2.1,DD1.1, DD2.4,   
DD2.5,DD2.6,C2,R3,VD1):                   
2. Clock generator (DDЗ.1-DDЗ.3,C1,R1, 
R2):                                      
3. Counting scheme (DDCH, DDS, DDb):             
4. Output INT signal generator    
(DD2.2,DD2.3,C3,R4,VD2).                  
 When an “abnormal” request arrives,
jerking to input DD2.1 trigger state
DD1.1 is set to logical one.  This
will entail the launch of a clock generator
torus (5n.DDЗ.2=1) and unlocking the state
a counter with an initial zero state
niya (14n.DDЧ,DDS=0).  Generator connected
to the accumulation input (+1) of the DDCH counter.    
So, the counting process begins - at the output
dakh counters DDCh, DDS "sort out" code-
high combinations in binary eight-bit
code. This process will continue until
until the matching circuit assembled
on m/s DDb and jumpers supplied by you,
will not “catch” the last extreme combination
units. If, in addition to jumpers, you use
shafts and additional inverters (which is more
correct) then the edge codeword
(and therefore the magnitude of the delay) will
will be determined not only by the first “meetings”
"valued" one bits, but also zero bits.
If there has not yet been a match, then the output
DDb will be log.1 - “no match”.  When
coincidence of the required and current combination
Log.0 will be generated at the output. The fact of the transition
Yes, the output of DDb from 1 to 0 is registered
short pulse transmitter on the element
max DD2.2,DD2.3,C3,R4,VD2 - onoutput
DD2.3 is formed "delayed" INT-for-
request, which arrives at the input INT pro-
processor.                                  
 It should be noted here that this formation
vator   completely   repeats  scheme in 
PENTAGONe, so no reason to worry
about the length of a short pulse - it will be
kim (one must assume normal), as well as
before.                                   
After generation output signal
the clock generator continues to work properly
work with his next tact at the exits
counters, another code combination appears
nation one unit larger than necessary
May. In this case, the matching circuit (8n.DDb)
again it displays log.1 - “no match”.    
The fact of transition of the coincidence circuit from 0 to 1
"caught" by another co-former
short pulses (DD2.4-DD2.6, C2, R3, VD1),
at the output of which a  pulse is formed
log.0. This pulse drives trigger DD1.1
to the initial state - at the output Q(5n.)
log.0 is set, which stops
clock generator, output non-Q(6n.)
log.1 is set, which resets
counters DDЧ, DDS to zero state. Pos-
After this, the circuit is ready to accept a new request
millet... That's all.                       
It should be noted that in this scheme
there is one drawback - the amount of delay
will "float" for a period of time equal to
one period of the clock generator. Therefore,
it is advisable to set the maximum
possible frequency of the generator, and the delays are
use older jumpers.  If
Instead of capacitor C1, a quartz
new resonator, then there are no problems with “swimming”
there will be.                                    
                                          
------------------------------------------
                                          
           Goodbye again!          
                                          
══════════════ ════════════════════════════
                                          
                  * * *