Iron - The Story of the Pentagon 1024SL.

<b>Iron</b> - The Story of the Pentagon 1024SL.
   The Story of a pentagon

   We will talk about the new reincarnation of penta
estrus, known as the Pentagon-1024SL, when
Think and sold by fellow King Of
Evil'om in the world bearing the name of Alex Zhabin.
More information about the project told his website:
 http://pentagon1024.narod.ru

   After learning about this project, I decided to buy
author naked fee, and after buying some
potormozil and finally assembled the computer.
Whereof action full grablehozhdeny and yar
FIR discoveries, I decided to tell on the page
 ples Info Guide Number 9.

   I'll note that I tell about the versions
PC 1.41 - that which is collected on
rassypuhe. At the moment, King Of Evil
introduces version 2.x, which is the basis
 are two PLMki.

   So, here we go:


         I. Soldering iron - your enemy!


   Here it is, shining moment! Before you lies
fresh, but the bare board, a in Sidir in
poke sidyuk issued by King Of Evil'om (in
Further, in order to save life the keyboard
algebra, just KOE:) you with the card, and
which there are principled and Mon
platemaking scheme, a list of chips and other non-
crawled in the assembly of elements pikadovsky
File wiring boards and other erundistika.

   First, of course, it's worth checking fee
visually with a magnifying glass, as recommended KOE.
Next - go to the market / in the shop and bought
tsya detalyami.Odnako should not rush until
tsya better carefully 'povtykat' in principle
tally chart - just in case there is something
superfluous or not necessary to you?

   For me over was the controller, we
shki (throwing DD65, DD66, DD67 and obvjaza
ku; numbering here and below - on the principle of
exponentially circuit board version 1.41). The reason -
I never understood for what it is the mouse - COM?
PS / 2? Nowhere it is written ...

   KOE: For the mouse COM. There it is drawn
connector, you can guess or I
ask.

   By the way, if you order a card from the author
torus in the mail and decided to do without it
controller, tell him not to send for
stitched PIC.

   Also, more than I thought the printer buffer
DD69 and tape comparator (DA1) from
the entire harness. Following the buffer, the printer
oblivion went tying DD60: 1, and for
tape input - and the tape you
course (part of strapping DD35). Looking ahead,
I will say that the chain of generation flash (on the DD34: 3)
I also sent into oblivion, because it
simply does not work:) And if wages
Botha, needed a flash, not synchronized
LIMITED with HR?

   KOE: For component values, these
in the scheme, glitches arise. Occasionally (1
times for 5 cards) requires the selection transistor
pa on the static coefficient of transmission
base current.

   I made a flash generator on another
K561IE10 by soldering it on top of already nalichest
sponding DD8, establishments at her signal KSI, which
ry is divided by 32, turning into a signal
flash.

   Despite the removal of the buffer printer
keep in mind that wherever soldered DD69, yes
itself IR23 you can still come in handy
the process of finishing construction.

   Next, I decided to cut consumption
scheme to the maximum by setting a series of
KR1533 (74ALS 'in their language') whenever only
to the can. Not only managed to find in this series
the theory of IC IR16 (DD40, DD41, DD63), and
and LN2 (DD62). I left them a series of K555
(Or 74LS).

   KOE: LN2 necessarily be of a
indoor collector t.e.TTL, with 555 or
155 series. Otherwise, there would be problems with 5.25 di
skovodami.

   Still, chip DD59 I found only cross
KR531 theory, we obtain from it a miniature utyu
JOC:) In addition, non-critical to quickly
action chains I put chips series
74HCT. It is worth to make a campaign against illiteracy
about this series. So:

   Series 74HCT - differs from other topics,
what's inside is fully CMOS, as, for example,
K561 series. At the same time, however, the similarities and for
ends by. Series 74HCT works when on
voltage of 5 volts (561st - from 3 to 15) and
has a performance at the series K155
(Slightly faster or slightly slower than in each
case). Since it is CMOS, it
of course, eating (in the temperate parts of
max switching) is smaller than even the Series
KR1533. CMOS also means that the output
levels - 0V and 5V, and not as a TTL. Input
levels in this series - TTLovskie. There
exists as a series of 74HC (and its Soviet analog
log - KR1564), which differs from 74HCT those
that its input levels - no TTL, but 'as
have '(the datasheet'u - the threshold can
be arbitrarily from 1 / 3 Vcc to 2 / 3 Vcc) and
Because of this a little more bystrodey
The corresponding. In addition, its staffing stress
Food - from 2 to 6 volts. In practice, this
means that when applied to the input signal
from conventional TTL circuits in the worst case
did not work, and at best - will decrease
Immunity. There are also KMOPserii 74AC and 74ACT (and their 
Soviet analogy w, respectively, KR1554 and KR1594), differ

sponding to 74HC * increased bystrodeyst
condition. This performance can be increased
cause interference zhutchayshim Nutrition
for such is not particularly well-diluted plate
You, as a fee pentagon-1024.

   KOE: With regard to the supply rails, then all
sverhakkuratno done. There are disadvantages,
related to the fact that the wiring circuits takto
Vågå clock generator and get
las rather long, but big problems
it does not cause.

   The fact that increased bystrodeyst
condition implies a sharp fronts. And now
Imagine - a conclusion of this m / c, loaded
adjoint is not one of a dozen picofarads, re
cluded a nanosecond time from 5 to
0 V.

   KOE: If it is not loaded with one dozen
pF, the sharp edges in principle
impossible, because output resistance of the LE
course.

   Naturally, with the supply rails at this point
uzhiraetsya current is not in one ten milliohms
per. A sharp pulse of current (with drop dead
dI / dt) on the bus inductance hilenkoy pi
Tania or (worse!) of land will jump
voltage, resulting in adjacent
Power ICs speech impediment to the WMO
Ladies, as a whole - will trudnoustranimy
glyukodrom.

   KOE: For replacement in CMOS, I recommend
74ACT/KR1554. No interference on nutrition is not
will it prejudice. Interference occur
when switching from a "prosazhivaniya"
voltage on the segment of the connector
Bus-power output to power micro
scheme. And if this segment
is reactive (inductive) impedance
of. If you look at the wiring board, the
we can conclude that as an asset
Noe and reactance interval
Cove Trunk negligible
(Reactance order of a few
nH). With proper arrangement of the blocks
kirovochnyh tanks no problem.
Where dI / dt is large does not matter,
as energy for pulse switching Be
is taken not from the power bus. Current at the time of 
transition switchover logic elements ensures

is called the charge accumulated in the lockup
emkosti.Dopustim that the output of the element loading
wives of 10 pF and rise time of 10 ns.
Then, when the output voltage
0 .. 5 V pulse of current at the output of the logical
element (and hence in the power line) will be
approximately 5 mA. And if you count the total
inductance of power bus segments of
bypass capacitors to vyvodla pi
Tania 100 nH (which greatly exaggerated), then
prosazhivanie voltage on this inductive
surface is only 0.05 V.

   ... So, the main chain, I did not touch
(For xs, how they affect change in more
more brake m / ch), and decided to replace only
to DD26, DD35, DD49 (at 74HCT174), DD36-39
(On 74HCT374), DD78 (74HCT373), DD42 (74HCT
257), DD60 (74HCT74) and DD43 (74HCT08), and,
Truth be told, to buy without much straining
(And did not want to strain) could only
74HCT374, 74HCT373, 74HCT74 and 74HCT08. About
last - we shall be specially (about
why here in any case, it is desirable to
put 74HCT). Also, I put the CMOS pro-
core processor - Z84C0010 (10 MHz), which was
I kindly zabaryzhen notorious you
Comrade CHRV (aka Roman Chuunin) =), and to
tory It is not heated during operation.

   But before than to rush to buy all sorts of
there resistors, diodes and other capacitors,
We recommend that you read this article ho
thee to the place where I talk about the formation
rovatele video.

   Other things to talk about the numerous
connectors facing the board. Power connector
(Oldest ideshnyh screws, 5.25 "drives
or simply bought at the store) in the hole
boards with their feet lezet.Ya not soldered in but
both connectors copper winding wire maxi
normal climbing into a hole diameter (ochis
TIV from his varnish) and then soldered sticking
pieces of wire into the board, not regretting the solder
(For pulling out of the connector strap
He feels and conveys to the board means
lnye mechanical efforts!). Connector under
simovnitsu - taken from old ptsshnyh mamok
and sawed along (there it is usually double,
under 2 CIMMYT). Can also be used but
zhevoy connector (from the ISA-slots, for example) under
walking size, however, will be with hemorrhagic
sticking there CIMMYT. Pro Connectors ZX-BUS
written on the docks. 3 identical connectors and 3
Response of them (is placed on the
plume) - Standard IDC-34. Socket of
core processor - I recommend to put the collet -
you may have peretykat processor
significantly more likely than other m / c on
socket.


           II. The birth of a masterpiece


   I started with the assembly of the clock generator
(Something like DD1, DD2, DD7, etc.), after
then checked for its functioning oscillator
lografom.Dalee a memory management scheme
Tew, multiplexers, she simovnitsa, video part (so you can get 
on TV visor at least a raster). In the process of soldering,

naturally, do not forget about switching system
chnyh capacitors on power supply, or even
what has already been collected, will fail. The case is
went directly to the shapers of the form
deosignalov (transistors VT2-VT5 and obvjaza
ka). Then again, I digress, dedicated
schennoe video outputs at different sintering.

   According to the standard scale video signal long
wives to be 1 volt. The level of 0.3 -
black level, 1.0 V - white level, and
0 V - the level of synchronization pulses. For
chrominance signals (RGB) scale well with
component of 1 volt. And now look at
scheme for each of the Spectrum (or a poke in the not-
On an oscilloscope), and be astonished - swing
signals will be 'like a god sent' (more precisely,
as he sent the author of the scheme), and almost certainly
- More than the standard. Of course,
good TV digest video signal
Wild amplitude (holding enhancement system
AGC), and often cope with mnogovolto
Vym signals RGB. Or fail, then
you get a terrible, wildly rich color
picture that no adjustment is not
lead to a normal appearance. What, for example,
I received, such as connecting a scorpion
to monitor Microvitec (who knows how to
TV and VGA-scan).

   ... In short, I decided this outrage leads
STI in the divine form. Namely, to make time
Mahi signals 'as expected', and on to
unloading of 75 ohms (again, standard), but still
output impedance of the shapers ETS
lat 75 ohms. What I got - see
scheme in the appendix.

   An important role in getting the right time
strides playing DD43 type 74HCT08 (instead of
KR1533LI1) having a level of logical unit
boundary of 5 V. It is easy to see that the application
Series 74HCT output signal levels from
put the values prescribed standard
(If you decide to check, remember that from the side
hand the TV signals are loaded resistivity
resistivity 75 ohms to the ground!).

   Transistors KT315 (KT361 and) I nastoyate
flax recommend never to use, and immediately
throw them in the trash =)

   KOE: Well, why is it so? KT315 - quite
normal shirpotrebnye tranzistory.S them
problems arise only when they
work in the evanescent modes. In the remaining
Mr. horosho.Ya all pretty with these transistors
tors worked hard, not too much difficulty
was. These transistors are undesirable impurities
to receive in a major equipment (which are solved
matters of life and death), because for them
normalized characteristics of reliability and
by acceptance.

   It is better to take the same consumer goods -
KT3102 (pnp - KT3107), which have large
leading coefficient of current gain greater bys
trodeystvie and higher load current (which
I have proposed for the scheme is very important!).
Better yet - put here Transistors
type 2N2222 - they are more 'oak'. All
these transistors have a different pinout
(KT315 - 'emitter-collector-base', KT3102,
KT3107 and 2N2222 - 'emitter-base-collector
torus'). But I have proposed updating and so
implies nehily raskolbas, so that
modified pinout - a trifle:)

   Share this video out, do not add
and many of the audio output. Firstly, done
Shannon delusional look dividing concentration
0.1 uF capacitors capacitance in audio circuits
(Beeper and Auca). Not to mention the fact that the Galilean
May (as a rule, the ferroelectric) Ke
ramika in the way of sound - sax and mazday, replacing
chu only that the lower classes with such capacitors
will be lost forever ... Bet there though
electrolytes to 10 uF! Still I've got Poon
ktik - I hate mixing scheme Auca
resistors. Just hate it =) Recommendation
through each channel as do the mixing in the scheme
Turbo-FM (see website www.nedopc.com).
Advantages of this scheme: The channels do not affect each
each other through resistors; selection coefficients
Mixing is performed by one resistor
on each factor, possibly connected
headphones directly to the output operatsionnika. Himself
I did not do such a revision, since TurboFM I already have, and 
I use it instead one hundred Auca:)



      III. The proud title of 'The Pentagon'


   Here we will focus on the many to
processing, some of which allow
computer somehow work, others -
increase its compatibility with the most
original Pentagon, who wrote under the
all the demos (yes, as it turned out, this computer
not quite 'pentagon', both the position invariant
one and the ports of memory!). So, here we go:

   The very first refinement concerns the signal
formation of blanking. Case that
synchronization pulses in video signals up to
lzhny be a (standard) 'surrounded' level
it black. It is the responsibility signal
are input to 2, 5, 10 and 13, DD43. On
Scheme 1.41 shows that this signal
initially passed through the element DD27: 2
but then the author has removed it from the circuit signal
signal (leaving the configuration of the input of the
element to the signals, as in the diagram), filed for
DD43 direct signal DSP. However,
DSP provides signal blanking only
for the horizontal sync pulses, but not
for frame synchronization pulses. Dan
LIMITED deficiency manifested itself on my Television Electron
results as follows: the curb with black telly catch
frame sync (when blanking and
was so), while white - do not catching. Detailed
study showed that the second coming on
DD27: 2 signal and provides just blanc
King during the clock staffing, and
error the author was that, instead of
an AND-NOT should be an element of I.

   KOE: The site is a matter opisano.Pri times
vodka board really an error.

   I corrected this error MGTFom and Freedoms
nym inverter DD76: 2. After that, my those
lectures became the 'train' fish personnel.

   The next revision of the conduct of the
ready computer when the power
and discharge of a TR-DOS. Specifically, after including
radiation power or discharge of a TR-DOS cleared
Xia screen and after a pause ў5 seconds flew
128th menu (speech about the standard firmware
- Trdos5.03 + + vasik128 vasik48, PZUshku with
KOE which will make the request to empty
of the board). Analysis showed that
resetting the power-on state
trigger TR-DOS (DD55: 2) - indefinite,
and when you reset button - no change. Bug
fix is obvious - it clings to the leg 10 through
diode ('plus' for this leg) to the circuit common
reset. Now, for any discharge begins
immediately to work vasik128.

   Several cosmetic improvements -
I had to pick up capacity capacitors
pa C10 bezglyuchnoy for writing to disk
(Raising it to 470 pF), and I stuck
series with the capacitor C1 (which
ry forms reset at power-up)
Resistor 15 ohms, thus avoiding the departing
on the screen of the band at the time of the closure button
reset. There is also a cosmetic refinements
include a decrease in resistance
resistance of the resistor R7 (for datasheet'u to
N-MOS CPU it generally must be
330 ohm!) Installation on reset circuit diode
('Plus' on the chain, 'net' on the bus
supply), which does not 'jump'
potential of this circuit over the potential nutrient
of when disconnecting thereof; and installation of the blocks
kirovochnogo capacitor for power bus
processor directly beneath it (the point
her under his sockets).

   Somehow, when reading the port # FE senior
bit - nul.Kak says jtn, it may at
lead to a glitch in some programmah.Zna
hence, separates the 14 th DD68 foot off the ground and ve
Shai on food. Data bus in the IM2 of TRDOSa - fragile, but it 
does not matter: the break system was driving, walking on the 4 
th leg DD24: 2, inserting eat free element "and" DD77: 4 and in 
addition plant for an output from the processor M1.


   For some reason the input of the BC2 Auca filed M1,
instead it is served on the decoder
port memory and Auca DD64. Is it good to 2
reasons: firstly, as seen from the chart
working with these strobes Auca (BDIR, BC1,
BC2), in noncommutative 'cunning' state bus
address at the time of reading of the vector in IM2
Auke can get some 'left'
register, and secondly, the scheme and the Turbo-Sound'a
Turbo-FM'a believe that the entrance BC2 not zadey
sponded. Solution again is simple - detachment
Vai M1 from the BC2 and hang there +5 V, and 6 th
feet DD64, respectively, separates A1
(Why would he?) And suem there brutally Authorit
vanny M1:). As a bonus, we get
Port # FC, employed in some (not
remember which) Deme.

   Now take a proverbial port
# FD. As can be seen under the scheme, it is not
completely. More precisely, it is not no ...
In general, if you make a LD A, # 10: OUT (# FD), A,
then A is written in # 1FFD, although the option
LD A, # 50: OUT (# FD), A pass, as ass
washed (port # FD in this way works
on kayah and a lot more than on, but not equal
bot to the pentagon-512k in the mode of 512k!).
But we have the same pentagon! How can that be! That
do? What's that ... Fixit!

   As can be seen from the scheme, and it follows from
describe the memory model, the derivation of the 1 in 5 th
bit port # 1FFD PC into a 128kilobaytny. I did that in this mode
appeared and 'honest' port # FD. This realization
ized as follows: signal A14, walking on two legs
DD64, breaks away from her, and the factories that
as A14, but missed a free
element "OR" DD70: 2. The other input of this
element is fed out again freely
element, "NOR" DD75: 4. This element
ment signal A14 ^ A15 (which is
zero if the appeal goes to the port # 1FFD and
per unit, if the request is for an honest port
# 7FFD), as well as 'cunning' lock signal.
Sly, he, in the sense that the inverted
and at the same time slightly delayed on the trigger
DD2: 2 (free). Namely, the reset input
this trigger signal is a master reset
RES, the input of the installation - filed 5V to shred
- The signal sampling port # 1FFD (the 15 th leg
DD64), and the input data - the lock signal
with the 10 th leg DD49. Inverted output of this
a trigger connected to the element of DD75: 4.

   The meaning of all this nonsense is as follows:
mode memory locking> 128k appeal to
Port # 7FFD runs as usual (since for
The signal A14 ^ A15 per unit, and hence
Indeed, A14 passes to decoder DD64
unchanged), but when accessing the port # 1FFD
(Also known as port # FD), instead of the signal A14 de
Encoder is a unit, turning the port # 1FFD
in # 7FFD. When accessing the ports Auca A14
is unchanged. A trigger is needed for
signal delay lock on this at
adultery, because otherwise after the inclusion of blocks
mounting plate in the same machine cycle will gate
and the port # 7FFD, writing unit there 5 th
bits and turning, so in general mode
48k =).

   Now, to include Port # FD and 128-th
mode, simply log on to 128 th BASIC and
write it OUT 8189,32. You will be told 0:1
O.K. Then go ahead and start any
stroke or ayeyk;)

   Although wages and similar to the Pentagon on the circuit
metal, and even sweep (the same preslovu
Tide 320 lines are taking place), cartoons
Laura and curb the effects are deeply
uh ... is not there:) A more thorough analysis
(Eg, program TEST4.30 and demoy
RAGE) yields the following - INT comes at 16
lines later in the vertical and desyatokdrugoy microseconds - 
horizontally. (Incidentally Tee, if Deme RAGE in the latter 
part of wait a very long time, the rotation still remains

becomes:). Vertical int vertaem back
elementary - served on the capacitor C2
instead of the signal from the 6 th leg DD12 inverted
en version of this signal by passing it
a free element DD75: 3. By Horizontal
ntali INT I put in 2 divided doses. First
filed for a tuft of DD19: 1 is not DSP, and SSI, after
Why INT rose almost to the place - now he's
was approximately 4 points before the required
position. If you are satisfied, then
on this and take it easy, I will not stop
and replace the printer buffer IR23,
first connecting its inputs and outputs
so to get the shift register.
Filed on her 7 MHz from near
DD40 and DD41, the input of the first stage gave
SSI and began to pick it out, from which
signal should apply for a tuft of DD19: 1. It turned
elk, enough to hold exactly SSI 4
tact (semimegagertsovyh) to INT rose
to its rightful place. You can nama
Drita scheme, such as 'printer's' trigger
Gero DD60: 1 (torn from him preliminary
but regarding the printer), and tucked him
tuft to get a shift by 4 pixels
or in some other way. For example, taking one more
IR16 and soldered it has some legs on
yuschiesya DD40 or DD41.


    IV. Pro blunt incandescent material

             everything else


   So, after this whole heap of improvements
board will look like an ordinary spectrum
Terumah ordinary user blunt Rusk
lennogo subject, although in this case will be
programmatically indistinguishable from the original
pentagona.Kakie conclusion is? Pres
First of all, it is obvious that the Pentagon-1024SL ve
rsii 1.41 - how to completely and the Pentagon. C
'Fair' by the Pentagon, from the viewpoint of Paul
transducer, his family did that only once
nimbly in 320 strochek.V rest of this - not
cue distinctive 'kompuntel'.

   KOE: The original charge was made exclusively
respect for my personal polzovaniya.A for
shows one card is impractical therefore
Therefore, all that happened. By the way, where you are
was when I did the version 1.4 (1.41)? After
ask what you want from the board, which must,
Why not nado.Realno interest on stage again
processing showed 2-3 people, although all
was extremely open. (See the archive
News on the site!) and on the Internet is not everything
ended, yet at the stage of a
from people with whom I corresponded, BOAD
Force information in Fido.Ya honored archive con
conferences - the same thing, almost none of the
interested.

   Despite the seeming horror of the situation,
it was quite easy to bring this plate
that the original pentagonu.Vse still ska
are called roots. By the way, the roots of this board
computed again jtn - its immediate
nym ancestor is a certain scheme, to which
swarm in the upper left corner of the Pentagon says
128K 1993

   KOE: And I do not skryval.Ishodnaya scheme
ma lies on the site. It was produced by
"Solon" in 1993-94. Then (in 1994 th
dy) name changed to GRM-128, and the scheme
remained the same.


              V. What's next?


   And then that's what. If you do not watch
zx.pk.ru, then I'll tell you what worked KOE
and prepares for 'serial' release ...

   KOE: 'Serial' registration (100-1000 pcs.)
will not. Will be ordered pack of 5 .. 10
boards.
... Board his new Pentagon, namely,
Version 2.1. On this board instead of almost
TNI m / c rassypuhi now live 2 PLISiny
- One EPM7128, another EPM3032 (two - the flow
Therefore, that one leg was not enough and the whole logic of
not climbed).

   KOE: And here is not so! I specifically
chose FPGAs from different series of self-serving
Mi development tasks like programming
firmware and device design
on different types of FPGA. In addition, such results
solution was optimal in relation
price / features.

   Live there and rassypuha, but rather in the mind
rennom quantity. Payment is made under this
Standard AT - can Screw in ptsshny
corpus. Its 'yuzersky' features: on-board
is PAL encoder, so that connection to those
levizoru not cause problem.Takzhe on the board
are audio power amplifiers - you can
just plug headphones or the speakers =)
Removed signals frame and line synchronization
Organization (KOE said that did not have legs
PLISiny to bring them). So what to monitor
torus now connect a big problem.
Increased the number of slots to 3 pieces. K
Unfortunately, the author is not going to give
vat source firmware PLISin (Sprinter
remember?).

   KOE: FPGA firmware files needed
to build (and repair in case of breakage)
boards are standard documents
ntatsii.V this kit does not include source code
firmware, which required only at the stage
development to make changes. I do not
v. to put to the public and
firmware source code, just do not want to
someone other than me, they tinkering. If
Who wants to read the source, then
share.

   I can only hope that the KOE,
after reading this article, try to make
so that its second version of the Pentagon was
no worse than the one that turned me on
all modifications.

   KOE: He's already better. Incidentally, the scheme for this
time was re-synthesized, fully
modeled in CAD FPGA. At this time
she rastaktovke and INTu exactly correspond
there exists a Pentagon-128 1991 year (done by
requests of workers), not 1993, as was the
the previous version.

   As for my 'ispoganennogo'
Pentagon, the plans for it are as follows:
sculpt a new PZUshku, in the shadow area to
Torah will lie QC3.11, and the menu is 128-st
BASIC will choose the mode 128k (where
I now have the port # FD). In addition, I
(In order of development PLISin) going to ETS
lat turbo mode on PLISine (EPM7064), at
than bezvaitovy (with a stop shred, as in
original Spectrum). Time will show that
this true.



   Well that's all. All this rigmarole scored
Uncle lvd from mayHem. E-mail him at
lvd@dgap.mipt.ru or 2:5020 / 835.1, for
go to channel # mhm at irc.forestnet.org
 or type in the ICQ number 77187846.

   Credits declared: most King
Of Evil'u, which muddied the new reinkar
nation of the Pentagon; CHRV, who helped with
zetnikom and gave many valuable suggestions, and, on
end jtn'u who gave too much price
GOVERNMENTAL advice and spent the archaeological research
route on the scheme of the pentagon
:)