06 февраля 2017

                      ATM-Turbo and others
                        by Alone Coder
                   corrected by Lord Vader

During the Soviet microcomputer revolution, the mass character
of production definitely was not provided by radio amateurs.
Most computers were made by factories(sometimes converted
defense plants, so a manual for one computer named "IMPULS" was
an official document 3.098.018 D15 with stamps and a list of
registered changes), and also by various firms.

One of such firms was an operating department"ATM-COMPUTER" of
"ATM" (that stood for"Associaciya Tvorcheskoy Molodezhi" - 
"Association of Creative Youth" ), later said to be namedAOZT 
"ATM" ("ATM" for"Association for Technics and 
Microelectronics"). They were based at VDNKh and Dom Kultury i
Tekhniki "Sozidatel" (center for culture and technics 
"Creator"). In 1990(Radio 11-1990 p.30) they sold books, 
circuit diagrams, flashed ROMs - and produced a ZX Spectrum 128
compatible computer with a disk drive and a printer port. They
mentioned specially that it had 128K of RAM, maybe because
Hobbit was advertized as 128K (RAM+ROM). This computer is easily 
identified as "Pentagon 128" also known as "Pentagon 2+" and
"ATM-128"(МНОГОФУНКЦИОНАЛЬНЫЙ КОМПЬЮТЕР ATM-TURBO. - АТМ-ПРЕСС, 
1992 p.4)."ATM" mentioned it later in their ads as their
development(Radio 9-1991 p.75). One known layout even has"ATM
1991" label on it:
http://sblive.narod.ru/ZX-Spectrum/Pentagon128k/Pentagon128k.htm

Of course, a private enterprise could not match the production
numbers of factory-made Speccies, especially when Speccy ULA
clone TЗЧVG1 appeared. The number of Speccies by the end of the
era might exceed one million, as they were more widespread than
the official one million of BKOO10's, DVK's, and UKNC's
combined - but these Speccies were mostly tape-driven 48K
intended for gaming. The guys from"ATM" didn't target that
market. They wanted to sell Spectrum as an alternative to IBM
PC, for a minority user - a "serious" customer. 

The main rival of"ATM" in Moscow was not IBM, but a subsidiary
of Russian-Spanish-Lebanese joint venture"Kramis", named
"Profi"(Radio 8-1992 p.63, 8-1994). Later ads and resources
used another name:TOO "Condor" or "Condor (Kramis)", with
another location and phone - the location was "Library for youth
#214". Compare it withATM's location. Or compare with a phone 
in C-64 demo "Meaning of Death"(1993) - located in "Moscow City
Palace of Child (Youth) Creativity". 

The research of these names is like a detective story."Profi"
might be an unregistered part ofJV "Kramis" ( "Kramis" produced
wallpaper and woodwork, not computers), and"Condor" might be
its name after registration. The only likely"Condor" whose
traces can be found online isTOO "Condor" headed by Tadeusz
Radjusz, known Polish publisher who started with computer
business(see his interview in Wyborcza
http://wyborcza.pl/1,75248,140079.html ). This"Condor" is now
defunct - the address and phone are now occupied by political
organization named"Black Hundred" (its leader Alexander
Stillmark was a brother ofRadjusz's wife). If this Radjusz was 
the businessman behind Profi, we must mention his work in
Soviet-Polish "Computer" magazine (1990-1991) and the fact that
Radjusz was the first FidoNet user in Moscow (also around 1990):

! ----- цитата из FIDONET.HISTORY FAQ -----
!Там была семья: муж-поляк Тадеуш Радюш [так он печатался в
!журнале в русской транскрипции -- Stanislav Efimov,
!2:5020/525.1024] и жена-русская Лена Радюш. Собственно узлом
!и BBS занималась Елена.
!Стояло это все хозяйство у них дома в районе
!м."Краснопресненская".
!А еще был - а может еще и сейчас есть - советско-польский
!журнал "Компьютер", издававшийся здесь притом, что макеты
!делали в Польше.
!Елена сотрудничала с этим журналом. Русские статьи - примерно
!половина объема журнала - передавались отсюда туда mailer'ом
!прямыми звонками, а назад качались уже свеpстанные варианты.
!Ну, а BBS была вроде как для порядка - компьютерный журнал
!всеж-таки. Стояло все это хозяйство на 386/25, модем 2400/MNP,
!BT 2.20/2.30 и RA 0.04. Сам Тадеуш, хотя и фигурировал в
!nodelist'е и в заставках, к узлу не подходил, а занимался
!коммерцией.
! ----- конец цитаты -----

Radjusz wrote an article about Betadisk in this magazine.
There also were rumours thatRadjusz brought Beta 128 Disk
Interface in USSR, however the first clone was produced in 
early 1988 by other people -Sergey Pacyuk and Vyacheslav
Bogomyatov fromNTK "Plus", Moscow (after the schematics from 
some unknown Czech magazine). That same Beta 128 that became the
standard for "serious" clones in USSR being virtually unknown
outside, unlike the original Betadisk.

This"Kramis"/"Profi"/"Condor" released its Profi computer in
late spring of 1991(date from Turbo2+ book by MicroART; no
earlier ads known), and that was possibly the first mass
produced Spectrum clone with AY-8910/12 sound chip, and also
with 256 to 512K RAM, high resolution graphics and turbo mode.
Byte-01 computer produced in Brest had high resolution and a 
disk drive by 1991 with CP/M to appear by 1992(copyright 1990,
but still missing in Radio 7-92 p.26). AndSergey Zonov in
St.Petersburg developed ZS Scorpion with 256K RAM and(later,
under the influence of Hobbit - see interviews in ZX-Format #2
and #6) a shadow monitor... And everybody in this market
provided a printer connection.

The reader must know that there is a widespread misdating of
Scorpion launch, based onZonov's interview in ZX-Format #2. In 
fact,Zonov hadn't made the circuit layout by summer of 1991. If
we combine all the sources, he showed the prototype on Leningrad
board with extra wiring in October 1991(just as MOA and
Rodionov's book "ZX Spectrum for users and programmers" came out
- 16.10.91). Incidentally, "MEGA SOUND 128" demo byVasilyev
Anton was written just then (10.10.91 - seemingly the first
Russian Speccy music demo),Vasilyev Anton is known to work for
Zonov (his patch of Song in Lines special forZonov, 1993). 
Scorpion ROM copyright is 1992-1994, Shadow monitor copyright is 
1993-1994, Scorpion manual was printed in 1993, CP/M was adapted
for Scorpion in 1992-1994, and the first Scorpion ads were
mid-1993("new development" ad in "System programs for ZX
Spectrum" issued 15.8.93 by Piter, surprisingly mentioning CP/M
support). Moreover, iS-DOS programmerAlexey Leontyev (also from
St.Petersburg) even thought that Scorpion came out later than
ZX-Next 512(Open Letters, 28.7.99: he quotes date for 
nextS12.hlp - help for ZX-Next iS-DOS driver - as 11.10.93):

"Очень  обрадовался  за  Вас, узнав, что у
Вашего  ZXNEXT  целых  512  kB ОЗУ. Также
можно  порадоваться  и  за  многих других
владельцев   данного   компьютера.   Файл
nextS12.hlp  датирован  аж 11.10.93, т.е.
москвичи  получили полумегабайтную машину
чуть  ли  не раньше, чем в Питере сделали
256-килобайтную."

"ATM" couldn't make less. And the needed machine was developed
for"ATM" by its creative team "MicroART", based on their
version of Pentagon(see above, they quoted the fact once more
in their Turbo2+ book). Note that the original Pentagon (48K)
was not their work, it was invented in 1989 by a radio amateur
Vladimir DrozdovRAЗAO(information from Konstantin Sviridov).
"ATM" continued to sell Pentagons, too, as"Profi" sold 
"Nadezhda" 128K machine(Radio 8-1994). 

"ATM" also worked jointly with a firm named"INTER-LINK" that 
produced various connectors and TV adapters(Radio 8-1992 p.63).
ATM-Turbo (initially "АТМ-СР" that could possibly stand for 
"sovmestnaya razrabotka" - "joint development") was launched in
1991(Radio 9-1991, p.75).

Later"MicroART" parted with "ATM" and started to produce
computers on its own. They were named (ATM)Turbo 2 and Turbo 2+,
with little difference toATM's ATM-Turbo 2. There is a
copyright problem: for example, ATM 4.20 scheme has(c) MICROART
on it, while 4.50 and 6.2A (from some album) haven't one. Of
course, the schematics of ATM-Turbo 2 were undisputably
developed by"MicroART".

"MicroART" also made modems and ROM programmers, printed books,
and sold PCs. Now they produce electrical equipment.

                             * * *

What"MicroART" did actually to the machine?
The first was memory.

128K of RAM in Pentagon 128 are changed for 512K almost
mechanically. Hovewer, program adaptation was a big issue.
Programs could access any, even crazy, port numbers, they could
address port#FD, even with non-standard bits in it. And any
non-working program could make the "serious" customer angry.
Each of the clone vendors used their own port, so did ATM-Turbo
(#FDFD).

And they guessed wrong (there's no ideal port number) - so in
1992, in ATM-Turbo 2, they hid the port, addressing it as
Betadisk ports (in 2+ they also made 1 Megabyte RAM). This made 
good compatibility at last. This decision also allowed adding
three more ports to switch RAM in other 16K addressing windows.
Sadly this required rewriting all the specially written software.

In 1991-1992, the idea of rapidly growing business was in the
air. Every vendor thought that programmers would support his
standard and there will be a lot of software to sell their clone
with. And all the vendors reserved Spectrum 128K compatibility
despite the low number of 128K software then. Vendors were even
ready to sponsor software development, as it was in case of
Zonov."Condor" and "MicroART" had their own programmers in the 
team. But they needed an operating system.

While the commercial blockbuster iS-DOS hadn't made its launch,
there was only one option, CP/M.
This operating system was a standard for Z80, it had a lot of
software and was already used in Spectrum world at Timex
Sinclair 2048/2068, ZX Spectrum +3, also in Polish and Romanian 
clones, in Czech devices byLEC and Sinsoft, at Aerco FD-68
Floppy Disc Drive, and probably elsewhere. 

WhileMOA (Andrew Larchenko) was adapting CP/M for Scorpion, and
Caro (Kamil' Karimov) for Kvorum -- Profi and ATM-Turbo (as well
as Byte ) were already provided with CP/M. In case of ATM-Turbo
it was even in ROM (optionally) and with TR-DOS formatted disks
(also optionally).

There is a possibility that CP/M for ATM-Turbo was not ready by
1991 because it wasn't mentioned in the ads. The same with
Profi. 

Interestingly,MicroART's archives contained documentation for
СМ1800 (SM1800, Soviet 8080-compatible workstation from 1979) -- 
did the authors adapt CP/M from there?

CP/M had a big help from extra graphics. 

Unlike Timex, Profi, and Byte, extra graphics in ATM-Turbo was
designed to match that of IBM PC:320x200 and 640x200 (that was
better for standard 80 characters in CP/M ). Also even the first
version of ATM-Turbo had the high resolution graphics coloured.
It also had 16 colours per pixel mode. Profi was blamed for its
absence in parody Best 0 Super Demo (it was an answer for Profi
advertizement once made byTHD software company) in 1993, but it
was too late."Condor" added colours and was not ready to
continue the development, later switching to Amiga(see X-Ray
#01)."MicroART" took a risk and added a textmode in Turbo 2. It
proved to be simple, and the computer was still one board (Profi
had two). ATM-Turbo was designed for "Mikrosha" (Soviet
8080 clone based home microcomputer) case. ( Scorpion - for
"Korvet" case. Byte and Profi had their own cases.) 

How the colour-per-pixel mode was done on a Russian Speccy
clone? Quite simple.
1. Make a hardware mulicolor (just address multiplexing).
2. Make it work with frequency twice as high (the authors used
both RAM lines at once, to avoid stopping the CPU) and use one 
more address bit. So we have hires mode. 
3. Initially we read in attribute register and pixel register.
Instead of this, make it read only in attribute register and 
commute meander instead of pixel data. So we get the following 
bit placement: %RLRRRLLL (LLLL=left pixel, RRRR=right pixel). 
Later the 1бcol addon for Pentagon was made the same way. 
4. Finally detach the higher addresses from screen geometry
counters in separate counter, and change the border logic to 
make 320x200. Authors put the needed logic in simple 
programmable logic array (like Scorpion authors; Profi and 
ZX-Next used ROM for this). This was needed anyway -- first, to 
keep the chips number low having corrected TV raster (Pentagon's 
exotic raster was born to save chips, and it was a sensitive 
issue that time, discussed in ZX-Review), and second, to 
copy-protect the board. (Only later all the ATM-Turbo 
documentation was released in public domain.) One funny thing 
about this PLA logic was a possibility of hardware vertical 
scroll (see Catdemo and Info Guide #10), accessible no easier 
than Atari ST's one :) 
[http://www.pouet.net/topic.php?which=6844&page=1]

Graphics output was connected to a palette circuit and a SECAM
coder. That SECAM coder could be fromINTER-LINK, and it was
removed later.
The palette is organized as an external device. Colour number to
use both for display and for colour change is determined with
RGBI signal from the computer, and the data for colour is taken
from CPU databus when CPU accesses one of "disk controller"
ports. LateDDp's scheme to attach to Pentagons et al
(http://realddp.narod.ru/zx/palЧЧ4/index.htm) had 4096 colours
and used data also from higher part of address bus (they match
databus if you access the port viaOUT (255),A -- all the known
programs do it this way so it's compatible). But originally
there were 64 colours as in EGA. Just to convert PC software,
especially games.

As in Profi, the first ATM-Turbo games were copies of IBM PC
games.
However"MicroART" didn't stop at Color Lines and Mine. They
also ported Prince of Persia and Gobliiins. The ports were so
close that even the passwords matched the originals!
Commercial ATM-Turbo games were also written by"MacroSystems
Laboratory" and maybe somebody else.
All the commercial development ended around 1996. However in XXI
century, when new incarnations of ATM-Turbo appeared (initially
fromNedoPC ), the number of games reached 40, and the total
number of software titles is around 300
(http://alonecoder.nedopc.com/atmsoft.txt). Even if we don't
count software that requires only DAC.

Games needed graphics -- and sound.

Sound chip named AY-8910/12 was the only foreign part in
ATM-Turbo schematics that couldn't be substituted with parts 
made in USSR. (Expensive 27512 ROM chip could be changed to
smaller 57ЗRF2 to load the operating system from disk or even
tape.) However you could do without AY-8910/12. There was an
8-bit DAC in printer port (ZX Lprint's port#FB, as in Pentagon
128 ), and you could hear IBM PC's digital music with it. The 
replayer was sold as "Scream Tracker Player" or simply "COVOX".
Profi also had one, and Profi authors even supported stereo 
sound at the cost of non-standard printer ports.

Initially ATM-Turbo included a modem based on this DAC. One
known program to use it was"АОН-Секретарь" ("Phone number
detector and secretary"). But later the authors removed the
modem to sell it as an external device, keeping only DAC and ADC
on the board.

A lot of digital music must be saved somewhere, so the user
needed an HDD.

ATM-Turbo 2 contained seemingly the world's first IDE controller 
for ZX Spectrum, maybe even the first HDD controller for it
(Tyrsin'sMFM controller shown atIskraSoftheadquarters was 
also developed in 1992 - see ZX-Format #5: "IDE HDD"). This time
ATM-Turbo authors surpassed Profi authors. 

The first ad of ATM-Turbo 2 was published in Radio 10-1992 p.63
-- note 1992, not 1993 as it was believed before! Sadly IDE
controller was not mentioned in the ad so we don't know when it
got worked, possibly in 1993 when its schematics were published
(incomplete ATM 6.00 scheme has N11,N12 instead of IDE's
D32,D34) -- but still the first IDE controller for ZX Spectrum.

This controller was a full-scale IDE interface. CD-ROM could be
connected to it but CD-ROMs were beyond the reach. When CD-ROMs
became widespread, ATM-Turbo even played CD video -- not bad for
USSR technology!
HDD support was included in CP/M (i.e. in ROM), so the computer 
once again approached IBM PC possibilities.

These PCs became cheaper and cheaper and spoiled the business.
(Later MicroART, like other competitors, was also forced to sold
them -- see Radio 2-94 p.47. Profi authors sold Amigas.) At the
same time the popular 58-keyboard with Russian letters except
"Й" (who made it?) slowly disappeared, and IBM's consumer goods
were used instead. This needed a keyboard controller.

Authors installed in ATM-Турбо 2+ a microcontroller named
181бVEЗ1 ( i8031 compatible). Its firmware provided serial 
interface and XT keyboard support (later AT keyboard). A mouse
could be connected to serial interface, but software for this
setup is not yet found. Instead,Kamil' Karimov connected a
modem in 2005 and managed to ping the Internet.

The first versions of ATM-Turbo had a "turbo" switch that could
be installed in "Mikrosha" case or old PC case. But "Mikrosha"
cases turned to be rare, and newer PC cases had no "turbo"
button. Maybe this was the reason why the switch was changed to
a port (thanks to the fact that the ports became hidden). With
these hidden ports a "compatibility switch" (that was in Profi )
was also unneeded, and there was no place for it, too.

A side connector (as in Profi ) was also impossible to install
in а PC case. ATM-Turbo had no system bus, there were just a
couple of ports to attach a PROM programmer. So when General
Sound appeared in 1996, it had to be attached by soldering wires 
to the board. ATM-Turbo 2+ unit owned byMaksagor is one
example.Maksagor also said that Zorel (who is known for
new ATM-Turbo board) also develops another version of the
board, with old chips but with an expansion slot like Scorpion's
(ZX-BUS). There are already implementations of ATM-Turbo 2+ on 
modern FPGA hardware: ZX Evo and Pentagon 2.6ббLE. FPGA
emulators allow to make ATM-Turbo 2+ from its ancestor Pentagon 
just with changing about 200 lines of code.